# Analytics API Source: https://docs.magnific.com/api-reference/analytics/overview Track your team AI credit consumption and discover team resources with the Magnific Analytics API. Programmatic access to the same credit usage data available in your dashboard and CSV export. The Analytics API lets you retrieve your team's AI credit consumption data programmatically. Integrate usage data directly into your reporting pipelines, BI tools, or internal dashboards. Available exclusively for **Business** and **Enterprise** plans at no additional cost. It does not consume credits. ## Credit usage Query credit consumption over time with [`POST /v1/analytics/team-credit-usage`](/api-reference/analytics/team-credit-usage/query). Each response includes per-user breakdown with email addresses and per-project grouping automatically. ```bash theme={null} curl -X POST https://api.magnific.com/v1/analytics/team-credit-usage \ -H "X-Magnific-API-Key: YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "granularity": "day", "start_date": "2026-05-01", "end_date": "2026-05-31", "group_ids": [10] }' ``` ### Example response ```json theme={null} { "data": [ { "date": "2026-05-01T00:00:00+00:00", "consumptions": [ { "tool": "AI Image Models", "user_uses": 45, "user_credits": 1800, "user_usages": [ { "user_email": "designer@company.com", "user_uses": 30, "user_credits": 1200, "project_reference": "a1b2c3d4-1111-4000-8000-000000000001", "project_name": "Marketing Campaign", "group_name": "Design Team" }, { "user_email": "developer@company.com", "user_uses": 15, "user_credits": 600 } ] }, { "tool": "Asset Download", "user_uses": 10, "user_credits": 900 } ] } ] } ``` ### Filtering Narrow down results to specific subsets of your team: | Filter | What it does | Where to get the IDs | | -------------------- | ----------------------------------------------------- | --------------------------------------------------------------------------- | | `group_ids` | Show only consumption from members of specific groups | [`GET /v1/analytics/team-groups`](/api-reference/analytics/team-groups) | | `project_references` | Show only consumption within specific projects | [`GET /v1/analytics/team-projects`](/api-reference/analytics/team-projects) | | `api_key_ids` | Show only consumption from specific API keys | [`GET /v1/analytics/team-api-keys`](/api-reference/analytics/team-api-keys) | ### Time granularity The `granularity` parameter controls the time bucket size. Defaults to `day`. | Value | Description | Best for | | ------- | ------------------------------------------- | ----------------------------- | | `day` | One data point per calendar day | Short ranges (up to 30 days) | | `week` | One data point per ISO week (Monday–Sunday) | Medium ranges (1–3 months) | | `month` | One data point per calendar month | Quarterly or yearly overviews | | `year` | One data point per calendar year | Multi-year trend analysis | ### Date range limits * If no dates are provided, returns the **last 7 days** * **Enterprise**: maximum 365 days * **Business**: maximum 180 days ## Discovery endpoints Four companion endpoints let you list team resources and get the IDs needed for filtering credit usage: | Endpoint | Returns | | --------------------------------------------------------------------------- | ------------------------------------ | | [`GET /v1/analytics/team-members`](/api-reference/analytics/team-members) | Members with email, role, and status | | [`GET /v1/analytics/team-api-keys`](/api-reference/analytics/team-api-keys) | API keys with name and status | | [`GET /v1/analytics/team-groups`](/api-reference/analytics/team-groups) | Groups with name and member count | | [`GET /v1/analytics/team-projects`](/api-reference/analytics/team-projects) | Projects with name | Cache discovery results — team composition changes infrequently. ## Authentication All endpoints authenticate with the `X-Magnific-API-Key` header. Your team is identified automatically from the API key. See [Authentication](/authentication) for details. ## Rate limits All Analytics API endpoints are limited to **10,000 requests per day** (10,000 RPD) per API key on the **Business** and **Enterprise** plans, the only plans where the Analytics API is available. ## API reference
Query credit consumption over time with filters List members with email, role, and status List API keys with name and status List groups with member count List projects with name
## Frequently Asked Questions The same data available in your Credit Usage dashboard and CSV export: date, tool (display name), user email, usage count, credits consumed, project, and group. The difference is structured JSON instead of flat CSV rows. No. The Analytics API is included with your Business or Enterprise plan at no additional cost. Call the discovery endpoints first. For example, `GET /v1/analytics/team-groups` returns all your groups with their `group_id`. Pass those IDs into the `group_ids` filter when querying credit usage. The same applies to projects (`project_references`) and API keys (`api_key_ids`). Enterprise plans can query up to 365 days. Business plans can query up to 180 days. If no dates are provided, the API defaults to the last 7 days. # List team API keys Source: https://docs.magnific.com/api-reference/analytics/team-api-keys get /v1/analytics/team-api-keys List all API keys in your team with their name, status, and creation date. Use key IDs to filter credit usage reports. # Get team credit usage over time Source: https://docs.magnific.com/api-reference/analytics/team-credit-usage/query post /v1/analytics/team-credit-usage Query your team AI credit consumption over time. Filter by group, project, or API key with day, week, month, or year granularity. # List team groups Source: https://docs.magnific.com/api-reference/analytics/team-groups get /v1/analytics/team-groups List all groups in your team with their name and member count. Use group IDs to filter credit usage reports. # List team members Source: https://docs.magnific.com/api-reference/analytics/team-members get /v1/analytics/team-members List all members of your team with their email, role, and membership status. # List team projects Source: https://docs.magnific.com/api-reference/analytics/team-projects get /v1/analytics/team-projects List all projects in your team with their reference UUID and name. Use project references to filter credit usage reports. # Audio Isolation - List tasks Source: https://docs.magnific.com/api-reference/audio-isolation/audio-isolation-tasks get /v1/ai/audio-isolation Get the status of all audio isolation tasks # Audio Isolation - Extract sounds from audio/video Source: https://docs.magnific.com/api-reference/audio-isolation/isolate post /v1/ai/audio-isolation Isolate and extract specific sounds from audio or video files using SAM Audio AI technology. Describe the sound you want to isolate, and the API separates it from background noise. **Use cases:** - Extract speech from noisy recordings - Isolate musical instruments from a mix - Separate specific sound effects from video audio - Remove background noise while preserving target sounds **Input options:** - Provide either an `audio` URL/base64 or a `video` URL/base64 (mutually exclusive) - Supported audio formats: WAV, MP3, FLAC, OGG, M4A - Supported video formats: MP4, MOV, WEBM, AVI - For video input, use bounding box coordinates (x1, y1, x2, y2) to focus on a specific region **Output:** WAV audio file containing the isolated sound # SAM Audio - Audio Isolation API | Magnific API Source: https://docs.magnific.com/api-reference/audio-isolation/overview Isolate specific sounds from audio or video with SAM Audio. Extract vocals, speech, instruments, or sound effects using text descriptions. WAV output for production workflows. Powered by SAM Audio technology, this API isolates specific sounds from audio or video files using natural language descriptions. SAM Audio is an AI-powered audio isolation API that extracts specific sounds from audio or video files based on text descriptions. Describe what you want to isolate - vocals, speech, instruments, or sound effects - and receive a clean WAV file containing only that sound. The API supports both audio files (WAV, MP3, FLAC, OGG, M4A) and video files (MP4, MOV, WEBM, AVI) as input. ### Key capabilities * **Text-guided isolation**: Describe any sound to extract (e.g., "A person speaking", "Piano playing", "Dog barking") * **Multi-format input**: Accepts audio (WAV, MP3, FLAC, OGG, M4A) or video (MP4, MOV, WEBM, AVI) files * **Video localization**: Optional bounding box (`x1`, `y1`, `x2`, `y2`) to focus on specific areas in video * **Quality tuning**: Adjust `reranking_candidates` (1-8) to balance quality vs. latency * **Event detection**: Enable `predict_spans` for better isolation of non-ambient sounds * **WAV output**: High-quality WAV audio file with the isolated sound * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Music production**: Extract vocals from songs for remixes or karaoke tracks * **Podcast editing**: Isolate speech from background noise or music * **Film post-production**: Separate dialogue from ambient sounds for audio mixing * **Sound design**: Extract specific sound effects from video recordings * **Transcription services**: Clean up audio by isolating speech before transcription * **Instrument isolation**: Separate specific instruments from full band recordings ### Isolate audio with SAM Audio Submit an audio or video file with a text description of the sound to isolate. The service returns a task ID for async polling or webhook notification.
Create a new audio isolation task List all audio isolation tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------- | --------- | -------- | ------- | ------------------------------------------------------------------------------------- | | `description` | `string` | Yes | - | Text description of the sound to isolate (e.g., "A person speaking", "Piano playing") | | `audio` | `string` | No\* | - | URL or base64-encoded audio file (WAV, MP3, FLAC, OGG, M4A) | | `video` | `string` | No\* | - | URL or base64-encoded video file (MP4, MOV, WEBM, AVI) | | `x1` | `integer` | No | `0` | Bounding box left coordinate for video localization (0 = full frame) | | `y1` | `integer` | No | `0` | Bounding box top coordinate for video localization (0 = full frame) | | `x2` | `integer` | No | `0` | Bounding box right coordinate for video localization (0 = full frame) | | `y2` | `integer` | No | `0` | Bounding box bottom coordinate for video localization (0 = full frame) | | `sample_fps` | `integer` | No | `2` | Frame sampling rate for video (1-5 FPS) | | `reranking_candidates` | `integer` | No | `1` | Quality vs. latency trade-off (1-8, higher = better quality, slower) | | `predict_spans` | `boolean` | No | `false` | Enable for better isolation of non-ambient, event-based sounds | | `webhook_url` | `string` | No | - | URL for task completion notification | \*Either `audio` or `video` must be provided, but not both. ## Frequently Asked Questions SAM Audio is an AI-powered audio isolation API that uses text descriptions to identify and extract specific sounds from audio or video files. You submit a file with a description of the target sound (e.g., "A person speaking"), receive a task ID immediately, then poll for results or receive a webhook notification. The output is a WAV file containing only the isolated sound. For audio input: WAV, MP3, FLAC, OGG, and M4A formats. For video input: MP4, MOV, WEBM, and AVI formats. Files can be provided as URLs or base64-encoded strings. Be specific and descriptive. Good examples: "A person speaking", "Piano playing in the background", "Dog barking loudly", "Acoustic guitar strumming". Avoid vague descriptions like "music" or "noise" - instead specify what type of music or sound you want to isolate. The `reranking_candidates` parameter (1-8) controls the quality vs. speed trade-off. Higher values produce better isolation quality but take longer to process. Use `1` for fastest results, `8` for highest quality. Default is `1`. Enable `predict_spans` when isolating non-ambient, event-based sounds like speech, individual notes, or sound effects. Keep it disabled (default) for continuous ambient sounds like background music or environmental noise. For video input, you can specify a bounding box (`x1`, `y1`, `x2`, `y2`) to focus on sounds originating from a specific area of the frame. This is useful when you want to isolate audio from a particular person or object in the video. Set all values to `0` (default) to process the full frame. SAM Audio outputs a high-quality WAV audio file containing only the isolated sound. This uncompressed format is ideal for further editing or processing in audio production workflows. ## Best practices * **Description specificity**: Use detailed descriptions for better isolation accuracy * **Input quality**: Higher quality input audio/video produces better isolation results * **Quality tuning**: Start with `reranking_candidates=1` for testing, increase for production * **Event sounds**: Enable `predict_spans` for speech, music notes, or sound effects * **Video focus**: Use bounding boxes to isolate sounds from specific video regions * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Sound Effects](/api-reference/sound-effects/post-sound-effects)**: Generate sound effects from text descriptions * **[Lip Sync](/api-reference/lip-sync/latent-sync/overview)**: Synchronize lip movements with audio * **[OmniHuman 1.5](/api-reference/video/omni-human-1-5)**: Generate human animations driven by audio # Audio Isolation - Get task status Source: https://docs.magnific.com/api-reference/audio-isolation/task-by-id get /v1/ai/audio-isolation/{task-id} Get the status of one audio isolation task # Search enterprise audit-log events Source: https://docs.magnific.com/api-reference/audit-logs/events get /v1/analytics/audit-logs/events Search your enterprise audit-log events. Filter by actor, group, folder, event type, request, or date window, and paginate with page or cursor. # Audit Logs API | Magnific API Source: https://docs.magnific.com/api-reference/audit-logs/overview Pull your enterprise audit trail programmatically with the Magnific Audit Logs API. Track who did what, filter events, and stream your full history via cursor. Programmatic, admin/owner-only access to the same audit events available in your Enterprise dashboard — who did what, to which resource, and how access changed. The Audit Logs API lets you retrieve your enterprise team's audit events over HTTP, so you can stream them into a SIEM, build compliance reports, or power your own security monitoring. It exposes the same activity feed as the dashboard, projected to a customer-safe set of fields. Available exclusively for **Enterprise** plans with **audit logs enabled**, and only to team **admins and owners**. It does not consume credits. ## Key capabilities * **Complete activity feed**: every governance, project, and member action your team produces — role changes, shares, permission grants, API-key lifecycle, and more. * **Rich filtering**: narrow by actor (`member_id`), group (`group_id`), folder (`folder_id`), event type (`event_type`), a single request (`request_id`), or a date window. * **Cursor pagination**: walk your entire history reliably with an opaque `cursor` — no offset limits, no skipped or duplicated rows. * **Server-side name resolution**: project and folder names are resolved for you, so you don't have to join against other endpoints. * **Customer-safe payloads**: the free-form event data is narrowed to a vetted allow-list — provider internals, cross-system identifiers, and moderation flags never leave your boundary. * **No credit cost**: reading your own audit trail is free. ## Use cases * **SIEM / SOC ingestion**: forward audit events into Splunk, Elastic, or Datadog for centralized security monitoring. * **Compliance & audit reporting**: produce evidence of access changes and administrative actions for SOC 2, ISO 27001, or internal reviews. * **Access reviews**: reconstruct who granted which permission, to whom, and when. * **Incident investigation**: pivot on a `request_id` to pull every event emitted by a single operation. * **Anomaly detection**: stream events into your own pipeline to flag unusual role changes or bulk shares. ## Read audit events Fetch events with [`GET /v1/analytics/audit-logs/events`](/api-reference/audit-logs/events). Authenticate with your API key — your enterprise team and identity are resolved automatically. ```bash theme={null} curl "https://api.magnific.com/v1/analytics/audit-logs/events?date_from=2026-07-16&date_to=2026-07-31&per_page=100" \ -H "X-Magnific-API-Key: YOUR_API_KEY" \ -H "Accept: application/json" ``` ### Example response ```json theme={null} { "data": [ { "event_id": "b3d1f0a2-9c4e-4a1b-8f2d-1a2b3c4d5e6f", "date": "2026-07-30T09:14:52+00:00", "action": { "event_type": "team.member.role_changed", "verb": "update", "status": "success", "resource_id": "18422", "request_id": "req-7f3a9c1e", "correlation_id": "corr-2b8e", "old_state": { "role": "member" }, "new_state": { "role": "admin" }, "duration_ms": 42, "data": { "target_user_id": 10511077, "role": "admin", "operation": "grant" }, "project_name": null, "folder_name": null }, "member": { "type": "user", "user_id": 10499812, "email": "admin@company.com" }, "category": "member", "source": { "project": "magnific", "platform": "web", "referer": null } } ], "meta": { "pagination": { "current_page": 1, "per_page": 20, "last_page": 5, "total": 92 }, "next_cursor": "eyJzIjpbMTcyMjMyOF19" } } ``` ### Event structure Each event describes one action, the actor behind it, and where it came from. | Field | Description | | ---------- | ------------------------------------------------------------------------------------------------------------------------------------ | | `event_id` | Unique identifier of the event | | `date` | When it occurred (UTC, ISO-8601) | | `action` | What happened: `event_type`, `verb`, `status`, `old_state`/`new_state`, `duration_ms`, and a `data` payload with event-specific keys | | `member` | The actor: `type`, `user_id`, `email` | | `category` | The resource category the event belongs to | | `source` | Where it originated: `project`, `platform`, `referer` | The `action.data`, `action.old_state`, and `action.new_state` objects are free-form. On the API they are narrowed to a vetted allow-list — provider/inference internals, cross-system identifiers, internal billing state, and content-moderation flags are stripped and never returned. ### Filtering Combine any of these query parameters. Multi-value filters accept a single value, a comma-separated list (`member_id=1,2`), or repeated params (`member_id[]=1&member_id[]=2`). | Filter | What it does | | ----------------------- | --------------------------------------------------------------------------------------- | | `date_from` / `date_to` | Restrict to a date window (`YYYY-MM-DD`). Max **15 days**; defaults to the last 15 days | | `member_id` | Only events performed by specific actor user IDs | | `group_id` | Only events involving specific team groups | | `folder_id` | Only events on specific folder references | | `event_type` | Only specific event types | | `request_id` | Every event emitted by one request (incident pivots) | ### Pagination Two modes, depending on what you need: Use `page` and `per_page` (max 100) for browsing. `page * per_page` cannot exceed 10000 — narrow the date range or switch to a cursor for deeper reads. Pass `cursor` to iterate with no depth limit. `page` is ignored. Follow `meta.next_cursor` until it is `null` to walk your whole history. To sync your full trail, request the first page, then keep calling with the returned `meta.next_cursor` until it comes back `null`. ## Authentication Authenticate with the `X-Magnific-API-Key` header. Your enterprise team and identity are resolved automatically from the API key. The key must belong to a team **admin or owner** of an **Enterprise** team with **audit logs enabled**. See [Authentication](/authentication) for details. ## Retention The audit index retains at most **15 days** of events. Queries cannot span a wider window. For long-term retention, stream events into your own storage on a schedule. ## Rate limits The Audit Logs API is limited to **50,000 requests per day** per API key, with a short-term burst cap. Use cursor pagination and incremental syncs to stay well within budget. ## API reference
Search and stream your enterprise audit events with filters and cursor pagination
## Best practices * **Sync incrementally**: after a full backfill, poll a small recent window (e.g. the last hour) on a schedule instead of re-reading everything. * **Use the cursor for backfills**: page mode is capped at 10000 results; the cursor has no depth limit and never skips rows. * **Persist your own copy**: retention is 15 days — export to your storage if you need history beyond that. * **Pivot on `request_id`**: when investigating an incident, grab one event's `request_id` and re-query to see every related event. * **Handle 429**: back off and retry when you hit the daily or burst limit. ## Frequently Asked Questions It's a read-only HTTP endpoint that returns your enterprise team's audit events — who did what, to which resource, and how access changed. It exposes the same activity feed as the Enterprise dashboard, projected to a customer-safe set of fields. Only team **admins and owners** of an **Enterprise** team that has **audit logs enabled**. The caller is resolved automatically from the API key; if the key's owner is not an admin/owner, the request is rejected. The audit index retains at most 15 days. A single query cannot span more than 15 days, and it defaults to the last 15 days when no dates are provided. To keep history longer, export events into your own storage. Use cursor pagination. Request the first page, then keep calling with the `meta.next_cursor` value from the previous response until it returns `null`. Cursor mode has no depth limit and won't skip or duplicate rows. The event `data`, `old_state`, and `new_state` objects are free-form. On the public API they are narrowed to a vetted allow-list, so provider internals, cross-system identifiers, internal billing state, and content-moderation flags are stripped. The dashboard and CSV export still show the full object. No. Reading your own audit trail is included with your Enterprise plan at no additional cost. 50,000 requests per day per API key, plus a short-term burst cap. See [Rate Limits](/ratelimits) for details. ## Related APIs * **[Analytics API](/api-reference/analytics/overview)**: credit consumption over time, plus team member, API key, group, and project discovery endpoints. # AI Image Classifier - Detection API | Magnific API Source: https://docs.magnific.com/api-reference/classifier/overview Detect AI-generated images with the AI Image Classifier API. Get probability scores for authentic vs synthetic content. Instant results for content moderation workflows. Powered by advanced machine learning, this API analyzes images to determine the probability of AI generation with high accuracy. The AI Image Classifier is a detection API that analyzes images to determine whether they were created by artificial intelligence or are authentic photographs. Submit any image and receive a confidence score indicating the probability of AI generation vs. natural origin. The API returns instant results, making it ideal for real-time content moderation and verification workflows. ### Key capabilities * **Binary classification**: Returns probability scores for both `ai` and `not_ai` categories * **High accuracy detection**: Advanced ML model trained on diverse AI-generated and authentic images * **Multiple input formats**: Accepts base64-encoded images, URLs, or binary data * **Instant response**: Synchronous API with immediate results (no task polling required) * **Confidence scoring**: Probability values from 0 to 1 for precise threshold-based decisions * **Format support**: Works with common image formats (JPEG, PNG, WebP, GIF) ### Use cases * **Content moderation**: Automatically flag potentially AI-generated uploads on user platforms * **Editorial verification**: Verify image authenticity before publication in news or media * **Stock image curation**: Filter AI-generated content from authentic photography collections * **Social media compliance**: Detect synthetic content for platform policy enforcement * **Academic integrity**: Identify AI-generated images in research or educational submissions * **Legal and forensic analysis**: Support authenticity verification for evidence documentation ### Classify images with AI Image Classifier Submit an image to analyze whether it was generated by AI. The API returns probability scores instantly without requiring task polling.
Analyze an image for AI-generation probability
### Parameters | Parameter | Type | Required | Default | Description | | --------- | -------- | -------- | ------- | ------------------------------------------------------------ | | `image` | `string` | Yes | - | Image to analyze: base64-encoded string, URL, or binary data | ### Response The API returns an array of classification results with probability scores: | Field | Type | Description | | ------------- | -------- | ------------------------------------------------------ | | `class_name` | `string` | Classification category: `ai` or `not_ai` | | `probability` | `number` | Confidence score from 0 to 1 (higher = more confident) | **Example response:** ```json theme={null} { "data": [ { "class_name": "not_ai", "probability": 0.9489 }, { "class_name": "ai", "probability": 0.0511 } ] } ``` ## Frequently Asked Questions The AI Image Classifier is a detection API that uses machine learning to analyze visual patterns in images that are characteristic of AI generation. It examines features like texture consistency, artifact patterns, and statistical properties to determine whether an image was created by AI tools (like Midjourney, DALL-E, or Stable Diffusion) or is an authentic photograph. The API returns probability scores for both categories. The classifier accepts common image formats including JPEG, PNG, WebP, and GIF. Images can be provided as base64-encoded strings, direct URLs, or binary data in the request body. The API returns two probability scores that sum to approximately 1.0. A `not_ai` probability of 0.95 means the model is 95% confident the image is authentic. For content moderation, you can set thresholds based on your requirements - for example, flag images where `ai` probability exceeds 0.7 for review. The classifier is trained on a diverse dataset of AI-generated images from various models and authentic photographs. Accuracy depends on the image type and generation method. The probability scores help you make threshold-based decisions appropriate for your use case. For critical applications, consider combining with manual review. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and how to handle rate-limited requests. See the [Pricing page](/pricing) for current rates and subscription options. The classifier is optimized for detecting fully AI-generated images. Partially edited images (AI inpainting on real photos, filters, or composites) may produce mixed results. For best accuracy, use on complete images rather than crops or heavily processed versions. ## Best practices * **Image quality**: Submit high-resolution images when possible for more accurate analysis * **Threshold tuning**: Set classification thresholds based on your false-positive tolerance * **Batch processing**: For high-volume workflows, implement request queuing to respect rate limits * **Human review**: Use API scores to prioritize manual review rather than as sole decision criteria * **Error handling**: Implement retry logic with exponential backoff for 503 errors * **Caching**: Cache results for identical images to reduce API calls and costs ## Related APIs * **[Mystic](/api-reference/mystic/mystic)**: Generate high-quality AI images with Magnific's proprietary model * **[Image Upscaler](/api-reference/image-upscaler-creative/image-upscaler)**: Enhance image resolution with AI upscaling * **[Background Remover](/api-reference/remove-background/overview)**: Remove or replace image backgrounds # Analyzes an image to determine its likelihood of being AI-generated Source: https://docs.magnific.com/api-reference/classifier/post-ai-classifier post /v1/ai/classifier/image Accepts an image file as input and analyzes it to determine the probability that the image was generated by artificial intelligence, providing a confidence score. # Creations API Source: https://docs.magnific.com/api-reference/creations/overview Retrieve your most recent AI creations programmatically with the Magnific Creations API. Programmatic access to the most recent creations of the user behind your API key. The Creations API lets you retrieve a user's most recent creations — the files they have generated — enriched with the metadata of the creation that produced each one. Use it to surface a user's latest work in your own app, dashboards, or galleries. It does not consume credits. ## Recent creations Fetch the most recent creations with [`GET /v1/creations/recent`](/api-reference/creations/recent-creations). Results are paginated and ordered from newest to oldest. The identity is resolved automatically from the API key — you always get the creations of that key's user. ```bash theme={null} curl https://api.magnific.com/v1/creations/recent?per_page=3 \ -H "X-Magnific-API-Key: YOUR_API_KEY" ``` ### Example response ```json theme={null} { "data": [ { "id": 1337, "reference": "4fef9987-75d0-44d1-b336-65f42f883bff", "user_id": 159327, "name": "My awesome creation", "external_id": "8421", "tool_name": "text-to-image", "thumbnail": { "url": "https://cdn.freepik.com/thumbnails/8421.jpg" }, "created_at": "2026-05-01T12:30:00Z", "updated_at": "2026-05-01T12:30:00Z", "creation": { "id": 8421, "identifier": "a1b2c3d4", "family": "image", "tool": "text-to-image", "url": "https://cdn.freepik.com/creations/8421.png", "preview": "https://cdn.freepik.com/creations/8421-preview.png", "status": "completed", "public": false, "nsfw": false, "user_id": 159327, "created_at": "2026-05-01T12:30:00Z" } } ], "meta": { "pagination": { "current_page": 1, "per_page": 3 } } } ``` ### Query parameters | Parameter | Description | Default | | ----------- | -------------------------------------------------------------------- | ------- | | `page` | Page number to retrieve | `1` | | `per_page` | Number of creations per page | `5` | | `tool_name` | Filter by the tool that produced the creation (e.g. `text-to-image`) | — | ## Authentication Authenticate with the `X-Magnific-API-Key` header. The user is identified automatically from the API key — you cannot request another user's creations. See [Authentication](/authentication) for details. ## Rate limits The endpoint is limited to **30,000 requests per minute** (30,000 RPM) per API key. Exceeding it returns `429 Too Many Requests`. ## API reference
List the user's most recent creations, newest first
## Frequently Asked Questions Always the creations of the user associated with the API key. Identity is resolved server-side from the key, so a client cannot request another user's creations. No. The Creations API is a read-only endpoint and does not consume credits. Use the `page` and `per_page` query parameters. The response `meta.pagination` block reflects the current page and page size. Yes. Pass `tool_name` (for example `text-to-image`) to return only the creations produced by that tool. # Get the user's most recent creations Source: https://docs.magnific.com/api-reference/creations/recent-creations get /v1/creations/recent Retrieve the most recent creations of the user associated with your API key, enriched with each creation's metadata. Paginated, newest first. # Flows - Get flow definition Source: https://docs.magnific.com/api-reference/flows/get-flow get /v1/ai/flows/{flow-id} Returns the full definition of a flow, including its dynamic inputs and cost. **Inputs** define what values the caller must provide when running the flow: - `creation` type: accepts an image/media URL, base64-encoded data, or an existing creation identifier - `text` type: accepts a plain text string (prompt, label, etc.) - `number` type: accepts a numeric value - `select` type: accepts a value from predefined options Use each input's `api_key` field (recommended, human-friendly node name) as the identifier when calling the run endpoint. The UUID-based `id` field is still accepted for backward compatibility. The `tool_metadata.total_cost` field indicates the credit cost per execution. # Flows - Get run status Source: https://docs.magnific.com/api-reference/flows/get-run get /v1/ai/flows/runs/{run-id} Retrieve the current status and results of a flow execution run. Use this endpoint to **poll** the status of a run after calling the `POST /v1/ai/flows/{flow-id}/run` endpoint. The `run_id` path parameter corresponds to the `workflow_run_identifier` returned in the run response. **Polling strategy:** - Wait 2-3 seconds after starting the run before the first poll - Poll every 3-5 seconds - When `status` is `completed`, the `result` field contains download URLs - When `status` is `failed`, the `error_message` field describes the failure **Asset URLs** in the `result` field are temporary and valid for **12 hours**. **Statuses:** | Status | Description | |--------|-------------| | `pending` | Run is queued but has not started yet | | `running` | Flow is executing | | `completed` | Finished successfully, `result` is available | | `completed_with_errors` | Execution finished but some nodes failed | | `failed` | Execution failed, see `error_message` | | `cancelled` | Execution was cancelled | # Flows - List flows Source: https://docs.magnific.com/api-reference/flows/list-flows get /v1/ai/flows Returns all flows **published as tools** that the authenticated user can execute. Each item includes the flow identifier, name, and tool metadata. Flows are visual AI pipelines built in [Magnific Spaces](https://www.magnific.com/spaces). Each flow chains together multiple AI tools (upscale, generate, edit, etc.) and can be executed via API. **Use this endpoint to discover** runnable flows, then call the detail endpoint to get input definitions before running them. # Flows - List my flows Source: https://docs.magnific.com/api-reference/flows/list-my-flows get /v1/ai/me/flows Returns all flows **owned** by the authenticated user, including drafts and unpublished workflows. Unlike `GET /v1/ai/flows` which returns only flows published as tools, this endpoint returns every flow the user has created — regardless of whether it has been published as a tool. Each item includes the flow identifier and name. # Flows API Source: https://docs.magnific.com/api-reference/flows/overview Run visual AI pipelines via API. Chain together AI tools like upscale, generate, and edit into reusable Flows with dynamic inputs. Execute multi-step AI workflows with a single API call. Each Flow chains together AI tools into a reusable pipeline. Flows are visual AI pipelines created in [Magnific Spaces](https://www.magnific.com/spaces). Each Flow chains together multiple AI tools (upscale, generate, edit, transform, etc.) into a reusable pipeline. Once published, Flows can be executed programmatically via this API — you provide the required inputs (images, text prompts, etc.) and the Flow runs the entire pipeline, returning the final output. ### Key capabilities * **Dynamic inputs**: Each Flow defines its own inputs — images, text prompts, numbers, or selections * **Async execution**: Submit a run, then poll for status or receive a webhook notification * **Multi-output**: Flows can produce images, videos, or audio depending on the workflow ### How it works 1. **List flows** to discover available workflows 2. **Get flow details** to see required inputs and their types 3. **Run the flow** by providing values for each input, keyed by its `api_key` (recommended) or `id` (legacy) 4. **Poll or webhook** to get the results when execution completes
List all available flows List flows you own, including drafts Get a flow definition and its inputs Run a flow with your inputs Poll run status and results
### Input types | Input Type | Accepted Values | Example | | ---------- | ------------------------------------------------------- | --------------------------------- | | `creation` | Image URL, base64-encoded image, or creation identifier | `"https://example.com/photo.jpg"` | | `text` | Plain text string | `"A car in the forest"` | | `number` | Numeric value | `0.5` | | `select` | Value from predefined options | `"style_a"` | ### Example: Run a flow ```bash theme={null} curl -X POST "https://api.magnific.com/v1/ai/flows/uqzQLDr2Aw/run" \ -H "X-Magnific-Api-Key: YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "inputs": { "image": "https://example.com/my-sketch.jpg" }, "webhook": "https://your-server.com/webhook" }' ``` ### Identifiers Flows are identified by their SQID: | Format | Example | Description | | -------- | ------------ | ------------------------------------- | | **SQID** | `uqzQLDr2Aw` | Short unique identifier for each flow | Existing integrations that call the legacy `/v1/ai/apps` paths remain functional. New integrations should use the `/v1/ai/flows` endpoints documented here. ## More about Spaces Magnific Spaces — node-based AI workflow canvas [Magnific Spaces](https://www.magnific.com/spaces) is the infinite canvas where you build and publish Flows. Connect AI nodes visually — image generation, upscaling, style transfer, text processing, and more — into reusable workflows. Collaborate with your team in real time, then publish your workflow as a Flow to make it available via this API. Start building your own AI workflows on the infinite canvas. View your published Flows and copy their IDs to use with this API. ## Frequently Asked Questions Flows are visual AI pipelines built in Magnific Spaces. They chain together multiple AI tools (image generation, upscaling, style transfer, etc.) into a single reusable workflow. The API lets you execute these workflows programmatically. Call `GET /v1/ai/flows/{flow-id}` to get the full definition, including all required inputs with their types and labels. Use each input's `api_key` field (recommended) — a human-friendly identifier matching the node name — as the key in the `inputs` object when running the flow. The UUID-based `id` field is still accepted for backward compatibility (legacy). Execution time depends on the workflow complexity and the AI tools involved. Simple single-tool workflows may complete in 10-30 seconds. Multi-step workflows with video generation can take 1-5 minutes. Use webhooks for production integrations. See the [Rate Limits](/ratelimits) page for current limits by subscription tier. Each flow shows its credit cost in the `tool_metadata.total_cost` field. See the [Pricing page](/pricing) for credit rates. ## Related APIs * **[Image Upscaler](/api-reference/image-upscaler-creative/image-upscaler)**: Upscale images with AI enhancement * **[Text to Image](/api-reference/text-to-image/image-generation)**: Generate images from text prompts * **[Image to Video](/api-reference/image-to-video/overview)**: Convert images into video clips # Flows - Run a flow Source: https://docs.magnific.com/api-reference/flows/run-flow post /v1/ai/flows/{flow-id}/run Triggers the execution of a flow. Provide values for all required inputs defined in the flow's definition (see the GET endpoint). The `inputs` object maps each input's `api_key` (recommended) or its `id` (UUID, legacy) to its value: | Input Type | Accepted Values | |------------|----------------| | `creation` | Image URL (`https://...`), base64-encoded image string, or existing creation identifier | | `text` | Plain text string | | `number` | Numeric value | | `select` | Value from allowed options | **This endpoint is asynchronous.** It returns immediately with `status: "running"` and a `workflow_run_identifier`. Provide a `webhook` URL to receive notifications when the execution completes. The webhook receives two events: `initialized` (run started) and `finished`/`failed` (run completed). The webhook payload for a completed run includes a `result` object containing `images`, `videos`, and `audios` arrays with download URLs for the generated assets. # AI Icon Generation - Text-to-Icon API | Magnific API Source: https://docs.magnific.com/api-reference/icon-generation/overview Generate custom icons from text prompts with AI. Create PNG or SVG icons in 5 styles: solid, outline, color, flat, and sticker. Production-ready vector output for apps and websites. Generate custom icons from text descriptions using advanced AI models, with support for multiple styles and vector output formats. AI Icon Generation is a text-to-icon API that creates custom icons from natural language prompts. Describe the icon you need, and receive production-ready PNG or SVG files in your chosen style. The API supports 5 distinct visual styles and offers a preview workflow for rapid iteration before final rendering. ### Key capabilities * **Text-to-icon generation**: Create icons from natural language descriptions (e.g., "A cute cat icon", "Shopping cart with heart") * **Multiple output formats**: Export as PNG for raster graphics or SVG for scalable vector graphics * **5 visual styles**: Choose from `solid`, `outline`, `color`, `flat`, or `sticker` styles * **Preview workflow**: Generate quick previews before committing to final high-quality renders * **Quality tuning**: Adjust `num_inference_steps` (10-50) and `guidance_scale` (0-10) for fine control * **Async processing**: Webhook notifications for task completion * **Vector-ready output**: SVG format for infinite scalability without quality loss ### Use cases * **Mobile app development**: Generate consistent icon sets for iOS and Android applications * **Web design**: Create custom icons matching your brand identity and design system * **UI/UX prototyping**: Rapidly iterate on icon concepts during design exploration * **Marketing materials**: Generate unique icons for presentations, infographics, and social media * **Brand identity**: Create distinctive iconography that aligns with brand guidelines * **E-commerce**: Generate product category icons, feature badges, and navigation elements ### Generate icons with the API Create icons by submitting a text prompt with your desired style and format. Use the preview endpoint for quick iterations, then render the final output in your preferred format.
Generate a new icon from text prompt Generate a quick preview of the icon Download icon in PNG or SVG format
### Parameters | Parameter | Type | Required | Default | Description | | --------------------- | --------- | -------- | ------- | -------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the icon to generate (e.g., "A cute cat icon") | | `webhook_url` | `string` | Yes | - | URL for task completion notification | | `style` | `string` | No | `solid` | Visual style: `solid`, `outline`, `color`, `flat`, or `sticker` | | `format` | `string` | No | `png` | Output format: `png` or `svg` | | `num_inference_steps` | `integer` | No | `10` | Generation quality steps (10-50, higher = better quality, slower) | | `guidance_scale` | `number` | No | `7` | Prompt adherence strength (0-10, higher = stricter prompt following) | ### Style reference | Style | Description | Best for | | --------- | ---------------------------------- | -------------------------------- | | `solid` | Filled shapes with single color | UI icons, navigation elements | | `outline` | Line-based icons with strokes | Minimalist designs, light themes | | `color` | Multi-colored filled icons | Marketing, illustrations | | `flat` | Simplified shapes with flat colors | Modern app interfaces | | `sticker` | Decorative style with playful look | Social media, casual apps | ## Frequently Asked Questions AI Icon Generation is a text-to-icon API that uses advanced AI models to create custom icons from natural language descriptions. You submit a prompt describing the icon you want (e.g., "A shopping cart icon"), choose a style and format, and receive a task ID. Poll for results or receive a webhook notification when the icon is ready. The output is a production-ready PNG or SVG file. The API supports two output formats: PNG for raster graphics and SVG for scalable vector graphics. PNG is ideal for immediate use in web and mobile applications, while SVG provides infinite scalability without quality loss, perfect for responsive designs and print materials. Five visual styles are available: `solid` (filled shapes), `outline` (line-based), `color` (multi-colored), `flat` (simplified modern), and `sticker` (decorative playful). Each style suits different design contexts - solid and outline work well for UI elements, while color and sticker are better for marketing and casual applications. Be specific and descriptive. Good examples: "A cute cat icon", "Shopping cart with a heart", "Lightning bolt in a circle", "Envelope with notification badge". Include the subject and any distinctive features. Avoid overly complex descriptions - icons work best as simple, recognizable symbols. The preview endpoint (`/preview`) generates a quick draft for rapid iteration during the design process. The main generate endpoint (`/v1/ai/text-to-icon`) creates the production-ready icon. Use previews to explore different prompts and styles, then generate the final version once satisfied. `num_inference_steps` (10-50) controls generation quality - higher values produce more refined icons but take longer. `guidance_scale` (0-10) determines how strictly the AI follows your prompt - higher values create icons more closely matching your description, while lower values allow more creative interpretation. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and quotas. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Prompt clarity**: Use clear, concise descriptions focusing on the core concept of the icon * **Style consistency**: Use the same style across icon sets for visual coherence in your application * **Preview first**: Use the preview endpoint to iterate quickly before generating final renders * **SVG for scalability**: Choose SVG format when icons will be used at multiple sizes * **Quality vs. speed**: Start with default `num_inference_steps` (10) for testing, increase for production * **Guidance tuning**: Use higher `guidance_scale` (8-10) for literal interpretations, lower (4-6) for creative variations * **Webhook integration**: Use webhooks instead of polling for production applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Magnific Mystic](/api-reference/mystic/mystic)**: Generate full illustrations and images from text prompts * **[Image Upscaler](/api-reference/image-upscaler-creative/image-upscaler)**: Enhance icon resolution for high-DPI displays * **[Background Removal](/api-reference/remove-background/overview)**: Remove backgrounds from existing icons or images # AI Icon generation Source: https://docs.magnific.com/api-reference/icon-generation/post-generate-icon post /v1/ai/text-to-icon Create stunning icons in different styles and formats (png, svg) from text prompts using our advanced AI models. ## Description Generate an icon from a text prompt. # AI Icon preview generation Source: https://docs.magnific.com/api-reference/icon-generation/post-preview post /v1/ai/text-to-icon/preview Create stunning previews icons in different styles and formats (png, svg) from text prompts using our advanced AI models. # Download an icon preview Source: https://docs.magnific.com/api-reference/icon-generation/post-{format}-by-id post /v1/ai/text-to-icon/{task-id}/render/{format} Download the generated AI icon in the specified format png or svg. # Download an icon Source: https://docs.magnific.com/api-reference/icons/download-an-icon get /v1/icons/{id}/download Download the specified icon by its unique ID in the requested format and size. Download the original icon asset. See the [Icons API overview](/api-reference/icons/icons-api) for authentication and usage guidance. # Search and filter icons by specified order Source: https://docs.magnific.com/api-reference/icons/get-all-icons-by-order get /v1/icons Get a list of icons based on the provided parameters and ordering criteria. Use the Magnific Icons API to list icons ordered by relevance, popularity, or recency. See the [Icons API overview](/api-reference/icons/icons-api) for authentication and usage guidance. # Get detailed icon information by ID Source: https://docs.magnific.com/api-reference/icons/get-one-icon-by-id get /v1/icons/{id} Get detailed information about a specific icon identified by its unique ID. Fetch a specific icon by ID. See the [Icons API overview](/api-reference/icons/icons-api) for authentication and usage guidance. # Icons API Source: https://docs.magnific.com/api-reference/icons/icons-api Search, retrieve, and download icons programmatically with the Magnific Icons API. The Magnific Icons API lets you find and retrieve icons from our catalog with AI-powered search and download the assets for use in your applications. ### What you can do * List icons with sorting and filters * Get a single icon by ID * Download an icon asset ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication). ### Endpoints
List icons with sorting and filters Retrieve a single icon by ID Download an icon asset
### Use cases * Build an icon picker into your design tool or CMS * AI-powered search by keyword and sort by popularity * Programmatically download icons during build/deploy ### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to icons? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). # Ideogram Image Edit - List tasks Source: https://docs.magnific.com/api-reference/ideogram-image-edit/get-ideogram-image-edit get /v1/ai/ideogram-image-edit Get the status of all Ideogram Image Edit tasks # Ideogram Image Edit - Get task status Source: https://docs.magnific.com/api-reference/ideogram-image-edit/get-{task-id}-by-id get /v1/ai/ideogram-image-edit/{task-id} Get the status of a specific Ideogram Image Edit task # Ideogram Inpainting API Source: https://docs.magnific.com/api-reference/ideogram-image-edit/overview Edit specific areas of an image using Ideogram AI inpainting. Mask-based editing with multiple rendering speeds, MagicPrompt enhancement, style and character references. Powered by Ideogram AI, this API edits specific areas of an image using mask-based inpainting with prompt-guided generation. Ideogram Inpainting is an AI-powered image editing API that lets you modify specific regions of an image using a mask and a text prompt. Provide a black-and-white mask where black regions indicate the areas to edit, along with a prompt describing the desired changes. The API supports multiple rendering speeds (TURBO, DEFAULT, QUALITY), automatic prompt enhancement via MagicPrompt, and style/character reference images for consistent results. ### Key capabilities * **Mask-based editing**: Use a black-and-white mask to precisely define which areas of the image to modify * **Multiple rendering speeds**: Choose between TURBO (fastest), DEFAULT (balanced), or QUALITY (highest quality) * **MagicPrompt**: Automatically enhance your prompt for better results (AUTO, ON, or OFF) * **Style customization**: Apply style codes, style types (AUTO, GENERAL, REALISTIC, DESIGN), and style reference images * **Character consistency**: Use character reference images to maintain consistent characters across edits * **Color palette control**: Guide the color palette of generated content * **Reproducible results**: Optional seed parameter (0-2147483647) for consistent output across requests * **Flexible input**: Accepts HTTPS URLs or base64-encoded images (max 10MB per image) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Object replacement**: Replace objects in a scene with AI-generated alternatives * **Background editing**: Modify or replace backgrounds while preserving subjects * **Content removal**: Remove unwanted elements from images seamlessly * **Creative retouching**: Edit specific areas with creative prompts for artistic effects * **Product photography**: Modify product backgrounds or surroundings for e-commerce * **Character editing**: Maintain character consistency across multiple edited images ### Edit images with Ideogram Inpainting Submit an image with a mask and prompt to edit specific regions. The service returns a task ID for async polling or webhook notification.
Create a new inpainting task List all inpainting tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | --------- | -------- | --------- | ----------------------------------------------------------------------------------------------------------------------------- | | `image` | `string` | Yes | - | Image to edit. HTTPS URL or base64-encoded (JPEG, WebP, PNG, max 10MB) | | `mask` | `string` | Yes | - | Black-and-white mask image (same size as input). Black regions indicate areas to edit. HTTPS URL or base64-encoded (max 10MB) | | `prompt` | `string` | Yes | - | Text describing the desired changes to the image | | `rendering_speed` | `string` | No | `DEFAULT` | Rendering speed: `TURBO`, `DEFAULT`, or `QUALITY` | | `magic_prompt` | `string` | No | - | MagicPrompt enhancement: `AUTO`, `ON`, or `OFF` | | `style_type` | `string` | No | - | Style type: `AUTO`, `GENERAL`, `REALISTIC`, or `DESIGN` | | `style_codes` | `array` | No | - | List of style codes for image generation | | `style_reference_images` | `array` | No | - | Images to use as style references (URLs or base64) | | `character_reference_images` | `array` | No | - | Images for character consistency (URLs or base64) | | `color_palette` | `object` | No | - | Color palette to guide generated content | | `seed` | `integer` | No | random | Seed for reproducibility (0-2147483647) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Ideogram Inpainting is an AI-powered image editing API. You provide an image, a black-and-white mask indicating which areas to edit, and a text prompt describing the desired changes. The API generates new content in the masked regions that blends seamlessly with the rest of the image. You receive a task ID immediately, then poll for results or receive a webhook notification. The mask must be the same size as the input image. It should be a black-and-white image where black regions indicate the areas you want to edit. White regions will remain unchanged. Supported formats are JPEG, WebP, and PNG (max 10MB). There are three rendering speeds: **TURBO** is the fastest with lower quality, **DEFAULT** provides balanced speed and quality, and **QUALITY** is slower but produces the highest quality results. The default is DEFAULT. MagicPrompt automatically enhances your text prompt for better generation results. Set it to `AUTO` to let the model decide, `ON` to always use it, or `OFF` to disable it. This can significantly improve output quality with minimal effort. Yes. Use the `character_reference_images` parameter to provide reference images of the character you want to maintain. Combined with `style_type`, this helps ensure consistent characters across multiple edits. The API accepts JPEG, WebP, and PNG images up to 10MB each. Both HTTPS URLs and base64-encoded strings are supported for images, masks, and reference images. The output is returned as a downloadable image URL. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Precise masks**: Create clean, well-defined masks for the best results. Avoid feathered or blurry edges * **Descriptive prompts**: Be specific about what you want in the edited region (e.g., "red roses in a garden" instead of just "flowers") * **MagicPrompt for quick results**: Use `magic_prompt: "AUTO"` to let the model enhance your prompt automatically * **Rendering speed tradeoffs**: Use TURBO for quick previews, DEFAULT for production, and QUALITY for final outputs * **Style references**: Provide style reference images when you need a specific visual style across edits * **Seed for consistency**: Use the same seed value when you need reproducible results across multiple API calls * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Ideogram Image Expand](/api-reference/image-expand/ideogram/overview)**: Expand images beyond their boundaries with Ideogram * **[Image Upscaler](/api-reference/image-upscaler-creative/post-image-upscaler)**: Enhance image resolution and quality * **[Remove Background](/api-reference/remove-background/post-beta-remove-background)**: Remove image backgrounds for compositing * **[Reimagine Flux](/api-reference/text-to-image/reimagine-flux/post-reimagine-flux)**: Reimagine images with Flux # Ideogram Image Edit - Edit an image using inpainting Source: https://docs.magnific.com/api-reference/ideogram-image-edit/post-ideogram-image-edit post /v1/ai/ideogram-image-edit Edit an image using Ideogram AI's inpainting capabilities. Provide an image and a mask to specify the areas to edit, along with a prompt describing the desired changes. **Key features:** - Inpainting: Edit specific areas of an image using a mask - Multiple rendering speeds: TURBO, DEFAULT, or QUALITY - MagicPrompt: Automatically enhance your prompt for better results - Style customization: Use style codes, style types, and reference images - Character reference: Use reference images to maintain character consistency **Supported formats:** JPEG, WebP, PNG (max 10MB each) # Change Camera - Transform image perspective Source: https://docs.magnific.com/api-reference/image-change-camera/change-camera post /v1/ai/image-change-camera Transform an image by changing the camera angle using AI. Adjust horizontal rotation (0-360 degrees), vertical tilt (-30 to 90 degrees), and zoom level (0-10) to generate a new image as if the camera had been repositioned around the subject. This is an asynchronous endpoint. After submitting a request, use the task ID to poll for results or provide a `webhook_url` to receive a notification when processing completes. **Camera controls:** - **Horizontal angle** (`horizontal_angle`): Rotate the viewpoint 0-360 degrees around the subject. `0` = front view, `90` = right side, `180` = back view, `270` = left side. - **Vertical angle** (`vertical_angle`): Tilt the camera from -30 (looking up) to 90 (bird's eye view). `0` = eye level. - **Zoom** (`zoom`): Adjust from `0` (wide shot, full scene) to `10` (close-up). **Use cases:** Product photography with multiple angle views, architectural visualization, creative image manipulation, and generating consistent multi-angle views of objects and scenes. # Change Camera - List tasks Source: https://docs.magnific.com/api-reference/image-change-camera/change-camera-tasks get /v1/ai/image-change-camera Retrieve the status of all Change Camera tasks for the authenticated user. Returns a list of tasks with their current status, creation time, and result URLs for completed tasks. # Change Camera API Source: https://docs.magnific.com/api-reference/image-change-camera/overview Change camera angle and perspective of any image with the Change Camera API. Adjust horizontal rotation (0-360), vertical tilt (-30 to 90), and zoom (0-10) for multi-angle views. Transform any image by repositioning the virtual camera with precise control over horizontal rotation, vertical tilt, and zoom level. Change Camera is an image editing API that transforms the camera perspective of any image using AI. Provide an input image and specify horizontal rotation (0-360 degrees), vertical tilt (-30 to 90 degrees), and zoom level (0-10) to generate a new image as if the camera had been physically repositioned around the subject. The API accepts JPG, PNG, and WebP images via publicly accessible HTTPS URLs and outputs in PNG or JPEG format. Processing is asynchronous with support for both polling and webhook notifications. ### Key capabilities * **360-degree horizontal rotation**: Rotate the viewpoint around the subject from 0 (front) through 90 (right side), 180 (back), 270 (left side), to 360 degrees * **Vertical tilt control**: Tilt the camera from -30 degrees (looking up) through 0 (eye level) to 90 degrees (bird's eye view) * **Adjustable zoom**: Control the distance from wide shot (`0`) through standard framing (`5`) to close-up (`10`) * **Output format options**: Choose between lossless PNG or compressed JPEG output * **Reproducible results**: Optional seed parameter for consistent output across requests * **Async processing**: Webhook notifications or polling for task completion ### Camera controls reference | Control | Parameter | Range | Default | Key positions | | ----------------------- | ------------------ | --------- | ------- | ------------------------------------------------ | | **Horizontal rotation** | `horizontal_angle` | 0-360 | 0 | `0` front, `90` right, `180` back, `270` left | | **Vertical tilt** | `vertical_angle` | -30 to 90 | 0 | `-30` looking up, `0` eye level, `90` bird's eye | | **Zoom** | `zoom` | 0-10 | 5 | `0` wide shot, `5` medium, `10` close-up | ### Use cases * **Product photography**: Generate multiple angle views of a product from a single photo for e-commerce listings * **Architectural visualization**: View buildings and interiors from different perspectives without reshooting * **Creative image manipulation**: Explore alternative viewpoints of scenes, portraits, and objects * **3D asset previewing**: Create multi-angle previews from a single reference image * **Marketing materials**: Produce varied camera angles for advertising campaigns from one source image * **Content creation**: Generate consistent multi-angle views of subjects for social media and editorial content ### Transform images with Change Camera Submit an image URL with camera angle parameters to create a new transformation task. The service returns a task ID for async polling or webhook notification.
Create a new camera angle transformation task List all Change Camera tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------ | --------- | -------- | ------- | ------------------------------------------------------------------------------ | | `image` | `string` | Yes | - | Input image URL (publicly accessible HTTPS). Supported formats: JPG, PNG, WebP | | `horizontal_angle` | `integer` | No | `0` | Horizontal rotation: `0` (front) to `360` degrees around the subject | | `vertical_angle` | `integer` | No | `0` | Vertical tilt: `-30` (looking up) to `90` (bird's eye view) degrees | | `zoom` | `integer` | No | `5` | Zoom level: `0` (wide shot) to `10` (close-up) | | `output_format` | `string` | No | `png` | Output format: `png` (lossless) or `jpeg` (compressed) | | `seed` | `integer` | No | random | Seed for reproducibility (minimum: 1) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions The Change Camera API transforms the camera perspective of any image using AI. You submit an image URL with desired camera angle parameters (horizontal rotation, vertical tilt, and zoom), receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The AI generates a new image as if the camera had been repositioned around the subject. The Change Camera API accepts images via publicly accessible HTTPS URLs. Supported input formats are JPG, PNG, and WebP. Output is available in PNG (lossless, default) or JPEG (compressed) format, controlled by the `output_format` parameter. The `horizontal_angle` parameter rotates the camera around the subject from 0 to 360 degrees. Key positions: `0` is the front view, `90` is the right side, `180` is the back view, `270` is the left side, and `360` returns to the front (equivalent to `0`). The default is `0` (front view). The `vertical_angle` parameter tilts the camera up or down relative to the subject. Values range from `-30` (looking up at the subject from below) through `0` (eye level, default) to `90` (bird's eye view looking straight down). A value of `45` provides a moderate downward angle. Yes. Use the `seed` parameter with the same value across requests. Combined with identical image URLs and camera parameters, the API produces consistent output. This is useful for fine-tuning angles iteratively. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Start with defaults**: Begin with default values (`horizontal_angle=0`, `vertical_angle=0`, `zoom=5`) and adjust one parameter at a time to understand the effect * **Subtle angle changes**: Small adjustments (15-30 degrees horizontal, 10-20 degrees vertical) often produce the most realistic results * **Input quality**: Use high-resolution, well-lit images for best perspective transformation quality * **Seed for iteration**: Use the `seed` parameter when fine-tuning angles to isolate the effect of each parameter change * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Image Upscaler](/api-reference/image-upscaler-precision-v2/overview)**: Enhance image resolution before or after camera angle transformation * **[Style Transfer](/api-reference/image-style-transfer/image-styletransfer)**: Apply artistic styles to images after changing the perspective * **[Remove Background](/api-reference/remove-background/overview)**: Remove backgrounds for clean subject isolation before camera transformation # Change Camera - Get task status Source: https://docs.magnific.com/api-reference/image-change-camera/task-by-id get /v1/ai/image-change-camera/{task-id} Retrieve the status and result of a specific Change Camera task by its task ID. When the task status is `completed`, the response includes the URL of the generated image with the new camera angle. # Get the status of all image expand tasks Source: https://docs.magnific.com/api-reference/image-expand/get-flux-pro get /v1/ai/image-expand/flux-pro # Get the status of all image expand seedream v4.5 tasks Source: https://docs.magnific.com/api-reference/image-expand/get-seedream-v4-5 get /v1/ai/image-expand/seedream-v4-5 # Get the status of one image expand seedream v4.5 task Source: https://docs.magnific.com/api-reference/image-expand/get-seedream-v4-5-task get /v1/ai/image-expand/seedream-v4-5/{task-id} # Get the status of one image expand task Source: https://docs.magnific.com/api-reference/image-expand/get-{task-id}-by-id get /v1/ai/image-expand/flux-pro/{task-id} # Ideogram - Expand image Source: https://docs.magnific.com/api-reference/image-expand/ideogram/expand-image post /v1/ai/image-expand/ideogram This endpoint allows you to expand an image using the Ideogram AI model. The image will be expanded based on the provided pixel values for each side. If no prompt is provided, the model will auto-generate one based on the image content. # Get the status of all image expand ideogram tasks Source: https://docs.magnific.com/api-reference/image-expand/ideogram/ideogram-tasks get /v1/ai/image-expand/ideogram # Ideogram Image Expand API Source: https://docs.magnific.com/api-reference/image-expand/ideogram/overview Expand images beyond their original boundaries with Ideogram Image Expand. AI-powered outpainting with directional control up to 2048 pixels per side and optional prompt guidance. Powered by Ideogram AI, this API expands images beyond their original boundaries with intelligent content generation that blends seamlessly with the original. Ideogram Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries using Ideogram's generative model. Specify how many pixels to add on each side (left, right, top, bottom) up to 2048 pixels, and optionally provide a text prompt to guide the generated content. If no prompt is provided, the model auto-generates one based on the image content. The API produces seamless expansions that maintain visual consistency with the original image, making it ideal for adapting images to different aspect ratios or adding creative space around subjects. ### Key capabilities * **Directional expansion**: Independently control expansion on each edge (left, right, top, bottom) from 0 to 2048 pixels * **Auto-prompt generation**: Automatically generates a prompt from the image content when no prompt is provided * **Prompt-guided generation**: Optional text descriptions to guide what appears in expanded areas * **Seamless blending**: AI-generated content matches the style, lighting, and composition of the original image * **Reproducible results**: Optional seed parameter (0-2147483647) for consistent output across requests * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Social media adaptation**: Convert landscape images to portrait format (or vice versa) for different platforms * **Print production**: Add bleed area or extend backgrounds for large-format printing * **Design workflows**: Create additional canvas space for text overlays, logos, or design elements * **Photography enhancement**: Extend cropped photos to recover lost composition * **E-commerce**: Expand product images to fit standardized dimensions * **Marketing materials**: Adapt hero images to various banner sizes and aspect ratios ### Expand images with Ideogram Submit an image with expansion parameters to extend it in any direction. The service returns a task ID for async polling or webhook notification.
Create a new image expansion task List all Ideogram image expansion tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | -------------- | ------------------------------------------------------------------------------------------------------------------------ | | `image` | `string` | Yes | - | Base64-encoded image to expand | | `left` | `integer` | Yes | - | Pixels to expand on the left side (0-2048) | | `right` | `integer` | Yes | - | Pixels to expand on the right side (0-2048) | | `top` | `integer` | Yes | - | Pixels to expand on the top side (0-2048) | | `bottom` | `integer` | Yes | - | Pixels to expand on the bottom side (0-2048) | | `prompt` | `string` | No | auto-generated | Text prompt describing the desired expanded content. If not provided, the AI auto-generates a prompt based on the image. | | `seed` | `integer` | No | random | Seed for reproducibility (0-2147483647) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Ideogram Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries. You submit a base64-encoded image with pixel expansion values for each edge (left, right, top, bottom), receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The API generates new content that seamlessly blends with the original image. You can expand up to 2048 pixels on each edge (left, right, top, bottom) independently. All four directional parameters (left, right, top, bottom) are required. Set any edge to 0 if you do not want to expand in that direction. If no prompt is provided, the Ideogram model automatically generates a prompt based on the image content. This auto-prompt feature analyzes the existing image and produces contextually appropriate content for the expanded areas. The API accepts base64-encoded images. Common formats like JPEG, PNG, and WebP can be encoded and submitted. The output is returned as a downloadable image URL. Yes. Use the `seed` parameter with the same value (0-2147483647) across requests to generate reproducible results. Combined with identical images, expansion values, and prompts, you will get consistent outputs. Both APIs extend images beyond their original boundaries. Ideogram Image Expand features auto-prompt generation when no prompt is provided and requires all four directional parameters. Flux Pro Image Expand uses optional directional parameters with default values of 0. Choose based on your workflow preferences and output quality requirements. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Balanced expansion**: For best results, avoid extreme asymmetric expansions; gradual extensions maintain better visual coherence * **Prompt specificity**: Use descriptive prompts when you need specific content in expanded areas (e.g., "sunset sky with orange clouds" instead of just "sky") * **Auto-prompt for natural results**: Omit the prompt parameter to let the model analyze the image and generate contextually appropriate content * **Image quality**: Higher resolution input images produce better expansion results * **Seed for consistency**: Use the same seed value when you need reproducible results across multiple API calls * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Flux Pro Image Expand](/api-reference/image-expand/overview)**: Alternative image expansion API powered by Flux Pro * **[Image Upscaler](/api-reference/image-upscaler-creative/post-image-upscaler)**: Enhance image resolution and quality * **[Remove Background](/api-reference/remove-background/post-beta-remove-background)**: Remove image backgrounds for compositing # Get the status of one image expand ideogram task Source: https://docs.magnific.com/api-reference/image-expand/ideogram/task-by-id get /v1/ai/image-expand/ideogram/{task-id} # Flux Pro Image Expand - Outpainting API | Magnific API Source: https://docs.magnific.com/api-reference/image-expand/overview Expand images beyond their original boundaries with Flux Pro Image Expand. AI-powered outpainting with directional control up to 2048 pixels per side. Perfect for social media, print, and design workflows. Powered by Flux Pro, this API expands images beyond their original boundaries using AI-generated content that seamlessly blends with the original. Flux Pro Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries. Specify how many pixels to add on each side (left, right, top, bottom) up to 2048 pixels, and optionally provide a text prompt to guide the generated content. The API produces seamless expansions that maintain visual consistency with the original image, making it ideal for adapting images to different aspect ratios or adding creative space around subjects. ### Key capabilities * **Directional expansion**: Independently control expansion on each edge (left, right, top, bottom) from 0 to 2048 pixels * **Prompt-guided generation**: Optional text descriptions to guide what appears in expanded areas * **Seamless blending**: AI-generated content matches the style, lighting, and composition of the original image * **Flexible input**: Accept base64-encoded images for easy integration * **High-resolution output**: Support for substantial canvas extensions while maintaining quality * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Social media adaptation**: Convert landscape images to portrait format (or vice versa) for different platforms * **Print production**: Add bleed area or extend backgrounds for large-format printing * **Design workflows**: Create additional canvas space for text overlays, logos, or design elements * **Photography enhancement**: Extend cropped photos to recover lost composition * **E-commerce**: Expand product images to fit standardized dimensions * **Marketing materials**: Adapt hero images to various banner sizes and aspect ratios ### Expand images with Flux Pro Submit an image with expansion parameters to extend it in any direction. The service returns a task ID for async polling or webhook notification.
Create a new image expansion task List all image expansion tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ------- | -------------------------------------------------------------------------------------- | | `image` | `string` | Yes | - | Base64-encoded image to expand | | `prompt` | `string` | No | - | Text description to guide the expanded areas (e.g., "sunset sky", "forest background") | | `left` | `integer` | No | `0` | Pixels to expand on the left edge (0-2048) | | `right` | `integer` | No | `0` | Pixels to expand on the right edge (0-2048) | | `top` | `integer` | No | `0` | Pixels to expand on the top edge (0-2048) | | `bottom` | `integer` | No | `0` | Pixels to expand on the bottom edge (0-2048) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Flux Pro Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries. You submit a base64-encoded image with expansion values for each edge (left, right, top, bottom), receive a task ID immediately, then poll for results or receive a webhook notification. The API generates new content that seamlessly blends with the original image. You can expand up to 2048 pixels on each edge (left, right, top, bottom) independently. Set any edge to 0 if you do not want to expand in that direction. For example, to only extend the width, set `left` and `right` values while keeping `top` and `bottom` at 0. The API accepts base64-encoded images. Common formats like JPEG, PNG, and WebP can be encoded and submitted. The output is returned as a downloadable image URL. The optional `prompt` parameter guides what the AI generates in the expanded areas. For example, "blue sky with clouds" will generate sky content, while "brick wall" will generate a wall background. If omitted, the AI infers appropriate content from the existing image context. Yes, Flux Pro analyzes your image's style, lighting, color palette, and composition to generate expansions that blend seamlessly. The AI maintains visual consistency so the final image appears natural and cohesive. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Balanced expansion**: For best results, avoid extreme asymmetric expansions; gradual extensions maintain better visual coherence * **Prompt specificity**: Use descriptive prompts when you need specific content in expanded areas (e.g., "ocean waves" instead of just "water") * **Image quality**: Higher resolution input images produce better expansion results * **Edge content**: Ensure the edges of your original image have enough context for the AI to continue naturally * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Flux 2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: Generate images from text descriptions * **[Flux Kontext Pro](/api-reference/text-to-image/flux-kontext-pro/overview)**: Text-to-image with context understanding and image guidance * **[Image Upscaler](/api-reference/image-upscaler-creative/post-image-upscaler)**: Enhance image resolution and quality * **[Remove Background](/api-reference/remove-background/post-beta-remove-background)**: Remove image backgrounds for compositing # Flux Pro - Expand image Source: https://docs.magnific.com/api-reference/image-expand/post-flux-pro post /v1/ai/image-expand/flux-pro This endpoint allows you to expand an image using the AI Flux Pro model. The image will be expanded based on the provided parameters. # Seedream V4.5 - Expand image Source: https://docs.magnific.com/api-reference/image-expand/post-seedream-v4-5 post /v1/ai/image-expand/seedream-v4-5 This endpoint allows you to expand an image using the Seedream V4.5 AI model. The image will be expanded based on the provided pixel values for each side. If no prompt is provided, the model will auto-generate one based on the image content. # Seedream V4.5 Image Expand API Source: https://docs.magnific.com/api-reference/image-expand/seedream-v4-5-overview Expand images beyond their original boundaries with Seedream V4.5. AI-powered outpainting with directional control up to 2048 pixels per side, auto-prompt generation, and optional seed for reproducibility. Powered by Seedream V4.5, this API expands images beyond their original boundaries with intelligent content generation that blends seamlessly with the original. Seedream V4.5 Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries using the Seedream V4.5 generative model. Specify how many pixels to add on each side (left, right, top, bottom) up to 2048 pixels, and optionally provide a text prompt to guide the generated content. If no prompt is provided, the model auto-generates one based on the image content. The API produces seamless expansions that maintain visual consistency with the original image, making it ideal for adapting images to different aspect ratios or adding creative space around subjects. ### Key capabilities * **Directional expansion**: Independently control expansion on each edge (left, right, top, bottom) from 0 to 2048 pixels * **Auto-prompt generation**: Automatically generates a prompt from the image content when no prompt is provided * **Prompt-guided generation**: Optional text descriptions to guide what appears in expanded areas * **Seamless blending**: AI-generated content matches the style, lighting, and composition of the original image * **Reproducible results**: Optional seed parameter (0-2147483647) for consistent output across requests * **Flexible input**: Accepts HTTPS URLs or base64-encoded images (max 10MB) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Social media adaptation**: Convert landscape images to portrait format (or vice versa) for different platforms * **Print production**: Add bleed area or extend backgrounds for large-format printing * **Design workflows**: Create additional canvas space for text overlays, logos, or design elements * **Photography enhancement**: Extend cropped photos to recover lost composition * **E-commerce**: Expand product images to fit standardized dimensions * **Marketing materials**: Adapt hero images to various banner sizes and aspect ratios ### Expand images with Seedream V4.5 Submit an image with expansion parameters to extend it in any direction. The service returns a task ID for async polling or webhook notification.
Create a new image expansion task List all Seedream V4.5 image expansion tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | -------------- | ------------------------------------------------------------------------------------------------------------------------ | | `image` | `string` | Yes | - | Image to expand. Accepts an HTTPS URL or a base64-encoded image (max 10MB) | | `left` | `integer` | Yes | - | Pixels to expand on the left side (0-2048) | | `right` | `integer` | Yes | - | Pixels to expand on the right side (0-2048) | | `top` | `integer` | Yes | - | Pixels to expand on the top side (0-2048) | | `bottom` | `integer` | Yes | - | Pixels to expand on the bottom side (0-2048) | | `prompt` | `string` | No | auto-generated | Text prompt describing the desired expanded content. If not provided, the AI auto-generates a prompt based on the image. | | `seed` | `integer` | No | random | Seed for reproducibility (0-2147483647) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Seedream V4.5 Image Expand is an AI-powered outpainting API that extends images beyond their original boundaries. You submit an image (URL or base64) with pixel expansion values for each edge (left, right, top, bottom), receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The API generates new content that seamlessly blends with the original image. You can expand up to 2048 pixels on each edge (left, right, top, bottom) independently. All four directional parameters (left, right, top, bottom) are required. Set any edge to 0 if you do not want to expand in that direction. If no prompt is provided, the Seedream V4.5 model automatically generates a prompt based on the image content. This auto-prompt feature analyzes the existing image and produces contextually appropriate content for the expanded areas. The API accepts HTTPS URLs or base64-encoded images up to 10MB. Common formats like JPEG, PNG, and WebP are supported. The output is returned as a downloadable image URL. Yes. Use the `seed` parameter with the same value (0-2147483647) across requests to generate reproducible results. Combined with identical images, expansion values, and prompts, you will get consistent outputs. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Balanced expansion**: For best results, avoid extreme asymmetric expansions; gradual extensions maintain better visual coherence * **Prompt specificity**: Use descriptive prompts when you need specific content in expanded areas (e.g., "sunset sky with orange clouds" instead of just "sky") * **Auto-prompt for natural results**: Omit the prompt parameter to let the model analyze the image and generate contextually appropriate content * **Image quality**: Higher resolution input images produce better expansion results * **Seed for consistency**: Use the same seed value when you need reproducible results across multiple API calls * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Flux Pro Image Expand](/api-reference/image-expand/overview)**: Alternative image expansion API powered by Flux Pro * **[Ideogram Image Expand](/api-reference/image-expand/ideogram/overview)**: Alternative image expansion API powered by Ideogram * **[Image Upscaler](/api-reference/image-upscaler-creative/post-image-upscaler)**: Enhance image resolution and quality * **[Remove Background](/api-reference/remove-background/post-beta-remove-background)**: Remove image backgrounds for compositing # Relight - Get task status Source: https://docs.magnific.com/api-reference/image-relight/get-image-relight get /v1/ai/image-relight/{task-id} Get the status of the relight task # Relight - List tasks Source: https://docs.magnific.com/api-reference/image-relight/get-image-relight-task get /v1/ai/image-relight Get the status of all relight tasks # Overview Source: https://docs.magnific.com/api-reference/image-relight/image-relight
Original scene Night scene Snowy landscape Hell scene
Image Relight allows you to change the entire lighting of a scene and, optionally, the background using: 1. A prompt 2. A reference image 3. A light map This tool can simulate different lighting scenarios, enhance details, and create artistic effects. Our API lets you transform your images with custom prompts and fine-tuned parameters for impressive results. Image Relight uses Magnific.ai technology, now available as a comprehensive API service. ### Use Cases The Image Relight can be applied to a wide range of image types and industries, including: * Portraits: Change lighting to enhance facial features * Product photography: Adjust lighting for better presentation * Architectural renderings: Simulate different times of day * Film stills: Create dramatic lighting effects * Graphic design: Add creative lighting to designs * Nature photography: Enhance natural light conditions ### Selecting Light Transfer Method Choose one of these three ways to transfer light in Image Relight: 1. **Prompt** Describe the desired lighting using text. Specify characteristics like color, time of day, type and shape of light. Example: "A sunlit forest clearing at golden hour" or "Under the water" 2. **Reference Image** Provide an image to transfer lighting from. Image Relight will replicate the light, colors, and shadows from this reference. 3. **Lightmap** Create custom lighting with your own lightmap image: * Black represents absence of light * Lighter colors (including white) indicate presence, color, and shape of light sources
Relight an image using AI Get the status of a relighting task Get the status of all relighting tasks
### Image Input Best Practices For maximum quality results, how you send your image matters: | Action | Quality Impact | | ------------------------------------------ | -------------------------- | | Send URL of original image | ✅ Maximum quality | | Send base64 of original file read directly | ✅ Maximum quality | | Use `canvas.toDataURL('image/jpeg')` | ❌ \~8% quality loss | | Use `canvas.toDataURL('image/jpeg', 0.8)` | ❌ \~20% quality loss | | Resize image before sending | ❌ Significant quality loss | | Convert PNG → JPEG before sending | ❌ Quality loss | **Recommended:** Send the original image via URL whenever possible. This ensures the AI receives the highest quality input, which directly impacts output quality. ### Frequently Asked Questions Yes! Magnific.ai is part of Magnific and we are happy to offer the same technology in an API service. If you tried Magnific.ai before, you will get the same technology and its future updates. The cost per image relight is a fixed cost of €0.10 per operation. This ensures that you have complete transparency and control over your usage. # Relight - Adjust image lighting Source: https://docs.magnific.com/api-reference/image-relight/post-image-relight post /v1/ai/image-relight Relight an image using AI. This endpoint accepts a variety of parameters to customize the generated images. ## Important Upscaler endpoints are only available for premium API users. You can upgrade your account [here](https://www.magnific.com/api#pricing). ## Request # Style Transfer - List tasks Source: https://docs.magnific.com/api-reference/image-style-transfer/get-image-style-transfer get /v1/ai/image-style-transfer Get the status of all Style Transfer tasks # Style Transfer - Get task status Source: https://docs.magnific.com/api-reference/image-style-transfer/get-image-style-transfer-task get /v1/ai/image-style-transfer/{task-id} Get the status of the Style Transfer task # Overview Source: https://docs.magnific.com/api-reference/image-style-transfer/image-styletransfer
Original image Black and white Wild nature Wood and brick
Image Style Transfer is a powerful AI-driven tool that allows you to transform any image by applying various artistic styles. This technology goes beyond simple filters, enabling you to control the amount of style transferred and maintain structural integrity. With our API, you can: 1. Apply predefined artistic styles 2. Use custom images as style references 3. Fine-tune the style transfer process with advanced parameters 4. Transform images while preserving important details 5. Combine style transfer with upscaling for high-resolution results Whether you're a professional artist, game developer, or just looking to have fun with your photos, Image Style Transfer opens up a world of creative possibilities. ### Use Cases Image Style Transfer can be applied to various industries and creative projects, some examples are: * **Video Game Development**: Transform concept art and create stylized game assets * **Film and VFX**: Create dramatic lighting effects and stylized scenes * **Interior Design**: Visualize room makeovers with different design styles * **3D Rendering**: Texturize and light 3D renders using reference images * **Sketch Transformation**: Convert sketches into detailed, styled artwork ### Key Features * **Control Over Style Intensity**: Adjust the strength of the style transfer to find the perfect balance between the original image and the new style. * **Structure Preservation**: Maintain important details and structural elements of the original image while applying new styles. * **Prompt-Guided Transformations**: Use text prompts to further refine and direct the style transfer process. * **Multiple Style Engines**: Choose from various AI engines optimized for different types of style transfers. By leveraging the power of AI, Image Style Transfer enables creators to push the boundaries of visual art and design, saving hours of manual work and opening up new realms of creative possibility.
API Reference: Image Style Transfer
### Image Input Best Practices For maximum quality results, how you send your image matters: | Action | Quality Impact | | ------------------------------------------ | -------------------------- | | Send URL of original image | ✅ Maximum quality | | Send base64 of original file read directly | ✅ Maximum quality | | Use `canvas.toDataURL('image/jpeg')` | ❌ \~8% quality loss | | Use `canvas.toDataURL('image/jpeg', 0.8)` | ❌ \~20% quality loss | | Resize image before sending | ❌ Significant quality loss | | Convert PNG → JPEG before sending | ❌ Quality loss | **Recommended:** Send the original image via URL whenever possible. This ensures the AI receives the highest quality input, which directly impacts output quality. ### Frequently Asked Questions Yes! Magnific.ai is part of Magnific and we are happy to offer the same technology in an API service. If you tried Magnific.ai before, you will get the same technology and its future updates. The cost per image style transfer is a fixed cost of €0.10 per operation. This ensures that you have complete transparency and control over your usage. # Style Transfer - Transform image style Source: https://docs.magnific.com/api-reference/image-style-transfer/post-image-style-transfer post /v1/ai/image-style-transfer # Image to Prompt - List tasks Source: https://docs.magnific.com/api-reference/image-to-prompt/get-image-to-prompt get /v1/ai/image-to-prompt Get the status of all image-to-prompt tasks # Image to Prompt - Get task status Source: https://docs.magnific.com/api-reference/image-to-prompt/get-{task-id}-by-id get /v1/ai/image-to-prompt/{task-id} Get the status of one image-to-prompt task # Image to Prompt - AI Image Analysis API | Magnific API Source: https://docs.magnific.com/api-reference/image-to-prompt/overview Generate descriptive prompts from images with Image to Prompt API. Extract detailed text descriptions for AI workflows. Perfect for prompt engineering and image cataloging. Analyze any image and generate detailed text prompts that describe its content, style, and composition. Image to Prompt is an AI-powered API that analyzes images and generates descriptive text prompts. Submit any image and receive a detailed text description suitable for use with AI image generation models. The API extracts visual elements, artistic styles, compositions, and contextual details to create prompts that can reproduce or inspire similar images. ### Key capabilities * **Automatic image analysis**: AI extracts subjects, objects, colors, lighting, and composition from images * **Prompt-ready output**: Generated descriptions formatted for direct use with text-to-image models * **Flexible input**: Accepts images via URL or base64-encoded string * **Style detection**: Identifies artistic styles, photography techniques, and visual aesthetics * **Detail extraction**: Captures fine details including textures, materials, and environmental elements * **Multi-format support**: Works with common image formats (JPEG, PNG, WebP) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Prompt engineering**: Reverse-engineer prompts from reference images to recreate similar styles * **Image cataloging**: Generate searchable text descriptions for image libraries * **AI workflow automation**: Bridge image-to-text pipelines for automated content creation * **Style transfer preparation**: Extract style descriptions for consistent AI image generation * **Content moderation**: Generate text descriptions for image review and classification * **Accessibility**: Create alt-text descriptions for images in web applications ### Generate prompts with Image to Prompt Submit an image via URL or base64-encoded string. The service returns a task ID for async polling or webhook notification.
Create a new image analysis task List all image-to-prompt tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | --------------------------------------------------------------------------------------------------- | | `image` | `string` | Yes | - | Input image for analysis. Accepts URL or base64-encoded string (e.g., `data:image/jpeg;base64,...`) | | `webhook_url` | `string` | No | - | URL for task completion notification | ### Response When the task completes successfully, the response includes: | Field | Type | Description | | ----------- | -------- | --------------------------------------------------------------- | | `task_id` | `string` | Unique identifier for the task (UUID format) | | `status` | `string` | Task status: `CREATED`, `IN_PROGRESS`, `COMPLETED`, or `FAILED` | | `generated` | `array` | Array containing the generated prompt text | ## Frequently Asked Questions Image to Prompt is an AI-powered API that analyzes images and generates text descriptions suitable for AI image generation. You submit an image (via URL or base64), receive a task ID immediately, then poll for results or receive a webhook notification. The output is a detailed text prompt describing the image content, style, and composition. Image to Prompt accepts common image formats including JPEG, PNG, and WebP. Images can be provided as publicly accessible URLs or as base64-encoded strings with the appropriate data URI prefix (e.g., `data:image/jpeg;base64,...`). The generated prompts capture multiple aspects of the image including subjects, objects, colors, lighting, composition, artistic style, and fine details like textures and materials. The descriptions are formatted for direct use with AI image generation models like Flux, Seedream, or Mystic. Yes, the prompts generated by Image to Prompt are designed to work with any text-to-image AI model. Use them directly with Magnific API models like [Flux](/api-reference/text-to-image/flux-kontext-pro/overview), [Seedream](/api-reference/text-to-image/seedream-4/overview), or [Mystic](/api-reference/mystic/mystic), or with external image generation services. Processing time varies based on image complexity and current system load. Typical analysis completes within a few seconds. For production workflows, use webhooks instead of polling to receive instant notifications when processing completes. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Higher resolution images with clear subjects produce more detailed and accurate prompts * **URL accessibility**: Ensure image URLs are publicly accessible without authentication * **Base64 encoding**: Include the proper data URI prefix when using base64 (e.g., `data:image/jpeg;base64,`) * **Production integration**: Use webhooks instead of polling for scalable applications * **Prompt refinement**: Use generated prompts as starting points and refine them for specific use cases * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Improve Prompt](/api-reference/improve-prompt/post-improve-prompt)**: Enhance and refine text prompts for better AI generation results * **[Flux Kontext Pro](/api-reference/text-to-image/flux-kontext-pro/overview)**: Generate images from text prompts with advanced context understanding * **[Seedream 4](/api-reference/text-to-image/seedream-4/overview)**: High-quality text-to-image generation with detailed prompt support * **[Mystic](/api-reference/mystic/mystic)**: Magnific's flagship image generation model with style customization # Image to Prompt - Generate prompt Source: https://docs.magnific.com/api-reference/image-to-prompt/post-image-to-prompt post /v1/ai/image-to-prompt Generate descriptive prompts from input images using AI analysis # Get all Veo 3.1 I2V tasks Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1 get /v1/ai/image-to-video/veo-3-1 # Get all Veo 3.1 I2V Fast tasks Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1-fast get /v1/ai/image-to-video/veo-3-1-fast # Get Veo 3.1 I2V Fast task by ID Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1-fast-task get /v1/ai/image-to-video/veo-3-1-fast/{task-id} # Get all Veo 3.1 Lite I2V tasks Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1-lite get /v1/ai/image-to-video/veo-3-1-lite # Get Veo 3.1 Lite I2V task by ID Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1-lite-task get /v1/ai/image-to-video/veo-3-1-lite/{task-id} # Get Veo 3.1 I2V task by ID Source: https://docs.magnific.com/api-reference/image-to-video/get-veo-3-1-task get /v1/ai/image-to-video/veo-3-1/{task-id} # Happy Horse 1.1 - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1-1/generate post /v1/ai/image-to-video/happy-horse-1-1 Animate a single image into a high-quality AI video using Happy Horse 1.1, Alibaba's image-to-video model. The image is used as the first frame and an optional text prompt guides motion and style. **Key features:** - 720P and 1080P resolution support - Duration: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 seconds - Optional prompt up to 2500 characters # Happy Horse 1.1 I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1-1/happy-horse-1-1-i2v-tasks get /v1/ai/image-to-video/happy-horse-1-1 Retrieve the list of all Happy Horse 1.1 image-to-video tasks for the authenticated user. # Happy Horse 1.1 Image To Video API Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1-1/overview Animate images into AI videos with Happy Horse 1.1. 720P and 1080P resolution, 3-15 second durations, prompt-guided motion, from Alibaba ATH. Happy Horse 1.1 Image-to-Video animates a single image into a high-quality video, the latest iteration of Alibaba ATH's Happy Horse family with strong motion fidelity and prompt adherence. Happy Horse 1.1 Image-to-Video is an AI video generation API that animates a single still image into an MP4 video. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), the model uses your image as the first frame and an optional text prompt to guide motion and style, producing videos at 720P (1280x720) or 1080P (1920x1080) resolution with durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: Latest iteration of the Alibaba ATH Happy Horse video family * **First-frame animation**: Use any supported image as the starting frame for the video * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Prompt-guided motion**: Optional prompt (up to 2500 characters) controls motion and style * **Wide image format support**: JPEG, JPG, PNG, BMP, WEBP (300px+ on each side, max 10 MB) * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Product animation**: Bring product imagery to life with smooth motion and camera movements * **Marketing videos**: Animate brand visuals and hero shots into short-form video content * **Social media content**: Create video posts from static images for TikTok, Instagram, and YouTube * **Storyboarding**: Animate concept art or wireframes to preview motion sequences * **Photography to motion**: Add subtle motion to still photos for cinemagraphs and immersive content * **Creative exploration**: Experiment with motion prompts on the same image for varied results ### API operations Generate videos by submitting an image URL and optional prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new image-to-video generation task List all Happy Horse 1.1 I2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | --------- | ------------------------------------------------------------------------------------------------------------------ | | `image_url` | `string` | Yes | - | URL of the image to use as the first frame (JPEG/JPG/PNG/BMP/WEBP, 300px+, max 10 MB, aspect ratio 1:2.5 to 2.5:1) | | `prompt` | `string` | No | - | Text prompt to guide motion and style. Max 2500 characters | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | When the task completes, the generated MP4 video URL is returned in `data.generated`. The URL is valid for 24 hours — download and persist it to permanent storage immediately. ## Frequently Asked Questions Happy Horse 1.1 Image-to-Video is an AI model from Alibaba ATH that animates a single image into an MP4 video. It uses your image as the first frame and applies natural motion guided by an optional text prompt. It is the latest iteration of the Happy Horse video family. Happy Horse 1.1 I2V accepts JPEG, JPG, PNG, BMP, and WEBP images via publicly accessible URLs. Each side must be at least 300 pixels, the aspect ratio must be between 1:2.5 and 2.5:1, and the file size must not exceed 10 MB. No. The `prompt` parameter is optional but recommended. Without a prompt, the model generates natural motion based on the image content. With a prompt, you can describe specific motions, camera movements, or style cues for better control. Happy Horse 1.1 I2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Allowed values: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15. The output aspect ratio is determined automatically by the dimensions of the input `image_url`. Provide an image with the aspect ratio you want for the output video. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use high-resolution images with clear subjects and balanced lighting. Avoid heavily compressed or noisy inputs. * **Aspect ratio**: Provide an image already cropped to the target output ratio — output ratio matches the input image. * **Prompt writing**: Even though prompts are optional, adding motion and camera directions ("subject walks forward, slow zoom in") improves results. * **Duration selection**: Start with shorter durations (3-5 seconds) for quick iterations, then increase for final outputs. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.1 Text-to-Video](/api-reference/text-to-video/happy-horse-1-1/overview)**: Generate videos purely from text prompts * **[Happy Horse 1.1 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1-1/overview)**: Generate videos from up to 9 character reference images * **[Happy Horse 1.0 Image-to-Video](/api-reference/image-to-video/happy-horse-1/overview)**: Previous Happy Horse iteration, ranked #1 on the Artificial Analysis Video Arena (April 2026) # Happy Horse 1.1 I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1-1/task-by-id get /v1/ai/image-to-video/happy-horse-1-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.1 image-to-video task by its ID. # Happy Horse 1.0 - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1/generate post /v1/ai/image-to-video/happy-horse-1 Animate a single image into a high-quality AI video using Happy Horse 1.0, Alibaba's image-to-video model. The image is used as the first frame and an optional text prompt guides motion and style. **Key features:** - 720P and 1080P resolution support - Duration: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 seconds - Optional prompt up to 2500 characters # Happy Horse 1.0 I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1/happy-horse-1-i2v-tasks get /v1/ai/image-to-video/happy-horse-1 Retrieve the list of all Happy Horse 1.0 image-to-video tasks for the authenticated user. # Happy Horse 1.0 Image To Video API Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1/overview Animate images into AI videos with Happy Horse 1.0, ranked #1 on the Artificial Analysis Video Arena (April 2026). 720P and 1080P resolution, 3-15 second durations, prompt-guided motion. Happy Horse 1.0 Image-to-Video animates a single image into a high-quality video, with #1 ranking on the Artificial Analysis Video Arena (April 2026) for I2V quality and motion fidelity. Happy Horse 1.0 Image-to-Video is an AI video generation API that animates a single still image into an MP4 video. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), it currently ranks #1 on the Artificial Analysis Video Arena leaderboard (April 2026) for image-to-video. The model uses your image as the first frame and an optional text prompt to guide motion and style, producing videos at 720P (1280x720) or 1080P (1920x1080) resolution with durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: #1 on Artificial Analysis Video Arena (April 2026) for image-to-video * **First-frame animation**: Use any supported image as the starting frame for the video * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Prompt-guided motion**: Optional prompt (up to 2500 characters) controls motion and style * **Wide image format support**: JPEG, JPG, PNG, BMP, WEBP (300px+ on each side, max 10 MB) * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Product animation**: Bring product imagery to life with smooth motion and camera movements * **Marketing videos**: Animate brand visuals and hero shots into short-form video content * **Social media content**: Create video posts from static images for TikTok, Instagram, and YouTube * **Storyboarding**: Animate concept art or wireframes to preview motion sequences * **Photography to motion**: Add subtle motion to still photos for cinemagraphs and immersive content * **Creative exploration**: Experiment with motion prompts on the same image for varied results ### API operations Generate videos by submitting an image URL and optional prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new image-to-video generation task List all Happy Horse 1.0 I2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | --------- | ------------------------------------------------------------------------------------------------------------------ | | `image_url` | `string` | Yes | - | URL of the image to use as the first frame (JPEG/JPG/PNG/BMP/WEBP, 300px+, max 10 MB, aspect ratio 1:2.5 to 2.5:1) | | `prompt` | `string` | No | - | Text prompt to guide motion and style. Max 2500 characters | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Happy Horse 1.0 Image-to-Video is an AI model from Alibaba ATH that animates a single image into an MP4 video. It uses your image as the first frame and applies natural motion guided by an optional text prompt. As of April 2026, it ranks #1 on the Artificial Analysis Video Arena leaderboard for image-to-video quality. Happy Horse 1.0 I2V accepts JPEG, JPG, PNG, BMP, and WEBP images via publicly accessible URLs. Each side must be at least 300 pixels, the aspect ratio must be between 1:2.5 and 2.5:1, and the file size must not exceed 10 MB. No. The `prompt` parameter is optional but recommended. Without a prompt, the model generates natural motion based on the image content. With a prompt, you can describe specific motions, camera movements, or style cues for better control. Happy Horse 1.0 I2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Allowed values: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15. The output aspect ratio is determined automatically by the dimensions of the input `image_url`. Provide an image with the aspect ratio you want for the output video. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use high-resolution images with clear subjects and balanced lighting. Avoid heavily compressed or noisy inputs. * **Aspect ratio**: Provide an image already cropped to the target output ratio — output ratio matches the input image. * **Prompt writing**: Even though prompts are optional, adding motion and camera directions ("subject walks forward, slow zoom in") improves results. * **Duration selection**: Start with shorter durations (3-5 seconds) for quick iterations, then increase for final outputs. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.0 Text-to-Video](/api-reference/text-to-video/happy-horse-1/overview)**: Generate videos purely from text prompts * **[Happy Horse 1.0 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1/overview)**: Generate videos from up to 9 character reference images * **[Happy Horse 1.0 Video Edit](/api-reference/video-edit/happy-horse-1/overview)**: Edit videos with natural-language instructions * **[WAN 2.7 Image-to-Video](/api-reference/image-to-video/wan-2-7/overview)**: Alibaba WAN I2V with first+last frame and video continuation modes # Happy Horse 1.0 I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/happy-horse-1/task-by-id get /v1/ai/image-to-video/happy-horse-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.0 image-to-video task by its ID. # Kling Elements Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-pro/get-kling-elements-pro get /v1/ai/image-to-video/kling-elements-pro Get the list of the kling-elements-pro tasks # Kling Elements - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-pro/get-kling-elements-pro-task get /v1/ai/image-to-video/kling-elements/{task-id} Get the status of the kling-elements task # Kling Elements Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-pro/post-kling-elements-pro post /v1/ai/image-to-video/kling-elements-pro Generate a video from an image using the Kling Elements Pro model. ## Important The service allows up to 3 concurrent requests per user. # Kling Elements Standard - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-std/get-kling-elements-std get /v1/ai/image-to-video/kling-elements-std Get the list of the kling-elements-std tasks # Kling Elements - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-std/get-kling-elements-std-task get /v1/ai/image-to-video/kling-elements/{task-id} Get the status of the kling-elements task # Kling Elements Standard - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-elements-std/post-kling-elements-std post /v1/ai/image-to-video/kling-elements-std Generate a video from an image using the Kling Elements Std model. ## Important The service allows up to 3 concurrent requests per user. # Kling O1 Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1-pro post /v1/ai/image-to-video/kling-o1-pro # Kling O1 Pro - Create video with reference Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1-pro-video-reference post /v1/ai/image-to-video/kling-o1-pro-video-reference # Kling O1 Standard - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1-std post /v1/ai/image-to-video/kling-o1-std # Kling O1 Standard - Create video with reference Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1-std-video-reference post /v1/ai/image-to-video/kling-o1-std-video-reference # Kling O1 - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1-tasks get /v1/ai/image-to-video/kling-o1 Get the list of the kling-o1 tasks # Kling O1 – Image-to-Video API Source: https://docs.magnific.com/api-reference/image-to-video/kling-o1/overview Create cinematic videos with first-to-last frame interpolation using Kling O1. Smooth transitions, reference image support, and Pro or Standard quality modes for 5 or 10 second outputs. Generate smooth video sequences using first frame and last frame interpolation, with optional reference images for character and style consistency. Kling O1 is an image-to-video API that creates cinematic video sequences by interpolating between a first frame and last frame. It delivers smooth transitions with controlled start and end points, suitable for creative professionals and automated workflows. Choose between Pro quality for premium results or Standard for faster, cost-effective generation. ### Key capabilities * **Frame interpolation**: Generate smooth transitions between first and last frame images * **Reference image support**: Maintain character and style consistency with up to 7 reference images * **Dual quality modes**: Pro for professional-grade output, Standard for efficient generation * **Flexible durations**: 5-second or 10-second video outputs * **Multiple aspect ratios**: `16:9` (widescreen), `9:16` (vertical), `1:1` (square) * **Prompt guidance**: Text descriptions up to 2500 characters to guide motion and style * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Film and animation**: Create smooth scene transitions and animated sequences * **Marketing content**: Generate cinematic product reveals and brand videos * **Social media**: Produce vertical and square videos optimized for Instagram, TikTok, and YouTube Shorts * **E-commerce**: Animate static product images with controlled motion * **Storytelling**: Build narrative sequences with consistent character appearance * **Motion design**: Create abstract transitions and visual effects ### API endpoints Generate videos by submitting a request to one of the Kling O1 endpoints. The service returns a task ID for async polling or webhook notification.
Pro quality video from first/last frame Standard quality video from first/last frame Pro quality with reference images Standard quality with reference images List all Kling O1 tasks Get task status by ID
### Parameters #### Image-to-Video (first/last frame) | Parameter | Type | Required | Default | Description | | -------------- | --------- | -------- | ------- | --------------------------------------------------------------- | | `first_frame` | `string` | No\* | - | Starting image (URL or Base64). Min 300x300px, max 10MB | | `last_frame` | `string` | No\* | - | Ending image (URL or Base64). Same requirements as first\_frame | | `prompt` | `string` | No | - | Motion description up to 2500 characters | | `aspect_ratio` | `string` | No | `16:9` | Video ratio: `16:9`, `9:16`, or `1:1` | | `duration` | `integer` | No | `5` | Video length: `5` or `10` seconds | | `webhook_url` | `string` | No | - | URL for completion notification | \*At least one frame (first\_frame or last\_frame) is required. #### Video Reference | Parameter | Type | Required | Default | Description | | ------------------ | --------- | -------- | ------- | ---------------------------------------- | | `prompt` | `string` | Yes | - | Video description up to 2500 characters | | `reference_images` | `array` | No | - | Up to 7 reference images for consistency | | `aspect_ratio` | `string` | No | `16:9` | Video ratio: `16:9`, `9:16`, or `1:1` | | `duration` | `integer` | No | `5` | Video length: `5` or `10` seconds | | `webhook_url` | `string` | No | - | URL for completion notification | ## Frequently Asked Questions Kling O1 is an image-to-video model that generates smooth video sequences through frame interpolation. You provide a first frame, last frame, or both, and the model creates a cinematic transition between them. Optionally, add a text prompt to guide the motion style. Processing is asynchronous: submit a request, receive a task ID, then poll for results or receive a webhook notification. Pro mode generates higher-fidelity videos with better motion quality, detail retention, and temporal consistency. Standard mode offers faster processing at lower cost, suitable for prototyping or high-volume use cases. Both support the same parameters and output formats. Kling O1 accepts JPEG and PNG images via URL (publicly accessible) or Base64 encoding. Requirements: minimum resolution 300x300px, maximum file size 10MB, aspect ratio between 1:2.5 and 2.5:1. The video reference endpoints accept up to 7 reference images that help maintain character and style consistency. Use these when you need the same subject to appear consistently across multiple video generations or shots. Processing time varies based on duration and mode. Typical generation: 30-120 seconds for 5-second videos, longer for 10-second outputs. Pro mode takes longer than Standard. Use webhooks for production workflows instead of polling. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Pricing varies by mode (Pro vs Standard) and duration. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use high-resolution images (1080p or higher) with clear subjects and balanced lighting for best results * **Frame similarity**: When using both first and last frames, ensure similar compositions for smoother interpolation * **Prompt specificity**: Describe the desired motion clearly (camera movement, subject action, mood) rather than just the scene * **Reference consistency**: For video reference mode, use images with similar lighting and style for better character consistency * **Production integration**: Implement webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 (service unavailable) responses ## Related APIs * **[Kling 2.6](/api-reference/image-to-video/kling-v2-6-pro)**: Latest Kling model with motion control capabilities * **[Kling 2.5 Turbo Pro](/api-reference/image-to-video/kling-v2.5-pro/overview)**: Fast cinematic generation with 5-10 second outputs * **[Kling 2.1 Pro](/api-reference/image-to-video/kling-v2.1-pro/overview)**: High-fidelity motion with strong temporal consistency * **[Kling Elements](/api-reference/image-to-video/kling-elements-pro/post-kling-elements-pro)**: Element-based video generation for specific use cases # Kling 1.6 Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-pro/get-kling-pro get /v1/ai/image-to-video/kling-pro Get the list of the kling-pro tasks # Kling 1.6 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-pro/get-kling-pro-task get /v1/ai/image-to-video/kling/{task-id} Get the status of the kling task # Kling 1.6 Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-pro/post-kling-pro post /v1/ai/image-to-video/kling-pro Generate a video from an image using the Kling 1.6 Pro model. ## Important The service allows up to 3 concurrent requests per user. # Kling 1.6 Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-std/get-kling-std get /v1/ai/image-to-video/kling-std Get the list of the kling-pro tasks # Kling 1.6 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-std/get-kling-std-task get /v1/ai/image-to-video/kling/{task-id} Get the status of the kling task # Kling 1.6 Standard - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-std/post-kling-std post /v1/ai/image-to-video/kling-std Generate a video from an image using the Kling 1.6 Std model. ## Important The service allows up to 3 concurrent requests per user. # Kling 2.6 Pro - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2-6-pro post /v1/ai/image-to-video/kling-v2-6-pro # Kling 2.6 Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2-6-pro-tasks get /v1/ai/image-to-video/kling-v2-6 # Kling 2.6 Pro - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2-6/task-by-id get /v1/ai/image-to-video/kling-v2-6/{task-id} # Kling 2.1 Standard - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-master/get-kling-v2-1-master get /v1/ai/image-to-video/kling-v2-1-master Get the list of the kling-v2-1-std tasks # Kling 2.1 Master - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-master/get-kling-v2-1-master-task get /v1/ai/image-to-video/kling-v2-1-master/{task-id} Get the status of the kling-v2-1-master task # Kling 2.1 Master - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-master/post-kling-v2-1-master post /v1/ai/image-to-video/kling-v2-1-master Generate a video from an image using the Kling 2.1 Master model. # Kling 2.1 Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-pro/get-kling-v2-1-pro get /v1/ai/image-to-video/kling-v2-1-pro Get the list of the kling-v2-1-pro tasks # Kling 2.1 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-pro/get-kling-v2-1-pro-task get /v1/ai/image-to-video/kling-v2-1/{task-id} Get the status of the kling-v2-1 task # Kling Pro v2.1 – Image‑to‑Video API Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-pro/overview Advanced image‑to‑video generation from a single image with higher fidelity, richer motion control, and strong temporal consistency for premium results. Kling Pro v2.1 builds on the stability of the Std variant with enhanced fidelity and more expressive motion. It delivers higher detail retention, smoother transitions, and stronger control over subject and camera movement—ideal for premium marketing assets, cinematic previews, and design showcases. Prefer general‑purpose results? See the Kling Std v2.1 overview. ### Key capabilities * High‑fidelity motion and detail from a single image * Strong temporal consistency with minimal flicker * Improved texture, edges, and lighting continuity * Better retention of fine details on faces, products, and environments ### Use cases * Premium product spots: dynamic hero shots with refined motion * Cinematic promos: richer parallax and lighting transitions * High-end social posts and ads: crisp detail and smooth movement * Design/3D previews: bring still renders to life with controlled motion * Brand storytelling: subtle camera work with consistent identity ### Generate a video with Kling Pro v2.1 Create an image-to-video task using the Kling v2.1 Pro model.
Generate video from an image List image-to-video tasks Check task status
## Frequently Asked Questions Use high‑resolution images with clear subjects and balanced lighting. PNG/JPEG are recommended. Avoid heavy compression and noisy inputs to maximize detail retention. Duration depends on API parameters. Kling Pro v2.1 is optimized for short clips suitable for premium previews, ads, and social formats. Yes. Pro supports expressive motion with strong temporal stability. Complex motion may require careful prompting, masks, or parameter tuning. Std focuses on dependable, general‑purpose results. Pro increases fidelity and motion expressiveness. Master targets top‑tier quality and advanced control for cinematic applications. ## Best practices for image‑to‑video (Pro) * Use clean, well‑lit images at sufficient resolution to capture micro‑detail. * Keep the main subject clear of clutter; plan composition for parallax. * Start with subtle motion; iterate to increase complexity and camera movement. * Protect brand/face detail—avoid occlusions and extreme crops. * Validate at target aspect ratios/durations and compress for delivery platforms. # Kling 2.1 Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-pro/post-kling-v2-1-pro post /v1/ai/image-to-video/kling-v2-1-pro Generate a video from an image using the Kling 2.1 Pro model. # Kling 2.1 Standard - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-std/get-kling-v2-1-std get /v1/ai/image-to-video/kling-v2-1-std Get the list of the kling-v2-1-std tasks # Kling 2.1 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-std/get-kling-v2-1-std-task get /v1/ai/image-to-video/kling-v2-1/{task-id} Get the status of the kling-v2-1 task # Kling Std v2.1 – Image‑to‑Video API Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-std/overview Image‑to‑video generation from a single image with smooth motion and temporal consistency. Ideal for product demos, concept visualizations, and social content. Kling Std v2.1 is a robust image-to-video model that generates short, consistent videos from a single image. It focuses on temporal stability, smooth motion, and faithful visual detail, making it a dependable choice for product demos, concept visualizations, and social media content. Need higher fidelity and more expressive motion? Try Kling Pro v2.1. ### Key capabilities * Smooth, realistic motion generation from one input image * Strong temporal consistency with minimal flicker * Natural color and lighting preservation * Good detail retention on faces, products, and environments ### Use cases * Product showcases: animate angles or subtle movements from a hero image * Marketing and social: create short video loops for ads and posts * Concept visualization: bring static artwork or renders to life * UI/UX motion previews: generate simple motion studies from mockups * Educational content: illustrate processes with gentle camera moves ### Generate a video with Kling Std v2.1 Create an image-to-video task using the Kling v2.1 Standard model.
Generate video from an image List image-to-video tasks Check task status
## Frequently Asked Questions Use high-resolution images with clear subjects, clean backgrounds, and balanced lighting. JPEG/PNG are recommended. Avoid heavy compression and noisy inputs. Video duration depends on model defaults and parameters from the API. Kling Std v2.1 is geared toward short clips ideal for previews and social formats. Yes. The model can synthesize gentle camera moves and subject motion while maintaining temporal stability. Complex motion may require careful prompting and tuning. Std focuses on reliable, general‑purpose quality with good consistency. Pro and Master variants target higher fidelity or more advanced motion control. Choose Std for dependable everyday use cases. ## Best practices for image‑to‑video * Start with clean, well‑lit images at sufficient resolution. * Keep subjects centered and avoid busy backgrounds for clearer motion focus. * Use moderate motion expectations; evaluate results before increasing complexity. * Ensure branding, faces, and key details are unobstructed for better consistency. * Validate outputs at target aspect ratios and durations for your platform. # Kling 2.1 Standard - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.1-std/post-kling-v2-1-std post /v1/ai/image-to-video/kling-v2-1-std Generate a video from an image using the Kling 2.1 Std model. # Kling 2.5 720p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/get-kling-v2-5-720p get /v1/ai/image-to-video/kling-v2-5-720p Get the list of the kling-v2-5-720p tasks # Kling 2.5 Pro - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/get-kling-v2-5-pro get /v1/ai/image-to-video/kling-v2-5-pro Get the list of the kling-v2-5-pro tasks # Kling 2.5 Pro - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/get-{task-id}-by-id get /v1/ai/image-to-video/kling-v2-5-pro/{task-id} Get the status of the kling-v2-5-pro task # Kling 2.5 720p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/get-{task-id}-by-id-720p get /v1/ai/image-to-video/kling-v2-5-720p/{task-id} Get the status of the kling-v2-5-720p task # Kling 2.5 Turbo Pro – Image‑to‑Video API Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/overview Cinematic image‑to‑video from a single image with smoother motion, sharper detail, stable style/color, and faster rendering. 5s and 10s outputs. Kling 2.5 Turbo Pro generates cinematic videos from a single image with dramatically improved motion quality, sharper detail, and stable style and color. It offers 5s and 10s outputs, faster rendering than previous versions, and more reliable prompt adherence—ideal for marketing assets, trailers, and high‑end social content. ### Choosing a resolution The model is served at two resolutions, one endpoint each. Both take the same request body and differ only in output resolution and price per second. | Resolution | Endpoint | When to use | | ---------- | --------------------------------------- | ----------------------------------------------------------------------- | | 1080p | `/v1/ai/image-to-video/kling-v2-5-pro` | Final deliverables, cinematic promos, anything that will be shown large | | 720p | `/v1/ai/image-to-video/kling-v2-5-720p` | Drafts, iteration on prompts, social formats where 720p is enough | ### Key capabilities * Smooth, expressive motion and professional camera work * Sharper, cinematic imagery with improved clarity and texture * Stable style and color; better subject consistency across frames * Strong instruction adherence for look, tone, and motion * Better static preservation and camera perspective response * Improved subject understanding and finer emotional expression * Cost‑efficient credit usage compared to earlier versions * 5s and 10s duration options ### Use cases * Short films, teasers, and cinematic promos * YouTube/TikTok/Instagram videos (short‑ and long‑form) * Product and brand videos with consistent visuals * Educational explainers and motion studies * Experimental art and abstract motion
Generate video from an image Check task status List image-to-video tasks Generate video from an image at 720p Check 720p task status List 720p tasks
## Frequently Asked Questions Both resolutions support 5s and 10s outputs. Only the output resolution and the price per second. Same model, same request body, same durations — pick `kling-v2-5-720p` while you iterate and `kling-v2-5-pro` for the final render. Use a high‑resolution image with clear subjects and balanced lighting. Avoid heavy compression or clutter that can reduce perceived motion quality. Audio support is planned in an upcoming upgrade of the model. Yes. The model targets professional‑grade cinematic quality with stable style/color and reliable prompt adherence. # Kling 2.5 720p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/post-kling-v2-5-720p post /v1/ai/image-to-video/kling-v2-5-720p Generate a video from an image using the Kling 2.5 model at 720p resolution. For 1080p, use `/v1/ai/image-to-video/kling-v2-5-pro`. # Kling 2.5 Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2.5-pro/post-kling-v2-5-pro post /v1/ai/image-to-video/kling-v2-5-pro Generate a video from an image using the Kling 2.5 Pro model. # Kling 2.0 - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2/get-kling-v2 get /v1/ai/image-to-video/kling-v2 Get the list of the kling-v2 tasks # Kling 2.0 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2/get-kling-v2-task get /v1/ai/image-to-video/kling-v2/{task-id} Get the status of the kling-v2 task # Kling 2.0 - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/kling-v2/post-kling-v2 post /v1/ai/image-to-video/kling-v2 Create a video from an image using the Kling v2 model ## Important The service allows up to 3 concurrent requests per user. # Kling 3.0 Turbo I2V 1080p - Generate video Source: https://docs.magnific.com/api-reference/image-to-video/kling-v3-turbo/generate-1080p post /v1/ai/image-to-video/kling-v3-turbo-1080p Generate AI video from a first-frame image using Kling 3.0 Turbo at 1080p resolution, a fast image-to-video model from the Kling 3.0 family. **Features:** - **Image-to-video**: Animate a static image into a dynamic video - **Optional prompt**: Guide the motion and animation style with text - **Resolution**: Fixed 1080p output - **Duration**: 3-15 seconds **How it works:** 1. Submit an image (URL or Base64) to use as the first frame, with an optional prompt 2. Receive a task ID to track progress 3. Poll the status endpoint or use a `webhook_url` to get notified when the task completes 4. Download the generated video from the result URL (valid for 24 hours) # Kling 3.0 Turbo I2V 720p - Generate video Source: https://docs.magnific.com/api-reference/image-to-video/kling-v3-turbo/generate-720p post /v1/ai/image-to-video/kling-v3-turbo-720p Generate AI video from a first-frame image using Kling 3.0 Turbo at 720p resolution, a fast image-to-video model from the Kling 3.0 family. **Features:** - **Image-to-video**: Animate a static image into a dynamic video - **Optional prompt**: Guide the motion and animation style with text - **Resolution**: Fixed 720p output - **Duration**: 3-15 seconds **How it works:** 1. Submit an image (URL or Base64) to use as the first frame, with an optional prompt 2. Receive a task ID to track progress 3. Poll the status endpoint or use a `webhook_url` to get notified when the task completes 4. Download the generated video from the result URL (valid for 24 hours) # Kling 3.0 Turbo Image To Video API Source: https://docs.magnific.com/api-reference/image-to-video/kling-v3-turbo/overview Animate a single image into video fast with Kling 3.0 Turbo. Dedicated 720p and 1080p generate endpoints, 3-15 second durations, and optional prompt-guided motion control. Kling 3.0 Turbo is the fast generation model from the Kling 3.0 family, animating a first-frame image into smooth, high-fidelity video with short processing times. Kling 3.0 Turbo image-to-video generates a video from a single first-frame image, optionally guided by a text prompt. It outputs high-fidelity video with durations from 3 to 15 seconds, and is optimized for fast generation, making it well suited for animating product shots, artwork, and social media content at scale. Resolution is selected at generation time: use the dedicated **720p** generate endpoint for standard high definition or the **1080p** generate endpoint for full high definition. Task status and listing use a single generic endpoint, regardless of the resolution. For the full model overview, including the text-to-video mode, see the [Kling 3.0 Turbo overview](/api-reference/text-to-video/kling-v3-turbo/overview). ### Key capabilities * **Image animation**: Turn a single first-frame image into motion * **Resolution by generate endpoint**: Dedicated `720p` and `1080p` generate endpoints * **Unified task tracking**: One generic list endpoint and one generic status endpoint for tasks of any resolution * **Fast generation**: Turbo-optimized pipeline for shorter processing times * **Flexible durations**: Any whole number from 3 to 15 seconds, default `5` * **Optional prompt**: Guide motion and style with up to 2500 characters, or omit it to infer motion from the image * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated MP4 video is returned via a URL valid for 24 hours ### Use cases * **Product videos**: Animate static product photography for ads and listings * **Social media content**: Bring images to life for Reels, Shorts, and Stories * **Creative exploration**: Add motion to illustrations, concept art, and designs * **Rapid prototyping**: Quickly preview how a still composition moves ## API Operations
Generate a 720p video from a first-frame image Generate a 1080p video from a first-frame image List all Kling 3.0 Turbo I2V tasks (any resolution) Get I2V task status and result by ID
### Endpoint structure | Operation | Endpoint | | ------------------ | ---------------------------------------------------- | | **Generate 720p** | `POST /v1/ai/image-to-video/kling-v3-turbo-720p` | | **Generate 1080p** | `POST /v1/ai/image-to-video/kling-v3-turbo-1080p` | | **List tasks** | `GET /v1/ai/image-to-video/kling-v3-turbo` | | **Get task** | `GET /v1/ai/image-to-video/kling-v3-turbo/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-720p` or `-1080p`) and is not a body parameter. | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | --------------------------------------------------------------------------------------------------------------------------- | | `image` | `string` | Yes | - | First-frame image as a public URL or Base64 string. JPG, JPEG, or PNG; min 300x300px; max 50MB; aspect ratio 1:2.5 to 2.5:1 | | `prompt` | `string` | No | - | Motion and style guidance. Up to 2500 characters. If omitted, motion is inferred from the image | | `duration` | `string` | No | `5` | Video length in seconds, any whole number from `3` to `15` | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Resolution is selected by the generate endpoint, not by a body parameter. Call `POST /v1/ai/image-to-video/kling-v3-turbo-720p` for standard high definition or `POST /v1/ai/image-to-video/kling-v3-turbo-1080p` for full high definition. Listing and status use the single generic endpoint `GET /v1/ai/image-to-video/kling-v3-turbo` (and `/{task-id}`). Image-to-video accepts JPG, JPEG, and PNG, provided as a publicly accessible URL or a Base64-encoded string (without the data URI prefix). Requirements: minimum 300x300 pixels, maximum 50MB, and an aspect ratio between 1:2.5 and 2.5:1. Only the first frame is supported. No. The prompt is optional in image-to-video. If you omit it, motion is inferred from the source image. Provide a prompt (up to 2500 characters) to guide the movement, camera behavior, and visual style. No. Image-to-video does not accept an aspect ratio parameter; the output ratio is inherited from the source image. Use text-to-video if you need to choose `16:9`, `9:16`, or `1:1`. Any whole-number duration from 3 to 15 seconds (default `5`). Shorter durations generate faster. Submit the request to receive a task ID, then poll the generic GET task endpoint or provide a `webhook_url` to be notified on completion. The result MP4 is delivered via a URL that remains valid for 24 hours. Rate limits depend on your subscription tier (see [Rate Limits](/ratelimits)). Pricing varies by resolution and duration (see [Pricing](/pricing)). ## Related APIs * **[Kling 3.0 Turbo (Text to Video)](/api-reference/text-to-video/kling-v3-turbo/overview)**: Generate video from a text prompt with aspect ratio control * **[Kling 3](/api-reference/video/kling-v3/overview)**: Standard Kling 3.0 model for higher-fidelity output * **[Runway Gen-4 Turbo](/api-reference/image-to-video/runway-gen4-turbo)**: Fast i2v alternative with cinematic quality # Kling 3.0 Turbo I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/kling-v3-turbo/task-by-id get /v1/ai/image-to-video/kling-v3-turbo/{task-id} Retrieve the status and result of a specific Kling 3.0 Turbo I2V task by its task ID, regardless of the resolution it was created at. # Kling 3.0 Turbo I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/kling-v3-turbo/tasks get /v1/ai/image-to-video/kling-v3-turbo Retrieve the list of all Kling 3.0 Turbo I2V tasks for the authenticated user, across all resolutions. # LTX Video 2.0 Fast - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-fast post /v1/ai/image-to-video/ltx-2-fast Generate a video from an image using the LTX Video 2.0 Fast model. **Features:** - Fast video generation with resolutions up to 4K (2160p) - Extended duration options: 6-20 seconds in 2-second increments - Uses the provided image as the first frame - Optional synchronized audio generation - **Note:** Durations longer than 10 seconds require 25 FPS and 1080p resolution # LTX Video 2.0 Fast I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-fast-tasks get /v1/ai/image-to-video/ltx-2-fast # LTX Video 2.0 Fast I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-fast/task-by-id get /v1/ai/image-to-video/ltx-2-fast/{task-id} # LTX Video 2.0 Pro - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-pro post /v1/ai/image-to-video/ltx-2-pro Generate a video from an image using the LTX Video 2.0 Pro model. **Features:** - High-quality video generation with resolutions up to 4K (2160p) - Duration options: 6, 8, or 10 seconds - Uses the provided image as the first frame - Optional synchronized audio generation - Reproducible results with seed parameter # LTX Video 2.0 Pro I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-pro-tasks get /v1/ai/image-to-video/ltx-2-pro # LTX Video 2.0 Pro I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/ltx-2-pro/task-by-id get /v1/ai/image-to-video/ltx-2-pro/{task-id} # MiniMax Hailuo 02 1080p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-1080p/get-minimax-hailuo-02-1080p get /v1/ai/image-to-video/minimax-hailuo-02-1080p Get the list of MiniMax Hailuo-02 1080p tasks # MiniMax Hailuo 02 1080p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-1080p/get-minimax-hailuo-02-1080p-task get /v1/ai/image-to-video/minimax-hailuo-02-1080p/{task-id} Get the status of the MiniMax Hailuo-02 1080p task # MiniMax Hailuo 02 1080p - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-1080p/post-minimax-hailuo-02-1080p post /v1/ai/image-to-video/minimax-hailuo-02-1080p Generate a video from text or image using the MiniMax Hailuo-02 1080p model. # MiniMax Hailuo 02 768p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-768p/get-minimax-hailuo-02-768p get /v1/ai/image-to-video/minimax-hailuo-02-768p Get the list of MiniMax Hailuo-02 768p tasks # MiniMax Hailuo 02 768p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-768p/get-minimax-hailuo-02-768p-task get /v1/ai/image-to-video/minimax-hailuo-02-768p/{task-id} Get the status of the MiniMax Hailuo-02 768p task # MiniMax Hailuo 02 768p - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-02-768p/post-minimax-hailuo-02-768p post /v1/ai/image-to-video/minimax-hailuo-02-768p Generate a video from text or image using the MiniMax Hailuo-02 768p model. # MiniMax Hailuo 2.3 1080p Fast - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p-fast/get-minimax-hailuo-2-3-1080p-fast get /v1/ai/image-to-video/minimax-hailuo-2-3-1080p-fast Get the list of MiniMax Hailuo 2.3 1080p Fast tasks # MiniMax Hailuo 2.3 1080p Fast - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p-fast/get-minimax-hailuo-2-3-1080p-fast-task get /v1/ai/image-to-video/minimax-hailuo-2-3-1080p-fast/{task-id} Get the status of the MiniMax Hailuo 2.3 1080p Fast task # MiniMax Hailuo 2.3 1080p Fast - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p-fast/post-minimax-hailuo-2-3-1080p-fast post /v1/ai/image-to-video/minimax-hailuo-2-3-1080p-fast Generate a video from text or image using the MiniMax Hailuo 2.3 1080p model with fast prompt optimization. # MiniMax Hailuo 2.3 1080p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p/get-minimax-hailuo-2-3-1080p get /v1/ai/image-to-video/minimax-hailuo-2-3-1080p Get the list of MiniMax Hailuo 2.3 1080p tasks # MiniMax Hailuo 2.3 1080p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p/get-minimax-hailuo-2-3-1080p-task get /v1/ai/image-to-video/minimax-hailuo-2-3-1080p/{task-id} Get the status of the MiniMax Hailuo 2.3 1080p task # MiniMax Hailuo 2.3 1080p - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-1080p/post-minimax-hailuo-2-3-1080p post /v1/ai/image-to-video/minimax-hailuo-2-3-1080p Generate a video from text or image using the MiniMax Hailuo 2.3 1080p model. # MiniMax Hailuo 2.3 768p Fast - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p-fast/get-minimax-hailuo-2-3-768p-fast get /v1/ai/image-to-video/minimax-hailuo-2-3-768p-fast Get the list of MiniMax Hailuo 2.3 768p Fast tasks # MiniMax Hailuo 2.3 768p Fast - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p-fast/get-minimax-hailuo-2-3-768p-fast-task get /v1/ai/image-to-video/minimax-hailuo-2-3-768p-fast/{task-id} Get the status of the MiniMax Hailuo 2.3 768p Fast task # MiniMax Hailuo 2.3 768p Fast - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p-fast/post-minimax-hailuo-2-3-768p-fast post /v1/ai/image-to-video/minimax-hailuo-2-3-768p-fast Generate a video from text or image using the MiniMax Hailuo 2.3 768p model with fast prompt optimization. # MiniMax Hailuo 2.3 768p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p/get-minimax-hailuo-2-3-768p get /v1/ai/image-to-video/minimax-hailuo-2-3-768p Get the list of MiniMax Hailuo 2.3 768p tasks # MiniMax Hailuo 2.3 768p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p/get-minimax-hailuo-2-3-768p-task get /v1/ai/image-to-video/minimax-hailuo-2-3-768p/{task-id} Get the status of the MiniMax Hailuo 2.3 768p task # MiniMax Hailuo 2.3 768p - Create video from text or image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-hailuo-2-3-768p/post-minimax-hailuo-2-3-768p post /v1/ai/image-to-video/minimax-hailuo-2-3-768p Generate a video from text or image using the MiniMax Hailuo 2.3 768p model. # MiniMax Video 01 Live - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/minimax-live post /v1/ai/image-to-video/minimax-live Generate a video from an image using MiniMax Video-01-Live model (Live Illustrations). **Features:** - Supports camera movements in square brackets: [Truck left], [Pan right], [Push in], [Pull out], [Zoom in], [Tracking shot], [Static shot] - Optional prompt optimization for better results - Works best with illustrations and artwork # MiniMax Video 01 Live - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/minimax-live-tasks get /v1/ai/image-to-video/minimax-live # MiniMax Video 01 Live - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/minimax-live/task-by-id get /v1/ai/image-to-video/minimax-live/{task-id} # Video Generation API Source: https://docs.magnific.com/api-reference/image-to-video/overview Generate AI videos from text prompts or images. Multiple models available including Kling, WAN, MiniMax, Runway, PixVerse, and LTX with various quality tiers and resolutions. Generate videos from text prompts or images using state-of-the-art AI models. Choose the model that best fits your quality, speed, and budget requirements. The Video Generation API provides access to multiple AI video models through a unified async workflow: submit a request, receive a task ID, then poll for results or get notified via webhook. ### Available models | Model | Modes | Resolutions | Durations | Highlights | | ------------------------------------------------------------------ | -------- | ----------- | --------- | -------------------------------- | | **[Kling 3](/api-reference/video/kling-v3/overview)** | T2V, I2V | up to 1080p | 3-15s | Multi-shot, element consistency | | **[Kling 3 Omni](/api-reference/video/kling-v3-omni/overview)** | T2V, I2V | up to 1080p | 3-15s | Video reference for motion/style | | **[Kling O1](/api-reference/image-to-video/kling-o1/overview)** | T2V, I2V | up to 1080p | 5-10s | High-performance generation | | **[WAN 2.6](/api-reference/wan-v2-6-i2v/overview)** | T2V, I2V | 720p, 1080p | 5-15s | High resolution, long duration | | **[WAN 2.5](/api-reference/wan-2-5-t2v/overview)** | T2V, I2V | 480p-1080p | 5-15s | Multiple resolution tiers | | **[Runway Gen 4.5](/api-reference/video/runway-gen-4-5/overview)** | T2V, I2V | up to 1080p | 5-10s | Cinematic quality | | **[MiniMax Live](/api-reference/image-to-video/minimax-live)** | I2V | 720p, 1080p | 5s | Fast generation | | **[LTX 2.0](/api-reference/image-to-video/ltx-2-fast)** | T2V, I2V | 768p | 3-9s | Fast and Pro tiers | ### How it works All video models follow the same async pattern: 1. **POST** a generation request with your prompt and/or image 2. Receive a `task_id` immediately 3. **Poll** with `GET /v1/ai/{model}/{task_id}` or receive a **webhook** notification 4. Download the resulting video from the `video_url` in the response ### Common parameters | Parameter | Description | | -------------------------- | ------------------------------------------ | | `prompt` | Text description of the video to generate | | `image_url` / `image_list` | Source image(s) for image-to-video modes | | `aspect_ratio` | Video aspect ratio (`16:9`, `9:16`, `1:1`) | | `duration` | Video length in seconds | | `webhook_url` | URL for async completion notification | Parameter names and accepted values vary by model. Check each model's endpoint reference for exact specifications. ### Choosing a model * **Highest quality**: Kling 3 Pro, Runway Gen 4.5 * **Best value**: Kling 3 Standard, LTX 2.0 Fast * **Longest videos**: WAN 2.6 (up to 15s at 1080p) * **Fastest generation**: LTX 2.0 Fast, MiniMax Live * **Multi-scene narratives**: Kling 3 (multi-shot mode) * **Motion/style reference**: Kling 3 Omni (video reference input) # PixVerse V5 Transition - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-transition/get-pixverse-v5-transition get /v1/ai/image-to-video/pixverse-v5-transition List PixVerse-V5 Transition tasks. # PixVerse V5 Transition - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-transition/get-{task-id}-by-id get /v1/ai/image-to-video/pixverse-v5-transition/{task-id} # PixVerse V5 - Video transition Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-transition/post-pixverse-v5-transition post /v1/ai/image-to-video/pixverse-v5-transition Generate a video transition between two images using the PixVerse-V5 model. # PixVerse V5.5 - Video transition Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v5-5-transition/create post /v1/ai/image-to-video/pixverse-v5-5-transition Generate a video transition using the PixVerse-V5.5 model. Provide a `first_image_url` and an `end_image_url` to animate a seamless transition between the two frames. PixVerse-V5.5 transition supports native audio (`generate_audio_switch`), camera movement control and prompt reasoning enhancement (`thinking_type`). # PixVerse V5.5 Transition API Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v5-5-transition/overview Generate first-to-last frame video transitions with PixVerse V5.5. Interpolate between two images with camera movement, native audio, and 5-10s durations. ## Overview PixVerse V5.5 Transition is an image-to-video API that interpolates a video between two keyframes. Provide a `first_image_url` and an `end_image_url`, and PixVerse V5.5 generates a smooth transition between the two frames guided by your `prompt`. Both images are required. It supports camera movement presets, native synchronized audio, prompt reasoning, and 5, 8, or 10 second durations. ### Key capabilities * **First-to-last frame transition**: interpolate between the required `first_image_url` and `end_image_url` in a single request * **Camera movement presets**: 20 options including `zoom_in`, `zoom_out`, `pan_left`, `pan_right`, `crane_up`, `hitchcock`, `whip_pan`, `camera_rotation` * **Native synchronized audio**: enable `generate_audio_switch` to produce music, sound effects, or dialogue together with the video * **Prompt reasoning (`thinking_type`)**: choose `enabled` (default), `disabled`, or `auto` to control prompt rewriting * **Duration (5, 8, or 10 seconds)**: `5` (default), `8`, or `10`; 8s costs double, 10s is available up to 720p, and 1080p is limited to 5 or 8 seconds * **Resolutions**: `360p`, `540p`, `720p`, `1080p` * **Aspect ratios**: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` * **Async processing**: poll the task endpoint or receive a webhook notification on completion ### Use cases * **Story transitions**: scene-to-scene cuts and morphs that bridge two shots * **Product reveals**: transition between a closed and open, or before and after, product state * **Marketing and ads**: dynamic transitions with camera movement and synchronized audio * **Social content**: vertical `social_story_9_16` transitions for TikTok, Instagram Reels, and YouTube Shorts * **Creative experimentation**: stylized morphs (`anime`, `cyberpunk`, `clay`) between keyframes
Generate a first/last frame transition with PixVerse V5.5 Get task status and result by ID List all PixVerse V5.5 Transition tasks
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------- | -------------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Prompt describing the video transition to generate | | `first_image_url` | `string` (uri) | Yes | - | URL of the image to use as the first frame | | `end_image_url` | `string` (uri) | Yes | - | URL of the image to use as the last frame | | `resolution` | `string` | No | - | Output resolution: `360p`, `540p`, `720p`, `1080p` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` | | `duration` | `integer` | No | `5` | Video length in seconds: `5`, `8`, or `10`. 8s costs double; 10s up to 720p; 1080p limited to 5 or 8s | | `negative_prompt` | `string` | No | `""` | Negative prompt for the generation | | `style` | `string` | No | - | Visual style: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` | | `seed` | `integer` | No | - | Reproducibility seed; same seed + same prompt yields the same video | | `camera_movement` | `string` | No | - | Camera preset (e.g., `zoom_in`, `pan_left`, `crane_up`, `hitchcock`, `whip_pan`, `camera_rotation`) | | `generate_audio_switch` | `boolean` | No | `false` | When `true`, generates synchronized audio (music, SFX, or dialogue) with the video | | `thinking_type` | `string` | No | `enabled` | Prompt reasoning mode: `enabled`, `disabled`, or `auto` | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ## Frequently Asked Questions Provide a `first_image_url` and an `end_image_url`. PixVerse V5.5 interpolates a video between the two frames using your `prompt` as guidance, producing the transition in a single request. Yes. PixVerse V5.5 Transition requires both `first_image_url` (first frame) and `end_image_url` (last frame). Requests without an end frame are rejected. PixVerse V5.5 Transition supports `5` (default), `8`, or `10` second durations and resolutions of `360p`, `540p`, `720p`, and `1080p`. 8-second videos cost double, 10-second videos are available up to 720p, and 1080p is limited to 5 or 8 seconds. 20 presets are supported, including `zoom_in`, `zoom_out`, `pan_left`, `pan_right`, `crane_up`, `hitchcock`, `whip_pan`, and `camera_rotation`. Leave `camera_movement` unset for a static camera. Set `generate_audio_switch` to `true` and PixVerse V5.5 produces synchronized audio (background music, sound effects, or dialogue) together with the video in a single request. Rate limits and pricing depend on your subscription tier. See [Rate Limits](/ratelimits) and the [Pricing page](/pricing) for current values. ## Best practices * **Frame consistency**: choose two frames that share scene context (subject, framing, lighting) for the smoothest interpolation * **Input quality**: provide sharp, well-lit images with a clear subject for the strongest motion and style stability * **Camera movement**: combine a `camera_movement` preset with the transition for more dynamic results, or leave it unset for a clean morph * **Duration selection**: use `5` seconds for most transitions to reduce cost; request 10s only at 720p or lower * **Production integration**: use `webhook_url` instead of polling for scalable workflows * **Error handling**: implement retry with exponential backoff for 503 responses ## Related APIs * **PixVerse V6 Transition**: newer transition endpoint with flexible 1-15 second durations and multi-clip output. See the [PixVerse V6 Transition overview](/api-reference/image-to-video/pixverse-v6-transition/overview). * **PixVerse V5.5 (image-to-video)**: animate a single image without a transition. See the [PixVerse V5.5 overview](/api-reference/image-to-video/pixverse-v5-5/overview). * **PixVerse V5.5 Text to Video**: generate a video from a text prompt with no input image. See the [PixVerse V5.5 Text to Video overview](/api-reference/text-to-video/pixverse-v5-5/overview). # PixVerse V5.5 Transition - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v5-5-transition/pixverse-v5-5-transition-tasks get /v1/ai/image-to-video/pixverse-v5-5-transition List PixVerse-V5.5 Transition tasks. # PixVerse V5.5 Transition - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v5-5-transition/task-by-id get /v1/ai/image-to-video/pixverse-v5-5-transition/{task-id} # PixVerse V6 - Video transition Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v6-transition/create post /v1/ai/image-to-video/pixverse-v6-transition Generate a video transition using the PixVerse-V6 model. Provide a `first_image_url` and an `end_image_url` to animate a seamless transition between the two frames. PixVerse-V6 transition supports flexible duration (`1`-`15` seconds), native audio (`generate_audio_switch`), multi-clip output (`generate_multi_clip_switch`), camera movement control and prompt reasoning enhancement (`thinking_type`). # PixVerse V6 Transition API Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v6-transition/overview Generate first-to-last frame video transitions with PixVerse V6. Interpolate between two images with flexible 1-15s durations, camera movement, and native audio. ## Overview PixVerse V6 Transition is an image-to-video API that interpolates a video between two keyframes. Provide a `first_image_url` and an `end_image_url`, and PixVerse V6 generates a smooth transition between the two frames guided by your `prompt`. Both images are required. Compared to PixVerse V5.5 Transition, version 6 unlocks flexible 1-15 second durations and adds multi-clip output, while keeping camera movement presets, native audio, and prompt reasoning. ### Key capabilities * **First-to-last frame transition**: interpolate between the required `first_image_url` and `end_image_url` in a single request * **Flexible duration (1-15 seconds)**: any integer length from `1` to `15`, with default `5` (previously fixed to `5`, `8`, or `10`) * **Camera movement presets**: 20 options including `zoom_in`, `zoom_out`, `pan_left`, `pan_right`, `crane_up`, `hitchcock`, `whip_pan`, `camera_rotation` * **Multi-clip with dynamic cameras**: enable `generate_multi_clip_switch` for multi-clip output with camera changes in a single generation * **Native synchronized audio**: enable `generate_audio_switch` to produce music, sound effects, or dialogue together with the video * **Prompt reasoning (`thinking_type`)**: choose `enabled` (default), `disabled`, or `auto` to control prompt rewriting * **Resolutions**: `360p`, `540p`, `720p`, `1080p` * **Aspect ratios**: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` * **Async processing**: poll the task endpoint or receive a webhook notification on completion ### Use cases * **Story transitions**: scene-to-scene cuts and morphs that bridge two shots in any duration * **Product reveals**: transition between a closed and open, or before and after, product state * **Marketing and ads**: dynamic transitions with camera movement, multi-clip output, and synchronized audio * **Social content**: vertical `social_story_9_16` transitions for TikTok, Instagram Reels, and YouTube Shorts * **Short films and teasers**: longer 10-15 second transition sequences with camera changes
Generate a first/last frame transition with PixVerse V6 Get task status and result by ID List all PixVerse V6 Transition tasks
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | -------------- | -------- | ----------------- | --------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Prompt describing the video transition to generate | | `first_image_url` | `string` (uri) | Yes | - | URL of the image to use as the first frame | | `end_image_url` | `string` (uri) | Yes | - | URL of the image to use as the last frame | | `resolution` | `string` | No | - | Output resolution: `360p`, `540p`, `720p`, `1080p` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` | | `duration` | `integer` | No | `5` | Video length in seconds. Any integer from `1` to `15` | | `negative_prompt` | `string` | No | `""` | Negative prompt for the generation | | `style` | `string` | No | - | Visual style: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` | | `seed` | `integer` | No | - | Reproducibility seed; same seed + same prompt yields the same video | | `camera_movement` | `string` | No | - | Camera preset (e.g., `zoom_in`, `pan_left`, `crane_up`, `hitchcock`, `whip_pan`, `camera_rotation`) | | `generate_audio_switch` | `boolean` | No | `false` | When `true`, generates synchronized audio (music, SFX, or dialogue) with the video | | `generate_multi_clip_switch` | `boolean` | No | `false` | When `true`, produces multi-clip output with dynamic camera changes in a single generation | | `thinking_type` | `string` | No | `enabled` | Prompt reasoning mode: `enabled`, `disabled`, or `auto` | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ## Frequently Asked Questions PixVerse V6 Transition adds flexible duration (any integer from `1` to `15` seconds instead of fixed `5`, `8`, or `10`) and multi-clip output (`generate_multi_clip_switch`). Camera movement, native audio, and prompt reasoning work the same as in V5.5 Transition. Provide a `first_image_url` and an `end_image_url`. PixVerse V6 interpolates a video between the two frames using your `prompt` as guidance, producing the transition in a single request. Both images are required. PixVerse V6 Transition accepts any integer duration from `1` to `15` seconds, with a default of `5`. You are no longer limited to fixed steps, so you can match the exact length of your transition. 20 presets are supported, including `zoom_in`, `zoom_out`, `pan_left`, `pan_right`, `crane_up`, `hitchcock`, `whip_pan`, and `camera_rotation`. Leave `camera_movement` unset for a static camera. When set to `true`, PixVerse V6 produces multi-clip output with dynamic camera changes inside a single generation, simulating cuts and camera moves without stitching multiple requests. Rate limits and pricing depend on your subscription tier. See [Rate Limits](/ratelimits) and the [Pricing page](/pricing) for current values. ## Best practices * **Frame consistency**: choose two frames that share scene context (subject, framing, lighting) for the smoothest interpolation * **Input quality**: provide sharp, well-lit images with a clear subject for the strongest motion and style stability * **Duration selection**: use the shortest duration that conveys the transition to reduce cost; reserve 10-15s for narrative sequences * **Camera and multi-clip**: combine `camera_movement` and `generate_multi_clip_switch` for dynamic, multi-shot transitions * **Production integration**: use `webhook_url` instead of polling for scalable workflows * **Error handling**: implement retry with exponential backoff for 503 responses ## Related APIs * **PixVerse V5.5 Transition**: prior transition endpoint with fixed `5`/`8`/`10` second durations and no multi-clip output. See the [PixVerse V5.5 Transition overview](/api-reference/image-to-video/pixverse-v5-5-transition/overview). * **PixVerse V6 (image-to-video)**: animate a single image or run a transition natively via `last_frame_image`. See the [PixVerse V6 overview](/api-reference/image-to-video/pixverse-v6/overview). * **PixVerse V6 Text to Video**: generate a video from a text prompt with no input image. See the [PixVerse V6 Text to Video overview](/api-reference/text-to-video/pixverse-v6/overview). # PixVerse V6 Transition - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v6-transition/pixverse-v6-transition-tasks get /v1/ai/image-to-video/pixverse-v6-transition List PixVerse-V6 Transition tasks. # PixVerse V6 Transition - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/pixverse-v6-transition/task-by-id get /v1/ai/image-to-video/pixverse-v6-transition/{task-id} # PixVerse V5 - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/pixverse/get-pixverse-v5 get /v1/ai/image-to-video/pixverse-v5 List PixVerse-V5 tasks filtered by resolution provided in the request body. # PixVerse V5 - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/pixverse/get-{task-id}-by-id get /v1/ai/image-to-video/pixverse-v5/{task-id} # PixVerse V5 – Image‑to‑Video API Source: https://docs.magnific.com/api-reference/image-to-video/pixverse/overview Generate cinematic videos from a single image with smooth, expressive motion, stable style and color, and strong instruction adherence. Faster rendering and multiple output options. ## Overview PixVerse V5 generates videos from a single image with cinematic quality and smooth, expressive motion. It improves rendering speed, maintains style and color across frames, and follows instructions reliably—ideal for creative exploration and production pipelines. ### Key capabilities * Smooth, natural motion from one input image * Cinematic imagery with sharper details and clarity * Stable style and color across sequences (subject consistency) * Strong instruction adherence for look, tone, and motion * Faster rendering versus prior versions, with multiple output options ### Use cases * Short films, teasers, and cinematic promos * YouTube/TikTok/Instagram content (short- and long‑form) * Product and brand videos with consistent subjects * Educational explainers and motion studies * Experimental art and abstract motion
Generate video from an image Check task status List image-to-video tasks
## Frequently Asked Questions Use a high‑resolution image with clear subjects and balanced lighting. Avoid heavy compression and cluttered backgrounds to maximize motion quality. Common outputs include 360p, 540p, 720p, and 1080p. For extreme resolutions, pair results with an upscaler to reach 4K. Yes. V5 improves style/color stability and subject consistency for more coherent sequences and narratives. # PixVerse V5 - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/pixverse/post-pixverse-v5 post /v1/ai/image-to-video/pixverse-v5 Generate a video using the PixVerse-V5 model. Resolution is specified in the request body. # Create video from image - Veo 3.1 Source: https://docs.magnific.com/api-reference/image-to-video/post-veo-3-1 post /v1/ai/image-to-video/veo-3-1 Generate a video from an image using Google Veo 3.1 model. Supports multiple resolutions (720p, 1080p, 4K) and optional audio generation. # Create video from image - Veo 3.1 Fast Source: https://docs.magnific.com/api-reference/image-to-video/post-veo-3-1-fast post /v1/ai/image-to-video/veo-3-1-fast Generate a video from an image using Google Veo 3.1 Fast model. Faster generation at a lower cost. # Create video from image - Veo 3.1 Lite Source: https://docs.magnific.com/api-reference/image-to-video/post-veo-3-1-lite post /v1/ai/image-to-video/veo-3-1-lite Generate a video from an image using Google Veo 3.1 Lite. A cost-efficient variant of Veo 3.1 supporting 720p and 1080p resolutions with optional native audio generation. # RunWay Gen4 Turbo - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/runway-gen4-turbo post /v1/ai/image-to-video/runway-gen4-turbo Generate a video from an image using RunWay Gen4 Turbo model. # RunWay Gen4 Turbo - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/runway-gen4-turbo-tasks get /v1/ai/image-to-video/runway-gen4-turbo # RunWay Gen4 Turbo - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/runway-gen4-turbo/task-by-id get /v1/ai/image-to-video/runway-gen4-turbo/{task-id} # Veo 3.1 Lite – Image-to-Video API | Magnific API Source: https://docs.magnific.com/api-reference/image-to-video/veo-3-1-lite/overview Animate images into cinematic videos with Google Veo 3.1 Lite. A cost-efficient variant supporting 720p and 1080p resolutions with optional native audio. Cost-efficient image-to-video generation powered by Google's Veo 3.1 Lite model with 720p and 1080p output and optional audio synthesis. Veo 3.1 Lite is the cost-efficient variant of Google's Veo 3.1 model. It animates a source image into a high-quality cinematic video at 720p or 1080p resolution, guided by a text prompt, with durations of 4-8 seconds and optional native audio generation. Choose Veo 3.1 Lite when you want the Veo 3.1 look at a lower cost for prototyping, iteration, and high-volume workflows. ### Key capabilities * **Image-driven generation**: Animate a source image into video guided by a text prompt * **Cost-efficient generation**: The lightweight Veo 3.1 variant for prototyping and high-volume workflows * **Multi-resolution output**: Generate videos in 720p or 1080p resolution * **Audio generation**: Optional synchronized audio for complete multimedia content * **Flexible durations**: 4, 6, or 8 second video outputs * **Aspect ratio control**: 16:9 (landscape) or 9:16 (portrait) formats * **Negative prompts**: Specify elements to avoid in generation * **Long prompts**: Up to 20,000 characters for detailed descriptions * **Reproducible results**: Seed parameter for consistent outputs ### Use cases * **Product animation**: Bring product photos to life for marketing and ads * **Social media**: Turn images into engaging short-form content for TikTok, Reels, and Shorts * **Storytelling**: Animate concept art, illustrations, and stills * **Advertising**: Produce video ads from a key visual * **Prototyping**: Quickly test motion from an image before committing to a higher-fidelity model ### Generate with Veo 3.1 Lite Submit a source image and a text prompt describing the motion you want. The service returns a task ID for async polling or webhook notification.
Create a new image-to-video task List all Veo 3.1 Lite tasks Get task status by ID
### Parameters | Parameter | Type | Required | Description | | ----------------- | --------- | -------- | ------------------------------------------------------------------------- | | `image` | `string` | Yes | Input image as a Base64-encoded string or a publicly accessible HTTPS URL | | `prompt` | `string` | Yes | Text describing the video to generate (max 20,000 chars) | | `negative_prompt` | `string` | No | Text describing what to avoid in the video (max 2,000 chars) | | `duration` | `integer` | No | Video duration: `4`, `6`, or `8` seconds (default: `8`) | | `resolution` | `string` | No | Output resolution: `"720p"` or `"1080p"` (default: `"720p"`) | | `aspect_ratio` | `string` | No | Video format: `"16:9"` or `"9:16"` (default: `"16:9"`) | | `generate_audio` | `boolean` | No | Generate synchronized audio (default: `true`) | | `seed` | `integer` | No | Random seed for reproducibility (0-4294967295) | | `webhook_url` | `string` | No | URL for task completion notification | ## Frequently Asked Questions Veo 3.1 Lite is the cost-efficient variant of Veo 3.1. It supports 720p and 1080p resolutions (Veo 3.1 also offers 4K) and is optimized for lower-cost, high-volume generation. Use Veo 3.1 Lite for prototyping and iteration, and Veo 3.1 Standard when you need maximum fidelity or 4K output. Provide the `image` as a Base64-encoded string or a publicly accessible HTTPS URL. The image is used as the starting frame that Veo 3.1 Lite animates according to your prompt. Veo 3.1 Lite supports **720p** (HD) and **1080p** (Full HD). For 4K output, use the standard Veo 3.1 model. When `generate_audio` is set to `true` (the default), Veo 3.1 Lite synthesizes synchronized audio that matches the visual content. This includes ambient sounds, music, and effects appropriate to the scene. Generation time varies based on resolution, duration, and server load. Use webhooks for production workflows to avoid polling. Yes, set `aspect_ratio` to `"9:16"` to generate vertical videos optimized for TikTok, Instagram Reels, and YouTube Shorts. ## Best practices * **Input quality**: Provide a sharp, well-lit source image with a clear subject for the strongest motion * **Prompt detail**: Describe the motion, camera movement, and atmosphere you want applied to the image * **Negative prompts**: Use `negative_prompt` to exclude unwanted elements like "blurry, low quality, distorted" * **Resolution selection**: Choose 720p for fast iteration and 1080p for social media and final delivery * **Audio control**: Disable `generate_audio` if you plan to add custom audio in post-production * **Webhook integration**: Use webhooks instead of polling for scalable production applications ## Related APIs * **[Veo 3.1 Image-to-Video](/api-reference/image-to-video/veo-3-1/overview)**: The standard Veo 3.1 model with 4K support and Fast mode * **[Veo 3.1 Lite Text-to-Video](/api-reference/text-to-video/veo-3-1-lite/overview)**: Generate videos from a text prompt with Veo 3.1 Lite * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Alternative image-to-video with motion control * **[WAN 2.6](/api-reference/text-to-video/wan-v2-6-720p)**: Text-to-video at 720p and 1080p resolutions # Veo 3.1 – Image-to-Video API | Magnific API Source: https://docs.magnific.com/api-reference/image-to-video/veo-3-1/overview Transform images into cinematic videos with Google Veo 3.1. Supports 720p, 1080p, and 4K resolutions with optional audio. Standard and Fast modes available. State-of-the-art image-to-video generation powered by Google's Veo 3.1 model with multi-resolution output and optional audio synthesis. Veo 3.1 Image-to-Video transforms static images into dynamic, cinematic videos guided by text prompts. It supports resolutions up to 4K, durations of 4-8 seconds, and optional audio generation. Choose between **Standard** mode for maximum quality or **Fast** mode for quicker generation at lower cost. ### Key capabilities * **Image animation**: Bring any image to life with natural motion and dynamics * **Start and end frame**: Animate from `image` towards an optional `image_end`, and Veo 3.1 interpolates the transition between both frames * **Multi-resolution output**: Generate videos in 720p, 1080p, or 4K resolution * **Audio generation**: Optional synchronized audio for complete multimedia content * **Flexible durations**: 4, 6, or 8 second video outputs * **Aspect ratio control**: 16:9 (landscape) or 9:16 (portrait) formats * **Motion guidance**: Use prompts to direct how the image animates * **Two generation modes**: Standard (highest quality) and Fast (cost-efficient) * **Flexible input**: Accept Base64 encoded images or HTTPS URLs ### Use cases * **Product animation**: Animate product photos for e-commerce and advertising * **Social media content**: Transform static posts into engaging video content * **Portrait animation**: Bring photos of people to life with natural movement * **Art animation**: Animate artwork, illustrations, and digital creations * **Marketing visuals**: Convert campaign images into dynamic video ads * **Storytelling**: Create video sequences from storyboard images ### Generate with Veo 3.1 Standard High-quality video generation with maximum fidelity and detail.
Create a new image-to-video task List all Veo 3.1 tasks Get task status by ID
### Generate with Veo 3.1 Fast Faster generation at reduced cost, ideal for prototyping and high-volume workflows.
Create a fast image-to-video task List all Veo 3.1 Fast tasks Get task status by ID
### Parameters | Parameter | Type | Required | Description | | ----------------- | --------- | -------- | ----------------------------------------------------------------------------------------------------------------- | | `image` | `string` | Yes | Input image as Base64 string or HTTPS URL (must be publicly accessible). Used as the first frame | | `image_end` | `string` | No | Optional last frame as Base64 string or HTTPS URL. Veo 3.1 interpolates from `image` towards it. Requires `image` | | `prompt` | `string` | Yes | Text describing the motion and animation (max 20,000 chars) | | `negative_prompt` | `string` | No | Text describing what to avoid in the video | | `duration` | `integer` | No | Video duration: `4`, `6`, or `8` seconds (default: `8`) | | `resolution` | `string` | No | Output resolution: `"720p"`, `"1080p"`, or `"4k"` (default: `"720p"`) | | `aspect_ratio` | `string` | No | Video format: `"16:9"` or `"9:16"` (default: `"16:9"`) | | `generate_audio` | `boolean` | No | Generate synchronized audio (default: `true`) | | `seed` | `integer` | No | Random seed for reproducibility | | `webhook_url` | `string` | No | URL for task completion notification | ## Frequently Asked Questions Veo 3.1 accepts images via **HTTPS URLs** (must be publicly accessible) or **Base64 encoded strings**. Common formats like JPEG, PNG, and WebP are supported. **Standard mode** produces the highest quality output with maximum detail and natural motion, ideal for final production content. **Fast mode** generates videos more quickly at a lower cost, perfect for prototyping and high-volume workflows. Yes. Send an optional `image_end` alongside `image` and Veo 3.1 interpolates the transition, so the video starts on `image` and lands on `image_end`. It works in both **Standard** and **Fast** modes and at every resolution, including 4K. `image` remains required — the model needs a first frame to interpolate from, so `image_end` on its own returns a `400`. Both frames accept HTTPS URLs or Base64, and they can differ in format (for example a PNG first frame and a JPEG last frame). Use the prompt to describe the transition itself, such as *"the scene morphs smoothly from the first frame to the last"*. The prompt guides how the image animates. Describe the desired motion, such as *"The subject slowly turns their head and smiles"* or *"Camera slowly zooms in while leaves blow in the wind"*. Be specific about movement direction and speed. Veo 3.1 supports **720p** (HD), **1080p** (Full HD), and **4K** (Ultra HD). The output resolution is independent of your input image resolution. When `generate_audio` is enabled (default), Veo 3.1 synthesizes audio that matches the visual content and motion. This includes ambient sounds and effects appropriate to the scene. Yes, set `aspect_ratio` to `"9:16"` to generate vertical videos. For best results, use portrait-oriented source images when creating vertical content. ## Best practices * **Image quality**: Use high-resolution, well-lit images with clear subjects for best results * **Motion prompts**: Describe specific movements like "slowly pans left" or "subject walks forward" * **Negative prompts**: Use to avoid unwanted artifacts like "blurry, distorted, flickering" * **Resolution matching**: Consider your source image quality when selecting output resolution * **Audio control**: Disable `generate_audio` if adding custom audio in post-production * **End frame framing**: When using `image_end`, keep both frames in the same aspect ratio and with a consistent subject and lighting — the closer they are, the smoother the interpolation. `aspect_ratio` still governs the output, so a mismatch is adapted rather than rejected * **Webhook integration**: Use webhooks for production workflows to handle async completion * **Seed for consistency**: Use the same seed to reproduce similar animation results ## Related APIs * **[Veo 3.1 Text-to-Video](/api-reference/text-to-video/veo-3-1/overview)**: Generate videos from text prompts with Veo 3.1 * **[Veo 3.1 Reference-to-Video](/api-reference/reference-to-video/veo-3-1/overview)**: Generate videos with consistent characters using reference images * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Image-to-video with advanced motion control * **[Kling O1](/api-reference/image-to-video/kling-o1/overview)**: Alternative image-to-video model * **[MiniMax Live](/api-reference/image-to-video/minimax-live)**: Fast image-to-video generation # WAN 2.5 1080p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-1080p post /v1/ai/image-to-video/wan-2-5-i2v-1080p Generate a 1080p video from image using the WAN 2.5 model. # WAN 2.5 I2V 1080p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-1080p-tasks get /v1/ai/image-to-video/wan-2-5-i2v-1080p Get the list of WAN 2.5 Image-to-Video 1080p tasks # WAN 2.5 480p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-480p post /v1/ai/image-to-video/wan-2-5-i2v-480p Generate a 480p video from image using the WAN 2.5 model. # WAN 2.5 I2V 480p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-480p-tasks get /v1/ai/image-to-video/wan-2-5-i2v-480p Get the list of WAN 2.5 Image-to-Video 480p tasks # WAN 2.5 720p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-720p post /v1/ai/image-to-video/wan-2-5-i2v-720p Generate a 720p video from image using the WAN 2.5 model. # WAN 2.5 I2V 720p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5-i2v-720p-tasks get /v1/ai/image-to-video/wan-2-5-i2v-720p Get the list of WAN 2.5 Image-to-Video 720p tasks # WAN 2.5 I2V 1080p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5/task-by-id-1080p get /v1/ai/image-to-video/wan-2-5-i2v-1080p/{task-id} Get the status of a WAN 2.5 Image-to-Video 1080p task # WAN 2.5 I2V 480p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5/task-by-id-480p get /v1/ai/image-to-video/wan-2-5-i2v-480p/{task-id} Get the status of a WAN 2.5 Image-to-Video 480p task # WAN 2.5 I2V 720p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-5/task-by-id-720p get /v1/ai/image-to-video/wan-2-5-i2v-720p/{task-id} Get the status of a WAN 2.5 Image-to-Video 720p task # WAN 2.7 - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-7/generate post /v1/ai/image-to-video/wan-2-7 Generate a video from an image or extend an existing video using WAN 2.7. **Three generation modes:** - **First frame**: Provide `start_image_url` alone to animate from a starting image - **First + last frame**: Provide both `start_image_url` and `end_image_url` for controlled start-to-end animation - **Video continuation**: Provide `video_url` to extend an existing video, optionally with `end_image_url` as the target ending frame **Key features:** - 720P and 1080P resolution support - Optional audio-guided generation - Duration range: 2-15 seconds # WAN 2.7 Image To Video API Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-7/overview Generate AI videos from images with WAN 2.7. First frame, first+last frame, and video continuation modes. 720P and 1080P resolution, 2-15 second durations, and audio-guided generation. WAN 2.7 Image-to-Video supports three generation modes: animate from a first frame, control start-to-end animation, or extend existing videos with optional audio guidance. WAN 2.7 Image-to-Video is an AI video generation API that creates MP4 videos from images or extends existing videos. It supports three distinct generation modes: first-frame animation, first+last frame controlled animation, and video continuation. Output is available at 720P (1280x720) or 1080P (1920x1080) resolution with durations from 2 to 15 seconds. The model also supports optional audio input and automatic prompt expansion. ### Key capabilities * **Three generation modes**: First frame only, first+last frame, and video continuation * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Flexible durations**: 2 to 15 seconds of video output * **Audio-guided generation**: Provide a WAV or MP3 audio file (2-30 seconds, max 15MB) to guide video creation * **Prompt expansion**: AI optimizer expands short prompts into detailed scripts for richer output * **Video extension**: Continue an existing MP4/MOV video (2-10 seconds, max 100MB) with new content * **Image constraints**: Supports JPEG, PNG, BMP, WEBP images (240-8000px per side, max 20MB) * **Async processing**: Webhook notifications or polling for task completion ### Generation modes | Mode | Required inputs | Best for | | ---------------------- | ------------------------------------------ | ------------------------------------------------- | | **First frame** | `start_image_url` | Animating a still image with AI-generated motion | | **First + last frame** | `start_image_url` + `end_image_url` | Controlled transition between two keyframes | | **Video continuation** | `video_url` (optionally + `end_image_url`) | Extending an existing video clip with new content | ### Use cases * **Product animation**: Bring product images to life with smooth motion and camera movements * **Marketing videos**: Animate brand imagery into short-form video content * **Social media content**: Create video posts from static images for TikTok, Instagram, and YouTube * **Video extension**: Extend short clips into longer narratives using video continuation * **Storyboarding**: Animate concept art or wireframes to preview motion sequences * **Creative exploration**: Experiment with first+last frame mode for controlled visual transitions ### API operations Generate videos by submitting an image or video to the API. The service returns a task ID for async polling or webhook notification.
Create a new image-to-video generation task List all WAN 2.7 I2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------------------- | --------- | ----------- | --------- | ------------------------------------------------------------------------------------- | | `prompt` | `string` | No | - | Text description to guide video motion and style. Max 5000 characters | | `negative_prompt` | `string` | No | - | Elements to avoid (e.g., "blurry, watermark"). Max 500 characters | | `start_image_url` | `string` | Conditional | - | URL of first-frame image (JPEG/PNG/BMP/WEBP, 240-8000px, max 20MB) | | `end_image_url` | `string` | No | - | URL of last-frame image. Use with `start_image_url` or `video_url` | | `video_url` | `string` | Conditional | - | URL of existing video to extend (MP4/MOV, 2-10s, max 100MB) | | `audio_url` | `string` | No | - | URL of audio file (WAV/MP3, 2-30s, max 15MB) to guide generation | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 2 to 15. For video continuation, this is total output length | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `additional_settings.prompt_extend` | `boolean` | No | `true` | Enable AI prompt expansion for richer output | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions WAN 2.7 Image-to-Video is an AI video generation API developed by Alibaba. You submit an image or video URL, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The model generates MP4 video at 720P or 1080P resolution in durations from 2 to 15 seconds. **First frame**: Provide `start_image_url` alone to animate from a starting image. **First + last frame**: Provide both `start_image_url` and `end_image_url` for controlled start-to-end animation. **Video continuation**: Provide `video_url` to extend an existing video, optionally with `end_image_url` as the target ending frame. WAN 2.7 accepts JPEG, PNG, BMP, and WEBP images via publicly accessible URLs. Images must be 240-8000 pixels per side with an aspect ratio between 1:8 and 8:1, and a maximum file size of 20MB. Video continuation accepts MP4 and MOV files via publicly accessible URLs. Input videos must be 2-10 seconds long, 240-4096 pixels per side, and under 100MB. Provide a WAV or MP3 audio file URL via the `audio_url` parameter. The audio must be 2-30 seconds long and under 15MB. WAN 2.7 uses the audio to influence the visual content and motion of the generated video. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use high-resolution images with clear subjects and balanced lighting. Avoid heavily compressed or noisy inputs. * **First + last frame**: Ensure both images share a similar visual style and subject for smooth transitions. * **Video continuation**: Input video duration (2-10s) counts toward total output duration. Plan accordingly. * **Prompt writing**: Even though prompts are optional for I2V, adding motion and camera directions improves results. * **Negative prompts**: Always include: "blurry, low quality, watermark, text, distortion, extra limbs" * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[WAN 2.7 Text-to-Video](/api-reference/text-to-video/wan-2-7/overview)**: Generate videos purely from text prompts with WAN 2.7 * **[WAN 2.7 Reference-to-Video](/api-reference/reference-to-video/wan-2-7/overview)**: Generate videos featuring characters from reference images or videos * **[WAN 2.6 Image-to-Video](/api-reference/wan-v2-6-i2v/overview)**: Previous WAN generation with multi-shot sequences * **[Kling 2.5 Turbo Pro](/api-reference/image-to-video/kling-v2.5-pro/overview)**: Alternative I2V model with cinematic quality # WAN 2.7 I2V - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-7/task-by-id get /v1/ai/image-to-video/wan-2-7/{task-id} Retrieve the status and result of a specific WAN 2.7 image-to-video task by its ID. # WAN 2.7 I2V - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-2-7/wan-2-7-i2v-tasks get /v1/ai/image-to-video/wan-2-7 Retrieve the list of all WAN 2.7 image-to-video tasks for the authenticated user. # WAN 2.2 480p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-480p/get-wan-v2-2-480p get /v1/ai/image-to-video/wan-v2-2-480p Get the list of WAN 2.2 480p tasks # WAN 2.2 480p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-480p/get-wan-v2-2-480p-task get /v1/ai/image-to-video/wan-v2-2-480p/{task-id} Get the status of the WAN 2.2 480p task # WAN 2.2 480p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-480p/post-wan-v2-2-480p post /v1/ai/image-to-video/wan-v2-2-480p Generate a video from image using the WAN 2.2 480p model. # WAN 2.2 580p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-580p/get-wan-v2-2-580p get /v1/ai/image-to-video/wan-v2-2-580p Get the list of WAN 2.2 580p tasks # WAN 2.2 580p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-580p/get-wan-v2-2-580p-task get /v1/ai/image-to-video/wan-v2-2-580p/{task-id} Get the status of the WAN 2.2 580p task # WAN 2.2 580p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-580p/post-wan-v2-2-580p post /v1/ai/image-to-video/wan-v2-2-580p Generate a video from image using the WAN 2.2 580p model. # WAN 2.2 720p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-720p/get-wan-v2-2-720p get /v1/ai/image-to-video/wan-v2-2-720p Get the list of WAN 2.2 720p tasks # WAN 2.2 720p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-720p/get-wan-v2-2-720p-task get /v1/ai/image-to-video/wan-v2-2-720p/{task-id} Get the status of the WAN 2.2 720p task # WAN 2.2 720p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-2-720p/post-wan-v2-2-720p post /v1/ai/image-to-video/wan-v2-2-720p Generate a video from image using the WAN 2.2 720p model. # WAN 2.6 1080p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6-1080p post /v1/ai/image-to-video/wan-v2-6-1080p Generate a 1080p video from image using the WAN 2.6 model. # WAN 2.6 I2V 1080p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6-1080p-tasks get /v1/ai/image-to-video/wan-v2-6-1080p Get the list of WAN 2.6 Image-to-Video 1080p tasks # WAN 2.6 720p - Create video from image Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6-720p post /v1/ai/image-to-video/wan-v2-6-720p Generate a 720p video from image using the WAN 2.6 model. # WAN 2.6 I2V 720p - List tasks Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6-720p-tasks get /v1/ai/image-to-video/wan-v2-6-720p Get the list of WAN 2.6 Image-to-Video 720p tasks # WAN 2.6 I2V 1080p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6/task-by-id-1080p get /v1/ai/image-to-video/wan-v2-6-1080p/{task-id} Get the status of a WAN 2.6 Image-to-Video 1080p task # WAN 2.6 I2V 720p - Get task status Source: https://docs.magnific.com/api-reference/image-to-video/wan-v2-6/task-by-id-720p get /v1/ai/image-to-video/wan-v2-6-720p/{task-id} Get the status of a WAN 2.6 Image-to-Video 720p task # Upscaler Creative - Get task status Source: https://docs.magnific.com/api-reference/image-upscaler-creative/get-image-upscaler get /v1/ai/image-upscaler/{task-id} Get the status of the upscaling task # Upscaler Creative - List tasks Source: https://docs.magnific.com/api-reference/image-upscaler-creative/get-image-upscaler-task get /v1/ai/image-upscaler Get the status of all upscaling tasks # Magnific Upscaler Creative API Source: https://docs.magnific.com/api-reference/image-upscaler-creative/image-upscaler Best AI image upscaler to enlarge pictures without losing quality. Prompt-guided enhancement, stylization, and 2x/4x/8x/16x upscaling. Creative mode adds or infers new detail guided by your prompt, enabling stylized, higher‑impact upscales. The Magnific Upscaler enhances images to higher resolutions. This powerful tool does more than just increase image size - it also improves quality and adds detail. By leveraging advanced upscaling technology, our API allows you to transform your images with custom prompts and fine-tuned parameters, achieving impressive results. Need faithful upscaling without adding new elements? See Upscaler Precision. ### Key capabilities * Prompt‑guided enhancement that can introduce or infer new detail * Stylization options to match aesthetics or art directions * Controls to balance fidelity vs creativity * Resolution growth (2x/4x/8x/16x) with quality improvement * Ideal for concept art, marketing creatives, and visual ideation ### Use Cases The Magnific Upscaler can be applied to a wide range of image types and industries, including: * Portraits: Enhance facial details and skin textures * Illustrations: Sharpen lines and improve color vibrancy * Video game assets: Upscale textures and character models * Landscapes: Bring out intricate details in nature scenes * Science fiction imagery: Enhance futuristic and space-themed visuals * Fantasy art: Improve magical and mythical elements * Film stills: Increase resolution for better visual quality * Graphic design: Upscale logos and marketing materials * Architectural renderings: Enhance building details and textures, jumping from quick sketches to high quality renders * Interior design: Improve quality of room layouts and furniture details * Food photography: Bring out textures and colors in culinary images ### Upscale an Image with Magnific Create an upscaling task for an image.
Upscale an image using Magnific
## Image Input Best Practices For maximum quality results, how you send your image matters: | Action | Quality Impact | | ------------------------------------------ | -------------------------- | | Send URL of original image | ✅ Maximum quality | | Send base64 of original file read directly | ✅ Maximum quality | | Use `canvas.toDataURL('image/jpeg')` | ❌ \~8% quality loss | | Use `canvas.toDataURL('image/jpeg', 0.8)` | ❌ \~20% quality loss | | Resize image before sending | ❌ Significant quality loss | | Convert PNG → JPEG before sending | ❌ Quality loss | **Recommended:** Send the original image via URL whenever possible. This ensures the upscaler receives the highest quality input, which directly impacts output quality. ## Frequently Asked Questions Yes! Magnific.ai is part of Magnific and we are happy to offer the same technology in an API service. If you tried Magnific.ai before, you will get the same technology and its future updates. The cost per image upscale is based on the output image area in pixels. This depends on two key factors: * The dimensions of the input image * The upscale factor applied (e.g.: 2x, 4x, 8x, 16x) The larger the output image area, the higher the cost per upscale. Our pricing tiers are applied based on the final output size, so you have complete transparency and control over your usage. Some examples: | Input Size | Output Size | Upscale Factor | Price | | ---------- | ----------- | -------------- | ----- | | 640x480 | 1280x960 | 2x | €0.10 | | 640x480 | 2560x1920 | 4x | €0.20 | | 640x480 | 5120x3840 | 8x | €0.50 | | 1280x720 | 2560x1440 | 2x | €0.10 | | 1280x720 | 5120x2880 | 4x | €0.40 | | 1920x1080 | 3840x2160 | 2x | €0.20 | The pricing structure ensures that you only pay for the level of upscaling you need. Use Upscaler Precision for faithful super‑resolution without hallucinations (best for logos, UI, text, and product photos). Choose Creative when you want to add or infer detail and stylize results via prompts. # Upscaler Creative - Upscale image Source: https://docs.magnific.com/api-reference/image-upscaler-creative/post-image-upscaler post /v1/ai/image-upscaler This asynchronous endpoint enables image upscaling using advanced AI algorithms. Upon submission, it returns a unique `task_id` which can be used to track the progress of the upscaling process. For real-time production use, include the optional `webhook_url` parameter to receive an automated notification once the task has been completed. This allows for seamless integration and efficient task management without the need for continuous polling. ## Important For a comprehensive guide on this service, including use cases and detailed functionality, check out our [Image Upscaler Guide](/api-reference/image-upscaler-creative/image-upscaler). ## Request # Upscaler Precision V2 - List tasks Source: https://docs.magnific.com/api-reference/image-upscaler-precision-v2/get-image-upscaler-precision-v2 get /v1/ai/image-upscaler-precision-v2 Returns a list of all precision upscaler V2 tasks. Each task includes its ID, current status, and output URL if completed. # Upscaler Precision V2 - Get task status Source: https://docs.magnific.com/api-reference/image-upscaler-precision-v2/get-{task-id}-by-id get /v1/ai/image-upscaler-precision-v2/{task-id} Returns the current status and output URL of a specific precision upscaler V2 task. The output URL is included only if the task has completed successfully. # Upscaler Precision V2 – AI Image Upscaling API Source: https://docs.magnific.com/api-reference/image-upscaler-precision-v2/overview Upscale images with Precision V2. Advanced detail enhancement with sharpen, smart grain, and ultra detail controls. Scale 2x to 16x with photo and illustration flavors. Precision V2 delivers enhanced upscaling with fine-grained controls for sharpness, texture, and detail while preserving image fidelity. Upscaler Precision V2 is an advanced image upscaling API that enhances resolution while intelligently adding details and improving image quality. Building on the original Precision upscaler, V2 introduces granular control over sharpness, grain texture, and ultra-fine detail enhancement. The API supports multiple image types through specialized processing flavors optimized for photos, illustrations, and noisy images. Looking for creative, prompt-guided upscaling? See Upscaler Creative. ### Key capabilities * **Scale factors 2x to 16x**: Multiply image dimensions from 2x up to 16x (up to 256x total pixels) * **Sharpen control**: Adjustable sharpening intensity (0-100, default 7) for edge definition and clarity * **Smart grain**: Intelligent texture enhancement (0-100, default 7) to preserve natural film grain and prevent artificial smoothness * **Ultra detail**: Fine detail enhancement level (0-100, default 30) for textures, patterns, and micro-details * **Processing flavors**: Three optimized modes: `sublime` for illustrations, `photo` for photographs, `photo_denoiser` for noisy images * **Async processing**: Webhook notifications or polling for task completion * **Flexible input**: Accepts HTTPS URLs or base64-encoded images ### Use cases * **E-commerce**: Upscale product photos with enhanced sharpness while preserving color accuracy and texture * **Photography restoration**: Enhance old photos with smart grain to maintain natural film characteristics * **Print production**: Scale digital images for large-format printing with ultra detail for crisp output * **Illustration enhancement**: Upscale digital art with the sublime flavor for smooth gradients and vibrant colors * **Real estate**: Enhance property photos with the photo flavor for realistic detail preservation * **Archival digitization**: Restore scanned documents and photos with noise reduction via photo\_denoiser ### Upscale images with Precision V2 Create an upscaling task by submitting a request to the API. The service returns a task ID for async polling or webhook notification.
Create a new image upscaling task List all Precision V2 tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | -------------- | --------------------- | -------- | ------- | ------------------------------------------------------------------------- | | `image` | `string` | Yes | - | Source image as HTTPS URL or base64-encoded string | | `scale_factor` | `integer` or `string` | No | - | Scaling multiplier: 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, or 16 | | `sharpen` | `integer` | No | `7` | Sharpness intensity (0-100). Higher values increase edge definition | | `smart_grain` | `integer` | No | `7` | Grain/texture enhancement (0-100). Preserves natural film grain | | `ultra_detail` | `integer` | No | `30` | Fine detail level (0-100). Controls textures and micro-details | | `flavor` | `string` | No | - | Processing mode: `sublime`, `photo`, or `photo_denoiser` | | `webhook_url` | `string` | No | - | URL for completion notification with task status and output URL | ## Frequently Asked Questions Upscaler Precision V2 is an AI-powered image upscaling API that increases image resolution while intelligently enhancing details. You submit an image via URL or base64, receive a task ID, then poll for results or receive a webhook notification when processing completes. The API adds fine details based on the content type and your parameter settings. Precision V2 introduces granular control parameters that V1 does not have: sharpen (edge definition), smart\_grain (texture preservation), and ultra\_detail (micro-detail enhancement). V2 also supports processing flavors optimized for different image types. Use V2 when you need fine-tuned control over the upscaling output. The three flavors are: **sublime** for artistic and illustrated images with smooth gradients and vibrant colors; **photo** for photographic images preserving natural colors and realistic details; **photo\_denoiser** for photos with noise reduction, ideal for low-light or grainy photographs. Choose based on your source image type. **Sharpen** (default 7) controls edge clarity and definition. **Smart grain** (default 7) adds natural texture to prevent artificial smoothness. **Ultra detail** (default 30) enhances fine patterns and micro-details. All range from 0-100. Start with defaults and adjust based on your specific image needs. Precision V2 supports scale factors from 2 to 16. A scale factor of 2 doubles dimensions (4x total pixels), while 16 increases dimensions 16-fold (256x total pixels). Higher scale factors increase processing time and output file size. Rate limits vary by subscription tier. See the Rate Limits page for current limits by plan. Pricing is based on output image area in pixels, determined by input dimensions and scale factor. See the Pricing page for current rates and subscription options. **Precision V2** focuses on faithful upscaling with controlled detail enhancement, preserving the original content while adding sharpness and texture. **Creative** uses prompts to introduce or infer new visual elements and stylization. Choose Precision V2 when fidelity matters; choose Creative when you want artistic enhancement. ## Best practices * **Input quality**: Start from the highest-quality source available. Avoid heavily compressed images as artifacts may be amplified * **Flavor selection**: Use `sublime` for illustrations and digital art, `photo` for natural photographs, `photo_denoiser` for noisy or low-light images * **Parameter tuning**: Start with default values (sharpen: 7, smart\_grain: 7, ultra\_detail: 30) and adjust incrementally based on results * **Scale factor**: Begin with 2x-4x for most use cases. Higher factors (8x-16x) work best with high-quality source images * **Grain for realism**: Increase smart\_grain for photographic images to avoid the "plastic" look common in AI upscaling * **Production integration**: Use webhooks instead of polling for scalable, event-driven applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Upscaler Creative](/api-reference/image-upscaler-creative/image-upscaler)**: Prompt-guided upscaling with creative detail generation and stylization * **[Upscaler Precision](/api-reference/image-upscaler-precision/image-upscaler)**: Original precision upscaler without V2's granular controls # Upscaler Precision V2 - Upscale image Source: https://docs.magnific.com/api-reference/image-upscaler-precision-v2/post-image-upscaler-precision-v2 post /v1/ai/image-upscaler-precision-v2 Upscales an image while adding new visual elements or details (V2). This endpoint may modify the original image content based on the prompt and inferred context. # Upscaler Precision - List tasks Source: https://docs.magnific.com/api-reference/image-upscaler-precision/get-image-upscaler-precision get /v1/ai/image-upscaler-precision Returns a list of all precision upscaler tasks. Each task includes its ID, current status, and output URL if completed. # Upscaler Precision - Get task status Source: https://docs.magnific.com/api-reference/image-upscaler-precision/get-{task-id}-by-id get /v1/ai/image-upscaler-precision/{task-id} Returns the current status and output URL of a specific precision upscaler task. The output URL is included only if the task has completed successfully. # Upscaler Precision – High‑Fidelity Super‑Resolution (No Hallucinations) Source: https://docs.magnific.com/api-reference/image-upscaler-precision/image-upscaler High‑fidelity AI upscaling that preserves original content, small text/UI, and fine details without adding new elements. Ideal for logos, UI, and product photos. Precision mode focuses on faithful super‑resolution without hallucinations—best choice when preserving content is critical. The Magnific Upscaler Precision focuses on faithful, high-fidelity upscaling. It increases resolution while preserving the original content, texture, and structure of the image. Unlike creative upscaling, Precision prioritizes clarity, sharpness, and detail recovery without inventing new elements. Looking for stylized or prompt‑guided enhancement? See Upscaler Creative. With advanced super-resolution techniques, Upscaler Precision: * Preserves edges and fine details while minimizing halos * Reduces noise and compression artifacts * Maintains color consistency and natural look * Enhances readability for small text and UI elements ### Use cases (Precision) Upscaler Precision is ideal when you need quality improvements without altering the content: * E‑commerce product photos: sharpen edges, remove artifacts, preserve true colors * Branding and logos: upscale raster logos and icons with crisp lines * UI assets and screenshots: improve sharpness while keeping pixel-accurate layouts * Photography restoration and scans: reduce noise and recover fine detail * Technical diagrams and architectural drawings: preserve thin lines and text * Social media and web images: upscale for higher-DPI displays without distortion ### Upscale an image with Upscaler Precision Create a precision upscaling task for an image.
Upscale an image with Precision mode
## Frequently Asked Questions Yes! Magnific.ai is part of Magnific and we are happy to offer the same technology in an API service. If you tried Magnific.ai before, you will get the same technology and its future updates. The cost per image upscale is based on the output image area in pixels. This depends on two key factors: * The dimensions of the input image * The upscale factor applied (e.g.: 2x, 4x, 8x, 16x) The larger the output image area, the higher the cost per upscale. Our pricing tiers are applied based on the final output size, so you have complete transparency and control over your usage. Some examples: | Input Size | Output Size | Upscale Factor | Price | | ---------- | ----------- | -------------- | ----- | | 640x480 | 1280x960 | 2x | €0.10 | | 640x480 | 2560x1920 | 4x | €0.20 | | 640x480 | 5120x3840 | 8x | €0.50 | | 1280x720 | 2560x1440 | 2x | €0.10 | | 1280x720 | 5120x2880 | 4x | €0.40 | | 1920x1080 | 3840x2160 | 2x | €0.20 | The pricing structure ensures that you only pay for the level of upscaling you need. Precision preserves the original content and improves clarity, detail, and sharpness without inventing new elements. It is ideal for logos, UI assets, product photos, scans, and technical diagrams. Creative can introduce or infer new visual details guided by prompts or context. It is better for artistic enhancement, stylization, or imaginative detail addition. Choose Precision when fidelity is critical: brand consistency, readable text, pixel-accurate UI, and product authenticity. Choose Creative when you want stylistic enhancements or new visual elements. Prefer PNG for graphics/logos and high-quality JPEG/PNG for photos. Avoid heavily compressed sources. For best results, use inputs large enough to contain the essential detail (e.g., at least 256–512 px on the shortest edge), then upscale 2x–4x. Precision improves edge sharpness and micro-contrast, which can make small text and UI icons more legible. Extremely low-resolution or heavily blurred originals may still have limits. ## Best practices for Upscaler Precision Follow these recommendations to get the best, most faithful results: * Start from the highest-quality source you have (avoid heavy-compressed inputs). * Prefer PNG or high-quality JPEG; keep artifacts minimal in the input. * Choose moderate upscale factors first (2x–4x) to avoid amplifying noise; iterate if needed. * Disable any creative prompts or stylization when fidelity is critical. * For logos/UI, use crisp sources with strong contrast; avoid anti-aliased tiny originals. * When upscaling scans, pre-trim borders and rotate/crop to the correct orientation. * Compare results at 100% zoom to validate line sharpness and color consistency. ### Image Input Quality How you send your image directly impacts the output quality: | Action | Quality Impact | | ------------------------------------------ | -------------------------- | | Send URL of original image | ✅ Maximum quality | | Send base64 of original file read directly | ✅ Maximum quality | | Use `canvas.toDataURL('image/jpeg')` | ❌ \~8% quality loss | | Use `canvas.toDataURL('image/jpeg', 0.8)` | ❌ \~20% quality loss | | Resize image before sending | ❌ Significant quality loss | | Convert PNG → JPEG before sending | ❌ Quality loss | **Recommended:** Send the original image via URL whenever possible. This ensures the upscaler receives the highest quality input. # Upscaler Precision - Upscale image Source: https://docs.magnific.com/api-reference/image-upscaler-precision/post-image-upscaler-precision post /v1/ai/image-upscaler-precision Upscales an image while adding new visual elements or details. This endpoint may modify the original image content based on the prompt and inferred context. # Improve Prompt - List tasks Source: https://docs.magnific.com/api-reference/improve-prompt/get-improve-prompt get /v1/ai/improve-prompt Get the status of all improve-prompt tasks # Improve Prompt - Get task status Source: https://docs.magnific.com/api-reference/improve-prompt/get-{task-id}-by-id get /v1/ai/improve-prompt/{task-id} Get the status of one improve-prompt task # Improve Prompt - AI Prompt Enhancement API | Magnific API Source: https://docs.magnific.com/api-reference/improve-prompt/overview Enhance text prompts for AI image and video generation. Transform basic descriptions into detailed, optimized prompts with language support. Improve generation results automatically. Transform simple text descriptions into optimized prompts for AI image and video generation, improving output quality automatically. Improve Prompt is an AI-powered prompt enhancement API that transforms basic text descriptions into detailed, optimized prompts for AI generation. Submit a simple prompt like "a cat" and receive an enhanced version with specific details about lighting, composition, style, and atmosphere. The API supports both image and video generation prompts, with multi-language output capabilities. ### Key capabilities * **Dual-mode enhancement**: Optimize prompts for either image or video generation with mode-specific improvements * **Multi-language support**: Output enhanced prompts in any language using ISO 639-1 codes (e.g., `en`, `es`, `fr`, `de`) * **Creative generation**: Submit an empty prompt to generate creative ideas from scratch * **Maximum prompt length**: Input prompts up to 2,500 characters for comprehensive enhancement * **Context-aware improvements**: Adds relevant details for lighting, composition, style, mood, and technical parameters * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Text-to-image workflows**: Enhance prompts before sending to Flux, Mystic, or other image generation APIs * **Video generation preparation**: Optimize prompts for Kling, WAN, or other video generation models * **Creative brainstorming**: Generate detailed creative prompts from minimal input or ideas * **Non-English content creation**: Create prompts in your preferred language for localized projects * **Batch processing pipelines**: Automate prompt enhancement in production workflows * **Quality improvement**: Consistently achieve better AI generation results with optimized prompts ### Enhance prompts with Improve Prompt Submit a text prompt with the target generation type (image or video). The service returns a task ID for async polling or webhook notification.
Create a new prompt enhancement task List all prompt enhancement tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | ----------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text prompt to improve for AI generation. Can be empty to generate a creative prompt. Maximum length: 2,500 characters. | | `type` | `string` | Yes | - | Type of generation to optimize for: `"image"` or `"video"` | | `language` | `string` | No | `"en"` | Language code for the improved prompt (ISO 639-1 format, e.g., `"en"`, `"es"`, `"fr"`) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Improve Prompt is an AI-powered API that enhances basic text descriptions into detailed, optimized prompts for AI image or video generation. You submit a simple prompt (e.g., "a sunset") with the target type (image or video), receive a task ID immediately, then poll for results or receive a webhook notification. The output is an enhanced prompt with added details about composition, lighting, style, and technical parameters. The `type` parameter determines how the prompt is optimized. For `"image"` prompts, the API adds details about composition, lighting, color palette, and visual style. For `"video"` prompts, it includes motion descriptions, camera movements, temporal flow, and cinematic elements that help video generation models produce better results. Yes. Use the `language` parameter with any ISO 639-1 language code (e.g., `"es"` for Spanish, `"fr"` for French, `"de"` for German, `"ja"` for Japanese). The enhanced prompt will be returned in your specified language. Default is English (`"en"`). When you submit an empty prompt, the API generates a creative prompt from scratch. This is useful for brainstorming or when you want AI-generated creative ideas for image or video generation. Input prompts can be up to 2,500 characters. This allows you to provide detailed context that the API will further enhance and optimize. Prompt enhancement typically completes within 2-5 seconds. For production workflows, use webhooks instead of polling to receive immediate notification when tasks complete. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Be specific**: Even simple prompts benefit from context - include subject, action, or mood hints * **Choose the right type**: Use `"image"` for static visuals and `"video"` for motion content * **Language matching**: Set the language parameter to match your target audience or downstream model preferences * **Chain with generation**: Pipe enhanced prompts directly to Flux, Mystic, Kling, or other generation APIs * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Magnific Mystic](/api-reference/mystic/mystic)**: Generate photorealistic images in 1K/2K/4K with LoRA styles * **[Flux Pro](/api-reference/text-to-image/flux-pro-v1-1/post-flux-pro-v1-1)**: High-quality text-to-image generation with Flux Pro * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2.5-pro/overview)**: Generate videos from images with Kling * **[Image to Prompt](/api-reference/image-to-prompt/post-image-to-prompt)**: Reverse process - generate prompts from images # Improve Prompt - Enhance prompt Source: https://docs.magnific.com/api-reference/improve-prompt/post-improve-prompt post /v1/ai/improve-prompt Enhance user prompts for AI image or video generation using advanced AI models. - **Image prompts**: Improve a prompt for image generation - **Video prompts**: Improve a prompt for video generation # Kling O1 - Get task status Source: https://docs.magnific.com/api-reference/kling-o1/task-by-id get /v1/ai/image-to-video/kling-o1/{task-id} # Latent Sync - List tasks Source: https://docs.magnific.com/api-reference/lip-sync/latent-sync/get-latent-sync get /v1/ai/lip-sync/latent-sync Get the status of all Latent-Sync lip-sync tasks # Latent Sync - Get task status Source: https://docs.magnific.com/api-reference/lip-sync/latent-sync/get-{task-id}-by-id get /v1/ai/lip-sync/latent-sync/{task-id} Get the status of one Latent-Sync lip-sync task # Latent Sync – Lip Sync API Source: https://docs.magnific.com/api-reference/lip-sync/latent-sync/overview Advanced AI-powered lip synchronization technology for realistic facial animation and speech-to-video matching with high-quality, production-ready outputs. Powered by Latent Sync technology, this API provides state-of-the-art lip synchronization capabilities for creating realistic talking avatar videos from audio input. Latent Sync is an advanced AI-powered lip synchronization solution that creates realistic facial animations synchronized with audio input. It generates high-quality talking avatar videos by intelligently matching mouth movements to speech patterns, delivering natural and believable results suitable for production environments. ### Key capabilities * High-quality lip synchronization with natural facial movements * Support for multiple languages and accents * Realistic expression preservation during speech animation * Production-ready video outputs with consistent quality * Fast processing times optimized for real-time applications ### Use cases * Avatar creation for virtual presentations and digital content * Video dubbing and localization with synchronized lip movements * Interactive chatbots and virtual assistants with realistic speech * Educational content with animated instructors or characters * Marketing videos with personalized spokesperson animations
Create a Latent Sync lip sync task Check Latent Sync task status List Latent Sync tasks
## Frequently Asked Questions Latent Sync supports common audio formats including MP3, WAV, and AAC. The API automatically processes the audio to extract speech patterns for optimal lip synchronization. Latent Sync uses advanced AI models trained specifically for speech-to-lip mapping, providing highly accurate synchronization that maintains natural facial expressions and realistic mouth movements. Yes, you can provide your own base images or videos that will be animated with synchronized lip movements. The system works best with clear, front-facing portraits. Latent Sync supports multiple languages and can handle various accents and speech patterns. The AI model adapts to different linguistic characteristics for optimal results. Yes, Latent Sync generates production-quality videos suitable for commercial applications, marketing content, and professional presentations. # Latent Sync - Lip-sync video generation Source: https://docs.magnific.com/api-reference/lip-sync/latent-sync/post-latent-sync post /v1/ai/lip-sync/latent-sync Generate lip-synced video by synchronizing a video with an audio file using AI Latent-Sync technology # Veed Fabric 1.0 Fast - Generate talking video Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0-fast/generate post /v1/ai/lip-sync/veed-fabric-1-0-fast Generate a realistic talking video by combining a static portrait image with an audio file using Veed Fabric 1.0 Fast. This is the faster variant of Veed Fabric 1.0, optimized for reduced generation time while maintaining quality lip synchronization. Ideal for workflows requiring quick turnaround on talking head video generation. # Veed Fabric 1.0 Fast API Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0-fast/overview Generate talking videos with faster processing using Veed Fabric 1.0 Fast. Lip-synced MP4 output at 720p or 480p from a portrait image and audio file. Powered by Veed Fabric 1.0 Fast technology, this API generates talking head videos with optimized processing time for faster turnaround. Veed Fabric 1.0 Fast is an AI-powered lip synchronization API optimized for reduced generation time. It creates talking videos from a static portrait image and an audio file, producing an MP4 video with natural lip movements synchronized to the speech. This variant prioritizes faster processing while maintaining quality lip-sync results, making it suitable for workflows that require quick turnaround. Output resolution is configurable at 720p (1280x720) or 480p (854x480). ### Key capabilities * **Faster processing**: Optimized generation pipeline for reduced turnaround time compared to Veed Fabric 1.0 * **Image + audio input**: Combine a portrait photo with an audio file (MP3, WAV, M4A) to generate a talking video * **Resolution options**: Output at `720p` (1280x720) for production use or `480p` (854x480) for drafts * **Natural lip sync**: AI-driven mouth movement generation that matches speech patterns * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Rapid prototyping**: Quickly preview talking head videos before committing to full-quality generation * **High-volume production**: Process large batches of lip-sync videos with faster turnaround * **Real-time content**: Create time-sensitive marketing or social media videos * **A/B testing**: Generate multiple video variants quickly for testing different audio or images * **Live campaigns**: Produce personalized video responses at scale with minimal latency ### Generate talking video with Veed Fabric 1.0 Fast Submit a portrait image URL and audio file URL along with the desired resolution. The service returns a task ID for async polling or webhook notification.
Create a new fast lip-sync video generation task List all Veed Fabric 1.0 Fast tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | ------------------------------------------------------------------------------------------------------------------- | | `image_url` | `string` | Yes | - | URL of the portrait image to animate. Must be publicly accessible. Use a clear, front-facing photo for best results | | `audio_url` | `string` | Yes | - | URL of the audio file for lip synchronization. Supported formats: MP3, WAV, M4A. Must be publicly accessible | | `resolution` | `string` | Yes | - | Output video resolution: `"720p"` (1280x720) or `"480p"` (854x480) | | `webhook_url` | `string` | No | - | HTTPS URL for task completion notification | ## Frequently Asked Questions Veed Fabric 1.0 Fast is an optimized variant of Veed Fabric 1.0 designed for faster lip-sync video generation. You submit an image URL and audio URL via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The output is an MP4 video with the person speaking in sync with the provided audio. Veed Fabric 1.0 Fast accepts common image formats via a publicly accessible URL. For best lip-sync results, use a clear, front-facing portrait photo where the face is fully visible and well-lit. Veed Fabric 1.0 Fast supports MP3, WAV, and M4A audio formats. The audio file must be hosted at a publicly accessible URL. Two resolutions are available: `720p` (1280x720) for production-quality output and `480p` (854x480) for drafts or faster processing. Veed Fabric 1.0 Fast is optimized for reduced generation time while maintaining quality lip synchronization. Veed Fabric 1.0 prioritizes maximum generation fidelity. Choose Veed Fabric 1.0 Fast when speed and throughput are the priority, or Veed Fabric 1.0 when you need the highest quality output. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use well-lit, front-facing portrait photos with a clearly visible face for the most accurate lip sync * **Audio clarity**: Clean audio with minimal background noise produces better lip-sync accuracy * **Resolution choice**: Use `720p` for final production videos and `480p` for quick previews or prototyping * **Speed vs quality**: Use Veed Fabric 1.0 Fast for high-volume or time-sensitive workflows; switch to Veed Fabric 1.0 for maximum quality * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Veed Fabric 1.0](/api-reference/lip-sync/veed-fabric-1-0/overview)**: Higher fidelity lip-sync generation for quality-focused workflows * **[Latent Sync](/api-reference/lip-sync/latent-sync/overview)**: Alternative lip synchronization technology * **[Voiceover](/api-reference/voiceover/overview)**: Generate speech audio from text to use as lip-sync input * **[Sound Effects](/api-reference/sound-effects/overview)**: Generate sound effects from text descriptions # Veed Fabric 1.0 Fast - Get task status Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0-fast/task-by-id get /v1/ai/lip-sync/veed-fabric-1-0-fast/{task-id} Retrieve a specific Veed Fabric 1.0 Fast lip-sync task by its ID, including generation status and result URL when completed. # Veed Fabric 1.0 Fast - List tasks Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0-fast/veed-fabric-1-0-fast-tasks get /v1/ai/lip-sync/veed-fabric-1-0-fast Retrieve all Veed Fabric 1.0 Fast lip-sync tasks for the authenticated user. # Veed Fabric 1.0 - Generate talking video Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0/generate post /v1/ai/lip-sync/veed-fabric-1-0 Generate a realistic talking video by combining a static portrait image with an audio file using Veed Fabric 1.0. The model produces a lip-synced video (MP4) where the person in the image speaks naturally in sync with the provided audio. Ideal for creating talking head videos from a single photo and voice recording. # Veed Fabric 1.0 API Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0/overview Generate realistic talking videos from a portrait image and audio file with Veed Fabric 1.0. Lip-synced MP4 output at 720p or 480p resolution for marketing, e-learning, and content creation. Powered by Veed Fabric 1.0 technology, this API generates realistic talking head videos by synchronizing lip movements to audio input. Veed Fabric 1.0 is an AI-powered lip synchronization API that creates realistic talking videos from a static portrait image and an audio file. The model analyzes the speech patterns in the audio and animates the person in the image with natural lip movements, producing an MP4 video where the subject appears to speak in sync with the provided audio. Output resolution is configurable at 720p (1280x720) or 480p (854x480). ### Key capabilities * **Image + audio input**: Combine a portrait photo with an audio file (MP3, WAV, M4A) to generate a talking video * **Resolution options**: Output at `720p` (1280x720) for production use or `480p` (854x480) for drafts and faster processing * **Natural lip sync**: AI-driven mouth movement generation that matches speech patterns accurately * **MP4 video output**: Generates a downloadable MP4 video file with the animated portrait * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing videos**: Create personalized spokesperson videos from a single photo * **E-learning**: Animate instructors for training materials and educational content * **Social media**: Generate talking head content at scale for multiple platforms * **Customer support**: Build video responses with consistent brand representatives * **Localization**: Produce dubbed videos with lip-synced speech in different languages ### Generate talking video with Veed Fabric 1.0 Submit a portrait image URL and audio file URL along with the desired resolution. The service returns a task ID for async polling or webhook notification.
Create a new lip-sync video generation task List all Veed Fabric 1.0 tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | ------------------------------------------------------------------------------------------------------------------- | | `image_url` | `string` | Yes | - | URL of the portrait image to animate. Must be publicly accessible. Use a clear, front-facing photo for best results | | `audio_url` | `string` | Yes | - | URL of the audio file for lip synchronization. Supported formats: MP3, WAV, M4A. Must be publicly accessible | | `resolution` | `string` | Yes | - | Output video resolution: `"720p"` (1280x720) or `"480p"` (854x480) | | `webhook_url` | `string` | No | - | HTTPS URL for task completion notification | ## Frequently Asked Questions Veed Fabric 1.0 is an AI lip-sync API that generates realistic talking videos from a static portrait image and an audio file. You submit an image URL and audio URL via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The output is an MP4 video where the person in the image speaks in sync with the provided audio. Veed Fabric 1.0 accepts common image formats via a publicly accessible URL. For best lip-sync results, use a clear, front-facing portrait photo where the face is fully visible and well-lit. Veed Fabric 1.0 supports MP3, WAV, and M4A audio formats. The audio file must be hosted at a publicly accessible URL. The model analyzes speech patterns in the audio to generate matching lip movements. Two resolutions are available: `720p` (1280x720) for production-quality output and `480p` (854x480) for drafts or faster processing. The resolution is set via the required `resolution` parameter. Veed Fabric 1.0 prioritizes generation quality with higher fidelity lip synchronization. Veed Fabric 1.0 Fast is optimized for reduced generation time while maintaining quality lip-sync results. Choose Veed Fabric 1.0 when output quality is the priority, or Veed Fabric 1.0 Fast when speed is more important. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use well-lit, front-facing portrait photos with a clearly visible face for the most accurate lip sync * **Audio clarity**: Clean audio with minimal background noise produces better lip-sync accuracy * **Resolution choice**: Use `720p` for final production videos and `480p` for quick previews or prototyping * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Veed Fabric 1.0 Fast](/api-reference/lip-sync/veed-fabric-1-0-fast/overview)**: Faster lip-sync generation with optimized processing time * **[Latent Sync](/api-reference/lip-sync/latent-sync/overview)**: Alternative lip synchronization technology * **[Voiceover](/api-reference/voiceover/overview)**: Generate speech audio from text to use as lip-sync input * **[Sound Effects](/api-reference/sound-effects/overview)**: Generate sound effects from text descriptions # Veed Fabric 1.0 - Get task status Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0/task-by-id get /v1/ai/lip-sync/veed-fabric-1-0/{task-id} Retrieve a specific Veed Fabric 1.0 lip-sync task by its ID, including generation status and result URL when completed. # Veed Fabric 1.0 - List tasks Source: https://docs.magnific.com/api-reference/lip-sync/veed-fabric-1-0/veed-fabric-1-0-tasks get /v1/ai/lip-sync/veed-fabric-1-0 Retrieve all Veed Fabric 1.0 lip-sync tasks for the authenticated user. # Music Generation - Generate from text Source: https://docs.magnific.com/api-reference/music-generation/generate post /v1/ai/music-generation Generate original music tracks from text descriptions using AI. Create high-quality music compositions based on your text prompts. Specify genre, mood, instruments, and tempo to get exactly the sound you need. Perfect for video production, game development, podcasts, and multimedia projects. **Tips for effective prompts:** - Include genre: "jazz", "electronic", "classical", "rock" - Describe mood: "upbeat", "melancholic", "energetic", "peaceful" - Mention instruments: "piano", "guitar", "drums", "synthesizer" - Add tempo hints: "slow", "fast-paced", "moderate groove" # Google Lyria - Generate music from text Source: https://docs.magnific.com/api-reference/music-generation/google-lyria/generate post /v1/ai/music-generation/google-lyria Generate original 30-second music tracks from text descriptions using Google Lyria. Create high-quality instrumental music by describing the genre, mood, instruments, and tempo you want. Ideal for video production, games, podcasts, and multimedia projects. Each request produces a single track with a fixed duration of 30 seconds. **Tips for effective prompts:** - Include genre: "lo-fi hip hop", "orchestral", "synthwave", "acoustic folk" - Describe mood: "uplifting", "melancholic", "energetic", "relaxed" - Mention instruments: "piano", "strings", "electric guitar", "808 drums" - Add tempo hints: "slow tempo", "upbeat", "driving rhythm" # Google Lyria - List tasks Source: https://docs.magnific.com/api-reference/music-generation/google-lyria/google-lyria-tasks get /v1/ai/music-generation/google-lyria Get the status of all Google Lyria music generation tasks # Google Lyria API Source: https://docs.magnific.com/api-reference/music-generation/google-lyria/overview Generate original 30-second music tracks from text with Google Lyria. Describe genre, mood, and instruments to create instrumental audio for video, games, and ads. Powered by Google Lyria, this API generates original instrumental music tracks from text descriptions. Google Lyria is an AI-powered music generation API that creates original instrumental music from text descriptions. Describe the genre, mood, instruments, and tempo you want, and receive an audio track delivered as a URL. Each request produces a single track with a fixed duration of 30 seconds, making it ideal for video production, games, podcasts, and multimedia projects. ### Key capabilities * **Text-guided generation**: describe any music style to generate (e.g., "Uplifting orchestral score with soaring strings", "Warm lo-fi hip hop beat with vinyl crackle") * **Fixed 30-second tracks**: each request produces a single 30-second instrumental track * **Negative prompt**: exclude unwanted instruments, styles, or moods with `negative_prompt` * **Reproducible output**: pass a `seed` to get a consistent result for the same prompt * **Audio URL output**: the generated track is delivered via a URL for download * **Async processing**: webhook notifications or polling for task completion ### Use cases * **Video production**: background music for videos and presentations * **Game development**: custom instrumental loops for games and apps * **Content creation**: unique music beds for podcasts, streams, and social media * **Advertising**: original background tracks for ads * **Film production**: temp tracks or production music for films ### Generate music with Google Lyria Submit a text description of the music you want to create. The service returns a task ID for async polling or webhook notification.
Create a new Google Lyria music generation task List all Google Lyria tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------- | -------------- | -------- | ------- | -------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the music to generate (1-2500 characters). Be specific about genre, mood, instruments, and tempo | | `negative_prompt` | `string` | No | - | Elements to avoid in the generated track (e.g., `"vocals, distortion, dissonance"`) | | `seed` | `integer` | No | - | Seed for reproducible generation; same prompt + seed returns a consistent result. Omit for a random seed | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ### Prompt tips For best results, include details about: * **Genre**: lo-fi hip hop, orchestral, synthwave, acoustic folk, etc. * **Mood**: uplifting, melancholic, energetic, relaxed * **Instruments**: piano, strings, electric guitar, 808 drums, etc. * **Tempo**: slow tempo, upbeat, driving rhythm **Example prompts:** * "Uplifting orchestral score with soaring strings and triumphant brass" * "Warm lo-fi hip hop beat with vinyl crackle and mellow keys" * "Energetic synthwave track with driving bass and retro leads" * "Peaceful acoustic folk with fingerpicked guitar and soft strings" ## Frequently Asked Questions Google Lyria is an AI-powered text-to-music API. You submit a text description of the music you want (genre, mood, instruments, tempo), receive a task ID immediately, then poll for results or receive a webhook notification. Each request returns a single 30-second instrumental track delivered via a URL. Every Google Lyria request produces a single track with a fixed duration of 30 seconds. There is no duration parameter. Google Lyria returns an audio track delivered via a URL that remains accessible for download after task completion. Pass the same `seed` together with the same `prompt`. Identical inputs return a consistent result across requests. Omit `seed` to get a random seed on each request. Use `negative_prompt` to describe elements to avoid, such as unwanted instruments, styles, or moods (for example `"vocals, distortion, harsh percussion"`). Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Prompt specificity**: use detailed descriptions of genre, mood, instruments, and tempo for better accuracy * **Negative prompts**: exclude unwanted elements to steer the composition * **Reproducibility**: fix `seed` to compare prompt variations against a stable baseline * **Production integration**: use webhooks instead of polling for scalable applications * **Error handling**: implement retry logic with exponential backoff for 503 errors ## Related APIs * **Lyria 3**: text-to-music with selectable `clip` or `pro` model variants and optional reference images. See the [Lyria 3 overview](/api-reference/music-generation/lyria-3/overview). * **[ElevenLabs Music](/api-reference/music-generation/overview)**: generate 10-240 second music tracks from text * **[Sound Effects](/api-reference/sound-effects/overview)**: generate sound effects from text descriptions * **[Audio Isolation](/api-reference/audio-isolation/overview)**: isolate specific sounds from audio files # Google Lyria - Get task status Source: https://docs.magnific.com/api-reference/music-generation/google-lyria/task-by-id get /v1/ai/music-generation/google-lyria/{task-id} Get the status of one Google Lyria music generation task # Lyria 3 - Generate music from text Source: https://docs.magnific.com/api-reference/music-generation/lyria-3/generate post /v1/ai/music-generation/lyria-3 Generate original music tracks from text descriptions using Lyria 3, with optional image references to guide mood and style. Describe the genre, mood, instruments, and tempo you want, and optionally attach reference images that inform the atmosphere of the composition. Ideal for video production, games, podcasts, and multimedia projects. **Tips for effective prompts:** - Include genre: "cinematic", "jazz", "electronic", "ambient" - Describe mood: "dramatic", "playful", "tense", "peaceful" - Mention instruments: "grand piano", "saxophone", "synth pads", "percussion" - Add tempo hints: "andante", "fast-paced", "steady groove" # Lyria 3 - List tasks Source: https://docs.magnific.com/api-reference/music-generation/lyria-3/lyria-3-tasks get /v1/ai/music-generation/lyria-3 Get the status of all Lyria 3 music generation tasks # Google Lyria 3 API Source: https://docs.magnific.com/api-reference/music-generation/lyria-3/overview Generate original music from text with Google Lyria 3. Choose clip or pro model variants and guide the mood with optional reference images for video, games, and ads. Powered by Google Lyria 3, this API generates original music tracks from text descriptions with selectable model variants. Google Lyria 3 is an AI-powered music generation API that creates original music from text descriptions. Describe the genre, mood, instruments, and tempo you want, choose between the `clip` and `pro` model variants, and optionally attach reference images to inform the atmosphere of the composition. The generated track is delivered as a URL, making it ideal for video production, games, podcasts, and multimedia projects. ### Key capabilities * **Text-guided generation**: describe any music style to generate (e.g., "Cinematic ambient track with warm synth pads and soft piano") * **Model variants (`clip` or `pro`)**: `clip` for faster generation optimized for short clips (default), `pro` for higher-fidelity, professional-quality tracks * **Reference images**: guide mood and style with optional reference images (PNG, JPEG, or WebP) via public URL or GCS format * **Audio URL output**: the generated track is delivered via a URL for download * **Async processing**: webhook notifications or polling for task completion ### Use cases * **Video production**: background music tailored to a scene's visual mood * **Game development**: custom soundtracks for games and apps * **Content creation**: unique music for podcasts, streams, and social media * **Advertising**: original background tracks and cues for ads * **Film production**: temp tracks or production music guided by reference imagery ### Generate music with Lyria 3 Submit a text description of the music you want to create, choose a model variant, and optionally attach reference images. The service returns a task ID for async polling or webhook notification.
Create a new Lyria 3 music generation task List all Lyria 3 tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------ | -------------- | -------- | ------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the music to generate (10-2000 characters). Be specific about genre, mood, instruments, and tempo | | `model` | `string` | No | `clip` | Model variant: `clip` (faster, optimized for short clips) or `pro` (higher-fidelity, professional-quality) | | `reference_images` | `array` | No | - | Optional reference images to guide mood and style. Each entry requires `image` (public URL or GCS) and `mime_type` (`image/png`, `image/jpeg`, or `image/webp`) | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | The `reference_images` array items have the following shape: | Field | Type | Required | Description | | ----------- | -------- | -------- | ---------------------------------------------------------------------------- | | `image` | `string` | Yes | Image URL (publicly accessible URL or GCS format) | | `mime_type` | `string` | Yes | MIME type of the reference image: `image/png`, `image/jpeg`, or `image/webp` | ### Prompt tips For best results, include details about: * **Genre**: cinematic, jazz, electronic, ambient, etc. * **Mood**: dramatic, playful, tense, peaceful * **Instruments**: grand piano, saxophone, synth pads, percussion, etc. * **Tempo**: andante, fast-paced, steady groove **Example prompts:** * "Cinematic ambient track with warm synth pads, soft piano, and a gentle build" * "Dramatic orchestral cue with rising strings and deep percussion" * "Playful jazz trio with upright bass, brushed drums, and grand piano" * "Tense electronic score with pulsing synths and a steady groove" ## Frequently Asked Questions Google Lyria 3 is an AI-powered text-to-music API built on Google's Lyria music generation model. You submit a text description of the music you want, choose a model variant, and optionally attach reference images. You receive a task ID immediately, then poll for results or receive a webhook notification. The generated track is delivered via a URL. `clip` (default) is optimized for faster generation of short clips, while `pro` produces higher-fidelity, professional-quality tracks. Select the variant with the `model` parameter. Attach one or more entries in `reference_images` to guide the mood and style of the composition. Each entry needs a publicly accessible `image` URL (or GCS format) and its `mime_type` (`image/png`, `image/jpeg`, or `image/webp`). Lyria 3 returns an audio track delivered via a URL that remains accessible for download after task completion. The `prompt` must be between 10 and 2000 characters. Be specific about genre, mood, instruments, and tempo for the best results. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Model selection**: use `clip` for fast iteration and short clips; switch to `pro` for final, professional-quality tracks * **Reference images**: attach imagery that captures the target atmosphere to steer mood and style * **Prompt specificity**: describe genre, mood, instruments, and tempo in detail for better accuracy * **Production integration**: use webhooks instead of polling for scalable applications * **Error handling**: implement retry logic with exponential backoff for 503 errors ## Related APIs * **Google Lyria**: text-to-music that produces fixed 30-second instrumental tracks. See the [Google Lyria overview](/api-reference/music-generation/google-lyria/overview). * **[ElevenLabs Music](/api-reference/music-generation/overview)**: generate 10-240 second music tracks from text * **[Sound Effects](/api-reference/sound-effects/overview)**: generate sound effects from text descriptions * **[Audio Isolation](/api-reference/audio-isolation/overview)**: isolate specific sounds from audio files # Lyria 3 - Get task status Source: https://docs.magnific.com/api-reference/music-generation/lyria-3/task-by-id get /v1/ai/music-generation/lyria-3/{task-id} Get the status of one Lyria 3 music generation task # Music Generation - List tasks Source: https://docs.magnific.com/api-reference/music-generation/music-generation-tasks get /v1/ai/music-generation Get the status of all music-generation tasks # ElevenLabs Music API Source: https://docs.magnific.com/api-reference/music-generation/overview Generate original music from text with ElevenLabs Music. Create 10-240 second tracks in any genre for videos, games, and ads. Powered by ElevenLabs technology, this API generates original music tracks from text descriptions. ElevenLabs Music is an AI-powered music generation API that creates original music tracks from text descriptions. Describe the style, mood, instruments, and tempo you want, and receive a high-quality MP3 audio file. The API supports customizable track durations from 10 seconds to 4 minutes, with typical generation times of 30-90 seconds. ### Key capabilities * **Text-guided generation**: Describe any music style to generate (e.g., "Upbeat jazz with piano and drums", "Cinematic orchestral soundtrack") * **Customizable duration**: Generate tracks from 10 seconds to 240 seconds (4 minutes) * **High-quality output**: Professional-grade MP3 audio files suitable for production use * **Genre flexibility**: Support for any music genre, style, or mood * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Video production**: Create background music for videos and presentations * **Game development**: Generate custom soundtracks for games and apps * **Content creation**: Produce unique music for podcasts, streams, and social media * **Advertising**: Create original jingles and background tracks for ads * **Film production**: Generate temp tracks or production music for films ### Generate music with ElevenLabs Submit a text description of the music you want to create along with the desired duration. The service returns a task ID for async polling or webhook notification.
Create a new music generation task List all music generation tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------- | --------- | -------- | ------- | ------------------------------------------------------------------------------------------------------ | | `prompt` | `string` | Yes | - | Text description of the music to generate (e.g., "Upbeat jazz with piano", "Melancholic piano ballad") | | `music_length_seconds` | `integer` | Yes | - | Duration of the generated track in seconds (10-240) | | `webhook_url` | `string` | No | - | HTTPS URL for task completion notification | ### Prompt tips For best results, include details about: * **Genre**: jazz, electronic, classical, rock, ambient, etc. * **Mood**: upbeat, melancholic, energetic, peaceful, dramatic * **Instruments**: piano, guitar, drums, synthesizer, strings, etc. * **Tempo**: slow, moderate, fast-paced, groove **Example prompts:** * "A Blues guitar solo with emotional bends and slides" * "Upbeat electronic dance music with synthesizers and heavy bass" * "Peaceful ambient piano with gentle strings in the background" * "Cinematic orchestral piece building to an epic climax" * "Modern corporate background music with light piano and subtle electronic elements" ## Frequently Asked Questions ElevenLabs Music is an AI-powered text-to-music API. You submit a text description of the music you want (genre, mood, instruments, tempo) along with the desired duration, receive a task ID immediately, then poll for results or receive a webhook notification. The output is a high-quality MP3 audio file. You can generate music tracks from 10 seconds (minimum) to 240 seconds / 4 minutes (maximum). The duration is specified in the `music_length_seconds` parameter. Be specific about genre, mood, instruments, and tempo. Good examples: "Upbeat jazz with piano and brushed drums", "Melancholic acoustic guitar ballad", "Epic orchestral soundtrack with building tension". Avoid vague prompts like just "music" or "song". ElevenLabs Music outputs MP3 audio files at professional quality. The generated audio is delivered via a URL that remains accessible for download after task completion. Generation typically takes 30-90 seconds depending on track duration. Shorter tracks (10-30 seconds) usually complete in under a minute, while maximum-length tracks (4 minutes) may take 1-2 minutes. For production workflows, use webhooks for efficient notification. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Prompt specificity**: Use detailed descriptions for better generation accuracy * **Duration planning**: Consider your use case when selecting track length * **Genre mixing**: Combine styles for unique results (e.g., "jazz-influenced electronic") * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Sound Effects](/api-reference/sound-effects/post-sound-effects)**: Generate sound effects from text descriptions * **[Audio Isolation](/api-reference/audio-isolation/overview)**: Isolate specific sounds from audio files * **[Lip Sync](/api-reference/lip-sync/latent-sync/overview)**: Synchronize lip movements to audio tracks * **[OmniHuman 1.5](/api-reference/video/omni-human-1-5)**: Generate human animations driven by audio # Music Generation - Get task status Source: https://docs.magnific.com/api-reference/music-generation/task-by-id get /v1/ai/music-generation/{task-id} Get the status of one music-generation task # Download music Source: https://docs.magnific.com/api-reference/music/download-music get /v1/music/{music-id}/download Get a direct download URL for a music audio file. Returns the music ID, title, and a CDN-hosted download link. Get a direct download URL for a music audio file. See the [Music API overview](/api-reference/music/overview) for authentication and usage guidance. # Get detailed music information by ID Source: https://docs.magnific.com/api-reference/music/get-music-by-id get /v1/music/{music-id} Retrieve full details for a music item including artist biography, genre and mood metadata, popularity score, and download statistics. Retrieve full details for a music item including artist biography, genre and mood metadata, popularity score, and download statistics. See the [Music API overview](/api-reference/music/overview) for authentication and usage guidance. # Magnific Music API Source: https://docs.magnific.com/api-reference/music/overview Search, filter, and download royalty-free music from the Magnific Music catalog via the Magnific API. The Magnific Music API gives you programmatic access to the Magnific Music catalog. Search music by keyword, filter by genre and mood, and download audio files for use in your applications. ### What you can do * Search music with full-text queries across titles and tags * Filter by genre, mood, premium status, and creation date * Sort results by relevance, popularity, creation date, duration, or tempo * Get detailed music metadata including artist biography, download stats, and playlist count * Download music audio files via CDN-hosted URLs ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication). ### Endpoints
Search and filter music Get detailed music information by ID Download a music audio file
### Parameters overview The search endpoint supports the following filters: | Parameter | Type | Description | | ----------------- | --------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `q` | `string` | Full-text search across titles and tags | | `genre` | `string` | Filter by genre name, comma-separated (e.g., `"Electronic,Ambient"`) | | `mood` | `string` | Filter by mood tag, comma-separated (e.g., `"Happy,Energetic"`) | | `include-premium` | `boolean` | Include premium content (default: `false`) | | `time_range` | `string` | Filter by creation date: `"7d"`, `"30d"`, or `"90d"` | | `order_by` | `string` | Sort order: `"relevance"`, `"popularity"`, `"-popularity"`, `"created_at"`, `"-created_at"`, `"title"`, `"-title"`, `"seconds"`, `"-seconds"`, `"bpm"`, `"-bpm"`. Defaults to `"relevance"` when `q` is provided, `"-popularity"` otherwise | | `limit` | `integer` | Results per page: 1-1000 (default: `10`) | | `offset` | `integer` | Pagination offset (default: `0`) | ### Use cases * Build a music browser or audio picker into your app or CMS * Search music by genre and mood for video or podcast background music * Programmatically download royalty-free music during content production pipelines * Create curated playlists filtered by popularity, tempo, or release date ### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to the music? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). #### What audio format is the music available in? Music is delivered as audio files via CDN-hosted download URLs. #### Can I filter for only free music? Yes. Set `include-premium` to `false` (the default) to return only free music. # Search and filter music Source: https://docs.magnific.com/api-reference/music/search-music get /v1/music Search the Magnific Music catalog. Filter by genre, mood, artist, premium status, and creation date range. Returns paginated results sorted by popularity by default. Use the Magnific Music API to search and filter music by genre, mood, artist, and more. See the [Music API overview](/api-reference/music/overview) for authentication and usage guidance. # Get loras Source: https://docs.magnific.com/api-reference/mystic/get-loras get /v1/ai/loras # Mystic - List tasks Source: https://docs.magnific.com/api-reference/mystic/get-mystic get /v1/ai/mystic Get the status of all Mystic tasks # Mystic - Get task status Source: https://docs.magnific.com/api-reference/mystic/get-mystic-task get /v1/ai/mystic/{task-id} Get the status of the Mystic task # Magnific Mystic API Source: https://docs.magnific.com/api-reference/mystic/mystic Photorealistic AI image generator with LoRA styles and characters. Generate production-ready 1K, 2K, and 4K images for marketing, e-commerce, and creative projects.
Generate photorealistic images in 1K/2K/4K with LoRA styles and characters.
Mystic AI image generation is a cutting-edge tool that produces stunningly realistic, high-resolution images. With options for 1K, 2K and 4K resolution, Mystic delivers sharp, detailed visuals that rival professional photography. From intricate facial features to complex textures, every element is rendered with exceptional clarity and precision. ### Stunning 1K, 2K and 4K quality Enjoy razor-sharp 1K, 2K and 4K resolution with Magnific Mystic. Generate high-definition images with zero effort—no extra steps or upscaling needed. Just describe your vision, and Mystic will create beautifully detailed visuals ready for any project, from web design to large-format printing. ### Effortless creation Transform your text descriptions into impressive, realistic results with custom parameters. Whether you need faces, hands, or text, everything is pixel-perfect and photo-realistic. Our state-of-the-art AI technology ensures that every generated image meets professional standards without requiring additional editing. ### Versatile applications Perfect for a wide range of uses, including: * Marketing materials that demand visual impact * E-commerce product showcases with lifelike detail * High-quality illustrations for publications * Concept art for film and game development * And much more! With Mystic AI, you're equipped to create visuals that captivate and convince, all at the click of a button.
Generate an image using Mystic Get the status of a Mystic image generation task Get a list of all Mystic image generation tasks, including their statuses and details. Returns a curated list of available LoRAs (Low-Rank Adaptation) for Mystic image generation.
### Image Reference Best Practices When using `structure_reference` or `style_reference` images, how you send them matters: | Action | Quality Impact | | ------------------------------------------ | -------------------------- | | Send URL of original image | ✅ Maximum quality | | Send base64 of original file read directly | ✅ Maximum quality | | Use `canvas.toDataURL('image/jpeg')` | ❌ \~8% quality loss | | Use `canvas.toDataURL('image/jpeg', 0.8)` | ❌ \~20% quality loss | | Resize image before sending | ❌ Significant quality loss | | Convert PNG → JPEG before sending | ❌ Quality loss | **Recommended:** Send reference images via URL whenever possible. Higher quality reference images produce better style and structure transfers. ### Frequently Asked Questions Magnific Mystic generates images in three resolution tiers: 1K (1024px), 2K (2048px), and 4K (4096px). Higher resolutions deliver more detail and are ideal for print or large-format displays. LoRAs (Low-Rank Adaptations) are style and character presets that customize Mystic's output. Use the GET /v1/ai/loras endpoint to retrieve available LoRAs, then include the LoRA ID in your generation request to apply specific styles or characters. Generation time depends on resolution: 1K images typically complete in 10-20 seconds, 2K in 20-40 seconds, and 4K in 40-90 seconds. Use webhooks for efficient notification of task completion. Mystic outputs high-quality PNG images. The generated image is delivered via a URL that remains accessible for download after task completion. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. # LoRAs training for custom characters Source: https://docs.magnific.com/api-reference/mystic/post-loras-characters post /v1/ai/loras/characters Create you own custom character using LoRAs training For now you can check the status of the training calling `v1/ai/loras`. We are working on it # Mystic LoRA Styles - Train custom style Source: https://docs.magnific.com/api-reference/mystic/post-loras-styles post /v1/ai/loras/styles Create you own custom style using LoRAs training For now you can check the status of the training calling `v1/ai/loras`. We are working on it # Mystic - Create image from text Source: https://docs.magnific.com/api-reference/mystic/post-mystic post /v1/ai/mystic Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. ## Important Mystic image generation mode is Magnific's exclusive advanced AI workflow for ultra-realistic, high-resolution images. Make sure you get your webhook set up on every request in order to retrieve your generations. # Happy Horse 1.1 - Create video from reference images Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1-1/generate post /v1/ai/reference-to-video/happy-horse-1-1 Generate a high-quality AI video guided by 1 to 9 reference images that define characters, objects or scenes using Happy Horse 1.1, Alibaba's reference-to-video model. **How to use references:** - Provide 1 to 9 reference images via `image_urls` - Reference each image in the prompt by its 1-based positional index (`character1`, `character2`, ...) - The order of `image_urls` matches the character index in the prompt **Key features:** - 720P and 1080P resolution support - 9 aspect ratio options - Duration range: 3-15 seconds - Optional bottom-right "Happy Horse" watermark - Prompt up to 5000 characters `image_urls` is an array of **objects**, each with a `url` field — not an array of plain strings. Passing strings returns `400 Bad Request`. **Example request body** ```json theme={null} { "prompt": "character1 and character2 walk together through a sunlit forest path", "image_urls": [ { "url": "https://example.com/character1.jpg" }, { "url": "https://example.com/character2.jpg" } ], "aspect_ratio": "16:9", "resolution": "1080P", "duration": 5 } ``` The array position sets the character index used in the prompt: the 1st item is `character1`, the 2nd is `character2`, and so on up to `character9`. # Happy Horse 1.1 R2V - List tasks Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1-1/happy-horse-1-1-r2v-tasks get /v1/ai/reference-to-video/happy-horse-1-1 Retrieve the list of all Happy Horse 1.1 reference-to-video tasks for the authenticated user. # Happy Horse 1.1 Reference To Video API Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1-1/overview Generate AI videos from up to 9 reference images with Happy Horse 1.1. Reference characters as character1..character9 in the prompt. 720P/1080P, 9 aspect ratios. Happy Horse 1.1 Reference-to-Video generates videos from up to 9 reference images, preserving character identity across scenes. Reference each image in the prompt with `character1`..`character9` tokens. Happy Horse 1.1 Reference-to-Video is an AI video generation API that creates MP4 videos guided by 1 to 9 reference images defining characters, objects, or scenes. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), the model maintains visual identity across the generated video. You reference each image in the prompt by its 1-based positional index (`character1`, `character2`, ..., up to `character9`). Output is available at 720P (1280x720) or 1080P (1920x1080) with durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: Latest iteration of the Alibaba ATH Happy Horse video family * **Up to 9 reference images**: Rich multi-character scenes from a single request * **Indexed prompt references**: Use `character1`..`character9` tokens in the prompt to place each reference * **9 aspect ratios**: `16:9`, `9:16`, `1:1`, `4:3`, `3:4`, `21:9`, `9:21`, `5:4`, `4:5` * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Optional watermark**: Bottom-right "Happy Horse" text watermark, off by default * **Long prompts**: Up to 5000 characters describing scene, motion, and character interactions * **Async processing**: Webhook notifications or polling for task completion ### How character references work 1. Provide 1 to 9 character images in the `image_urls` array (JPEG/JPG/PNG/WEBP, shortest side 400 px+ recommended, max 10 MB each) 2. The 1st item in the array is `character1`, the 2nd is `character2`, ..., up to `character9` 3. In the `prompt`, reference characters by these tokens to place them in the scene 4. Combine multiple references for multi-character scenes (e.g. "character1 and character2 share a meal") **Example prompt:** > "character1 and character2 walk together through a sunlit forest path, golden hour lighting" ### Use cases * **Multi-character narratives**: Create scenes with up to 9 distinct characters interacting * **Branded content**: Maintain consistent mascot, spokesperson, or brand identity across video campaigns * **Episodic content**: Reuse the same characters across multiple videos for series or storytelling * **Game cinematics**: Generate cutscenes featuring specific character designs * **Marketing campaigns**: Place product imagery (`character1` = product) into varied scene contexts * **Storyboarding**: Pre-visualize scenes with reference characters before live production ### API operations Generate videos by submitting reference images and a prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new reference-to-video generation task List all Happy Horse 1.1 R2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | -------------- | --------------- | -------- | --------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Scene description referencing characters as `character1`..`character9`. Max 5000 characters | | `image_urls` | `array` | Yes | - | Ordered array of 1 to 9 reference image objects, each `{ "url": "https://..." }` (JPEG/JPG/PNG/WEBP, max 10 MB each). Position determines character index. Objects, not plain strings | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"`, `"21:9"`, `"9:21"`, `"5:4"`, `"4:5"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3 to 15 | | `watermark` | `boolean` | No | `false` | Add a bottom-right "Happy Horse" text watermark | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | Note that `image_urls` takes an array of **objects** (`{ "url": "https://..." }`), not plain strings. Example request body: ```json theme={null} { "prompt": "character1 and character2 walk together through a sunlit forest path", "image_urls": [ { "url": "https://example.com/character1.jpg" }, { "url": "https://example.com/character2.jpg" } ], "aspect_ratio": "16:9", "resolution": "1080P", "duration": 5 } ``` When the task completes, the generated MP4 video URL is returned in `data.generated`. The URL is valid for 24 hours — download and persist it to permanent storage immediately. ## Frequently Asked Questions Happy Horse 1.1 Reference-to-Video accepts 1 to 9 reference images via the `image_urls` array. The first image is referenced as `character1`, the second as `character2`, and so on up to `character9`. Use position-based tokens that match the order of the `image_urls` array. The 1st image is `character1`, the 2nd is `character2`, etc. Example: "character1 and character2 walk together through a sunlit forest path while character3 watches from a distance." Reference images must be JPEG, JPG, PNG, or WEBP via publicly accessible URLs. Recommended shortest side is at least 400 pixels (720P or higher recommended). Maximum file size is 10 MB per image. The prompt can be up to 5000 characters. The upstream provider treats 5000 non-Chinese characters or 2500 Chinese characters as the effective ceiling. Prompts longer than 5000 characters are rejected by validation. When `watermark` is `true`, a fixed "Happy Horse" text watermark is rendered in the bottom-right corner of the output video. Default is `false` (no watermark). Happy Horse 1.1 R2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Reference images**: Use clear, well-lit images with the character or object prominently visible. Avoid busy backgrounds. * **Reference order**: The array order is the character index — list characters in the order you plan to reference them in the prompt. * **Prompt structure**: Explicitly name each character by token ("character1 walks left while character2 sits down") for predictable placement. * **Image consistency**: Use images at similar quality and lighting for smoother integration in multi-character scenes. * **Duration selection**: Start with shorter durations (3-5 seconds) for iteration, then increase for final outputs. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.1 Text-to-Video](/api-reference/text-to-video/happy-horse-1-1/overview)**: Generate videos from text prompts without character references * **[Happy Horse 1.1 Image-to-Video](/api-reference/image-to-video/happy-horse-1-1/overview)**: Animate a single image into video * **[Happy Horse 1.0 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1/overview)**: Previous Happy Horse iteration with up to 9 character references # Happy Horse 1.1 R2V - Get task status Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1-1/task-by-id get /v1/ai/reference-to-video/happy-horse-1-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.1 reference-to-video task by its ID. # Happy Horse 1.0 - Create video from reference images Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1/generate post /v1/ai/reference-to-video/happy-horse-1 Generate a high-quality AI video guided by 1 to 9 reference images that define characters, objects or scenes using Happy Horse 1.0, Alibaba's reference-to-video model. **How to use references:** - Provide 1 to 9 reference images via `image_urls` - Reference each image in the prompt by its 1-based positional index (`character1`, `character2`, ...) - The order of `image_urls` matches the character index in the prompt **Key features:** - 720P and 1080P resolution support - 5 aspect ratio options - Duration range: 3-15 seconds - Optional bottom-right "Happy Horse" watermark - Prompt up to 5000 characters # Happy Horse 1.0 R2V - List tasks Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1/happy-horse-1-r2v-tasks get /v1/ai/reference-to-video/happy-horse-1 Retrieve the list of all Happy Horse 1.0 reference-to-video tasks for the authenticated user. # Happy Horse 1.0 Reference To Video API Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1/overview Generate AI videos from up to 9 reference images with Happy Horse 1.0. Reference characters as character1..character9 in the prompt. 720P/1080P, 3-15 second durations. Happy Horse 1.0 Reference-to-Video generates videos from up to 9 reference images, preserving character identity across scenes. Reference each image in the prompt with `character1`..`character9` tokens. Happy Horse 1.0 Reference-to-Video is an AI video generation API that creates MP4 videos guided by 1 to 9 reference images defining characters, objects, or scenes. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), the model maintains visual identity across the generated video. You reference each image in the prompt by its 1-based positional index (`character1`, `character2`, ..., up to `character9`). Output is available at 720P (1280x720) or 1080P (1920x1080) with durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: From the same family that ranks #1 on the Artificial Analysis Video Arena (April 2026) * **Up to 9 reference images**: More references than competing models for richer multi-character scenes * **Indexed prompt references**: Use `character1`..`character9` tokens in the prompt to place each reference * **5 aspect ratios**: `16:9`, `9:16`, `1:1`, `4:3`, `3:4` * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Optional watermark**: Bottom-right "Happy Horse" text watermark, off by default * **Long prompts**: Up to 5000 characters describing scene, motion, and character interactions * **Async processing**: Webhook notifications or polling for task completion ### How character references work 1. Provide 1 to 9 character images in the `image_urls` array (JPEG/JPG/PNG/WEBP, shortest side 400 px+ recommended, max 10 MB each) 2. The 1st item in the array is `character1`, the 2nd is `character2`, ..., up to `character9` 3. In the `prompt`, reference characters by these tokens to place them in the scene 4. Combine multiple references for multi-character scenes (e.g. "character1 and character2 share a meal") **Example prompt:** > "character1 and character2 walk together through a sunlit forest path, golden hour lighting" ### Use cases * **Multi-character narratives**: Create scenes with up to 9 distinct characters interacting * **Branded content**: Maintain consistent mascot, spokesperson, or brand identity across video campaigns * **Episodic content**: Reuse the same characters across multiple videos for series or storytelling * **Game cinematics**: Generate cutscenes featuring specific character designs * **Marketing campaigns**: Place product imagery (`character1` = product) into varied scene contexts * **Storyboarding**: Pre-visualize scenes with reference characters before live production ### API operations Generate videos by submitting reference images and a prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new reference-to-video generation task List all Happy Horse 1.0 R2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | -------------- | --------- | -------- | --------- | ----------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Scene description referencing characters as `character1`..`character9`. Max 5000 characters | | `image_urls` | `array` | Yes | - | Ordered array of 1 to 9 reference images (JPEG/JPG/PNG/WEBP, max 10 MB each). Position determines character index | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3 to 15 | | `watermark` | `boolean` | No | `false` | Add a bottom-right "Happy Horse" text watermark | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Happy Horse 1.0 Reference-to-Video accepts 1 to 9 reference images via the `image_urls` array. The first image is referenced as `character1`, the second as `character2`, and so on up to `character9`. Use position-based tokens that match the order of the `image_urls` array. The 1st image is `character1`, the 2nd is `character2`, etc. Example: "character1 and character2 walk together through a sunlit forest path while character3 watches from a distance." Reference images must be JPEG, JPG, PNG, or WEBP via publicly accessible URLs. Recommended shortest side is at least 400 pixels (720P or higher recommended). Maximum file size is 10 MB per image. The prompt can be up to 5000 characters. The upstream provider treats 5000 non-Chinese characters or 2500 Chinese characters as the effective ceiling. Prompts longer than 5000 characters are rejected by validation. When `watermark` is `true`, a fixed "Happy Horse" text watermark is rendered in the bottom-right corner of the output video. Default is `false` (no watermark). Happy Horse 1.0 R2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Reference images**: Use clear, well-lit images with the character or object prominently visible. Avoid busy backgrounds. * **Reference order**: The array order is the character index — list characters in the order you plan to reference them in the prompt. * **Prompt structure**: Explicitly name each character by token ("character1 walks left while character2 sits down") for predictable placement. * **Image consistency**: Use images at similar quality and lighting for smoother integration in multi-character scenes. * **Duration selection**: Start with shorter durations (3-5 seconds) for iteration, then increase for final outputs. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.0 Text-to-Video](/api-reference/text-to-video/happy-horse-1/overview)**: Generate videos from text prompts without character references * **[Happy Horse 1.0 Image-to-Video](/api-reference/image-to-video/happy-horse-1/overview)**: Animate a single image into video * **[Happy Horse 1.0 Video Edit](/api-reference/video-edit/happy-horse-1/overview)**: Edit existing videos with natural-language instructions * **[WAN 2.7 Reference-to-Video](/api-reference/reference-to-video/wan-2-7/overview)**: Alibaba WAN R2V with up to 5 references and voice-guided generation # Happy Horse 1.0 R2V - Get task status Source: https://docs.magnific.com/api-reference/reference-to-video/happy-horse-1/task-by-id get /v1/ai/reference-to-video/happy-horse-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.0 reference-to-video task by its ID. # Create video with reference images - Veo 3.1 Source: https://docs.magnific.com/api-reference/reference-to-video/veo-3-1/generate post /v1/ai/reference-to-video/veo-3-1 Generate a video with character or object consistency using reference images. Maintains visual identity across scenes for storytelling and multi-scene projects. Supports 720p, 1080p, and 4K resolutions with native audio generation including dialogue and sound effects. Fixed 8-second duration at 24 FPS. # Veo 3.1 Reference-to-Video API | Magnific API Source: https://docs.magnific.com/api-reference/reference-to-video/veo-3-1/overview Generate videos with consistent characters and objects using reference images. Maintain visual identity across scenes for storytelling and multi-scene projects with Google Veo 3.1. Generate videos with consistent characters and objects using reference images. Perfect for storytelling and multi-scene projects. Reference-to-Video is a specialized video generation mode that maintains **visual identity consistency** across generated content. By providing 1-3 reference images, you can create videos where characters, objects, or subjects maintain their appearance throughout the scene. This is ideal for creating coherent narratives, character-based content, and multi-scene projects where visual consistency is critical. ### Key capabilities * **Character consistency**: Maintain visual identity of characters across video generation * **Multi-reference support**: Use 1-3 reference images for subject consistency * **Multi-resolution output**: Generate videos in 720p, 1080p, or 4K resolution * **Native audio generation**: Includes dialogue and sound effects synthesis * **Fixed 8-second duration**: Optimized duration at 24 FPS for cinematic quality * **Aspect ratio control**: 16:9 (landscape) or 9:16 (portrait) formats * **Negative prompts**: Specify elements to avoid in generation * **Long prompts**: Up to 20,000 characters for detailed scene descriptions ### Use cases * **Storytelling**: Create multi-scene narratives with consistent characters * **Brand mascots**: Generate videos featuring consistent brand characters * **Product showcases**: Maintain product appearance across different scenes * **Character animation**: Bring illustrated or photographed characters to life consistently * **Social media series**: Create episodic content with recurring characters * **Advertising campaigns**: Produce multiple ads with consistent spokesperson ### How it differs from Image-to-Video | Feature | Image-to-Video | Reference-to-Video | | -------- | ---------------------------- | -------------------------------------------- | | Input | Single image to animate | 1-3 reference images + prompt | | Purpose | Animate a specific image | Generate new scenes with consistent subjects | | Output | Animation of the input image | New video featuring reference subjects | | Duration | 4, 6, or 8 seconds | Fixed 8 seconds | | Modes | Standard and Fast | Single mode | ### Generate with Reference-to-Video Create videos with consistent characters and objects using reference images.
Create a reference-to-video task List all reference-to-video tasks Get task status by ID
### Parameters | Parameter | Type | Required | Description | | ----------------- | --------- | -------- | -------------------------------------------------------------------------- | | `image_urls` | `array` | Yes | Array of 1-3 reference image URLs (HTTPS, publicly accessible) | | `prompt` | `string` | Yes | Text describing the video scene with reference subjects (max 20,000 chars) | | `negative_prompt` | `string` | No | Text describing what to avoid in the video | | `resolution` | `string` | No | Output resolution: `"720p"`, `"1080p"`, or `"4k"` (default: `"720p"`) | | `aspect_ratio` | `string` | No | Video format: `"16:9"` or `"9:16"` (default: `"16:9"`) | | `generate_audio` | `boolean` | No | Generate audio with dialogue and effects (default: `true`) | | `seed` | `integer` | No | Random seed for reproducibility | | `webhook_url` | `string` | No | URL for task completion notification | ### Example request ```bash theme={null} curl -X POST "https://api.magnific.com/v1/ai/reference-to-video/veo-3-1" \ -H "x-magnific-api-key: YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "image_urls": [ "https://example.com/character-front.jpg", "https://example.com/character-side.jpg" ], "prompt": "The character walks through a futuristic city at night, neon lights reflecting on wet streets", "negative_prompt": "blurry, low quality, distorted", "resolution": "1080p", "aspect_ratio": "16:9", "generate_audio": true }' ``` ## Frequently Asked Questions **Reference-to-Video** uses reference images to maintain visual consistency of subjects (characters, objects) while generating entirely new video scenes. **Image-to-Video** animates a single input image directly. Use Reference-to-Video when you need to create multiple scenes with the same character or object looking consistent. You can provide 1-3 reference images. Using multiple images from different angles improves consistency. For characters, include front-facing and profile views. For objects, include various angles to help the model understand the complete appearance. Good reference images are: * High resolution and well-lit * Show the subject clearly without obstructions * Include different angles when using multiple images * Have consistent appearance of the subject across images * Use HTTPS URLs that are publicly accessible The 8-second duration at 24 FPS is optimized for reference-to-video generation, providing enough time for meaningful scenes while ensuring high-quality consistency of the reference subjects throughout the video. Currently, Reference-to-Video is available in a single mode optimized for quality and consistency. Unlike Text-to-Video and Image-to-Video, there is no Fast variant for Reference-to-Video. When `generate_audio` is enabled (default), the model generates synchronized audio including dialogue and sound effects appropriate to the scene. If your reference subject is a person and the prompt describes them speaking, the audio will include synthesized dialogue. ## Best practices * **Multiple reference angles**: Provide 2-3 images showing different angles of your subject for best consistency * **Clear subjects**: Use reference images where the subject is clearly visible and unobstructed * **Consistent lighting**: Reference images with similar lighting produce more coherent results * **Descriptive prompts**: Describe how the reference subject should act in the scene * **Scene context**: Include environment and action details in your prompt * **Negative prompts**: Use to avoid quality issues like "blurry, distorted, inconsistent features" * **Webhook integration**: Use webhooks for production workflows to handle async completion ## Related APIs * **[Veo 3.1 Text-to-Video](/api-reference/text-to-video/veo-3-1/overview)**: Generate videos from text prompts without reference images * **[Veo 3.1 Image-to-Video](/api-reference/image-to-video/veo-3-1/overview)**: Animate a single image into video * **[Kling 2.6 Motion Control](/api-reference/video/kling-v2-6-motion-control-pro)**: Transfer motion from reference videos * **[RunWay Act Two](/api-reference/video/runway-act-two)**: Character performance with reference video # Get Veo 3.1 Reference-to-Video task by ID Source: https://docs.magnific.com/api-reference/reference-to-video/veo-3-1/task-by-id get /v1/ai/reference-to-video/veo-3-1/{task-id} Retrieve the status and results of a specific reference-to-video generation task. # Get all Veo 3.1 Reference-to-Video tasks Source: https://docs.magnific.com/api-reference/reference-to-video/veo-3-1/veo-3-1-tasks get /v1/ai/reference-to-video/veo-3-1 Retrieve a list of all reference-to-video generation tasks for Veo 3.1. # WAN 2.7 - Create video from reference characters Source: https://docs.magnific.com/api-reference/reference-to-video/wan-2-7/generate post /v1/ai/reference-to-video/wan-2-7 Generate a video featuring characters from reference images or videos using WAN 2.7. Maintains visual identity of referenced characters across the generated video. **How to use references:** - Provide character images via `image_urls` and/or character videos via `video_urls` - Combined total of image and video references must not exceed 5 - Reference characters in the prompt as "Image 1", "Image 2", "Video 1", etc. - Optionally include `reference_voice` per character for voice-guided generation **Key features:** - 720P and 1080P resolution support - 5 aspect ratio options (or use `start_image_url` to set aspect ratio from an image) - Duration range: 2-10 seconds # WAN 2.7 Reference To Video API Source: https://docs.magnific.com/api-reference/reference-to-video/wan-2-7/overview Generate AI videos featuring characters from reference images or videos with WAN 2.7. Up to 5 character references, voice-guided generation, 720P and 1080P resolution, 2-10 second durations. WAN 2.7 Reference-to-Video generates videos featuring specific characters from reference images or videos, maintaining visual identity and supporting optional voice references. WAN 2.7 Reference-to-Video is an AI video generation API that creates MP4 videos featuring characters from reference images or videos. You provide up to 5 character references (images and/or videos combined), then describe a scene in the prompt using labels like "Image 1" or "Video 1" to place those characters. The model maintains visual identity of referenced characters across the generated video. Output is available at 720P (1280x720) or 1080P (1920x1080) resolution with durations from 2 to 10 seconds. ### Key capabilities * **Character references**: Provide up to 5 combined character images and videos for identity preservation * **Prompt-based character placement**: Reference characters as "Image 1", "Image 2", "Video 1" in the prompt * **Voice-guided generation**: Optionally include `reference_voice` audio per character for voice-guided output * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **5 aspect ratios**: `16:9`, `9:16`, `1:1`, `4:3`, `3:4` (or auto-detect from `start_image_url`) * **Flexible durations**: 2 to 10 seconds of video output * **Start frame control**: Optionally provide `start_image_url` to set the first frame and auto-detect aspect ratio * **Async processing**: Webhook notifications or polling for task completion ### How character references work 1. Provide character images via `image_urls` (JPEG/PNG/BMP/WEBP, 240-8000px, max 20MB each) 2. Provide character videos via `video_urls` (MP4/MOV, max 100MB each) 3. Combined total of images + videos must not exceed 5 4. Reference characters in the prompt using position labels: "Image 1", "Image 2", "Video 1", "Video 2" 5. Optionally add `reference_voice` audio URL per character for voice-guided generation **Example prompt:** > "Image 1 and Image 2 are walking together in a park while Video 1 plays guitar in the background." ### Use cases * **Consistent character videos**: Generate multiple videos with the same character across different scenes * **Multi-character narratives**: Create scenes with up to 5 characters interacting * **Branded content**: Maintain consistent mascot or spokesperson identity across video campaigns * **Voice-synchronized video**: Guide character motion using voice references for natural lip and gesture sync * **Social media series**: Create episodic content with recurring characters * **Virtual presenters**: Generate videos of a reference person in different settings ### API operations Generate videos by submitting character references and a prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new reference-to-video generation task List all WAN 2.7 R2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------------------- | --------- | ----------- | --------- | ------------------------------------------------------------------------------------------ | | `prompt` | `string` | Yes | - | Scene description referencing characters as "Image 1", "Video 1", etc. Max 5000 characters | | `negative_prompt` | `string` | No | - | Elements to avoid (e.g., "blurry, watermark"). Max 500 characters | | `image_urls` | `array` | Conditional | - | Character reference images. Each item has `url` (required) and optional `reference_voice` | | `video_urls` | `array` | Conditional | - | Character reference videos. Each item has `url` (required) and optional `reference_voice` | | `start_image_url` | `string` | No | - | First-frame image. If provided, overrides `aspect_ratio` with image dimensions | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 2 to 10 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `additional_settings.prompt_extend` | `boolean` | No | `true` | Enable AI prompt expansion for richer output | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions WAN 2.7 Reference-to-Video is an AI video generation API developed by Alibaba. You provide character reference images or videos along with a text prompt that describes a scene. The model generates a video featuring those characters while preserving their visual identity. You receive a task ID immediately, then poll for results or receive a webhook notification. You can provide up to 5 combined character references (images + videos). For example: 3 character images and 2 character videos, or 5 images and 0 videos. At least one image or video reference is required. Use position labels based on the order you provide references. Image references are labeled "Image 1", "Image 2", etc. Video references are labeled "Video 1", "Video 2", etc. Example: "Image 1 and Video 1 are having a conversation at a cafe." Each character reference (image or video) can include an optional `reference_voice` URL pointing to an audio file. The model uses this voice to guide character motion and lip movement in the generated video, creating more natural character animation. Image references: JPEG, PNG, BMP, WEBP (240-8000px per side, max 20MB). Video references: MP4, MOV (max 100MB). All files must be at publicly accessible URLs. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Character images**: Use clear, well-lit images with the character prominently visible. Avoid busy backgrounds. * **Character videos**: Shorter reference videos with clear character visibility produce better identity preservation. * **Prompt structure**: Explicitly name each character by label ("Image 1 walks left while Image 2 sits down") for predictable placement. * **Voice references**: Provide clean audio clips with minimal background noise for best voice-guided results. * **Duration selection**: Reference-to-Video supports 2-10 seconds. Start with shorter durations for iteration. * **Negative prompts**: Include: "blurry, low quality, watermark, text, distortion, extra limbs" * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[WAN 2.7 Text-to-Video](/api-reference/text-to-video/wan-2-7/overview)**: Generate videos from text prompts without character references * **[WAN 2.7 Image-to-Video](/api-reference/image-to-video/wan-2-7/overview)**: Animate images or extend existing videos * **[Kling 3 Omni](/api-reference/video/kling-v3-omni/overview)**: Alternative video generation model with reference video support # WAN 2.7 R2V - Get task status Source: https://docs.magnific.com/api-reference/reference-to-video/wan-2-7/task-by-id get /v1/ai/reference-to-video/wan-2-7/{task-id} Retrieve the status and result of a specific WAN 2.7 reference-to-video task by its ID. # WAN 2.7 R2V - List tasks Source: https://docs.magnific.com/api-reference/reference-to-video/wan-2-7/wan-2-7-r2v-tasks get /v1/ai/reference-to-video/wan-2-7 Retrieve the list of all WAN 2.7 reference-to-video tasks for the authenticated user. # Remove Background - Image Cutout API | Magnific API Source: https://docs.magnific.com/api-reference/remove-background/overview Remove image backgrounds with AI. Get transparent PNG up to 25 megapixels. Supports JPG/PNG input up to 20 MB. Instant synchronous results. This API is currently in **beta**. Features and endpoints may change as we continue to improve the service. Automatically remove backgrounds from images using AI-powered segmentation. Get clean cutouts with transparent backgrounds in seconds. Remove Background is an AI-powered background removal API that instantly separates subjects from their backgrounds. Upload any JPG or PNG image via URL and receive a transparent PNG with the background removed. The API supports images up to 20 MB and outputs high-resolution results up to 25 megapixels. This is a synchronous API that returns results immediately without polling. ### Key capabilities * **Instant processing**: Synchronous API returns results immediately without task polling * **High resolution output**: Full resolution up to 25 megapixels, preview up to 0.25 megapixels * **Transparent PNG output**: Clean cutouts with alpha channel transparency * **Multiple output URLs**: Get original, high-resolution, and preview versions in one response * **Large file support**: Accepts images up to 20 MB in size * **Format support**: Works with JPG and PNG input images * **URL-based input**: Submit images via public URL for easy integration * **Temporary URLs**: Response URLs valid for 5 minutes for secure access ### Use cases * **E-commerce product photos**: Remove backgrounds for consistent product catalogs and marketplace listings * **Graphic design workflows**: Create transparent assets for compositions and marketing materials * **Social media content**: Isolate subjects for custom backgrounds and branded visuals * **Photo editing apps**: Integrate background removal into mobile and web editing tools * **Print-on-demand services**: Prepare artwork for merchandise and custom products * **Real estate photography**: Isolate property elements for virtual staging ### Remove background from images Submit an image URL to remove its background. The API returns multiple URLs for different resolution outputs.
Remove background from an image
### Parameters | Parameter | Type | Required | Default | Description | | ----------- | -------- | -------- | ------- | ------------------------------------------------------------------------------- | | `image_url` | `string` | Yes | - | URL of the image whose background needs to be removed (JPG or PNG, up to 20 MB) | ### Response | Field | Type | Description | | ----------------- | -------- | ---------------------------------------------------------------------- | | `original` | `string` | URL of the original input image | | `high_resolution` | `string` | URL of the high-resolution image with background removed (up to 25 MP) | | `preview` | `string` | URL of the preview version (up to 0.25 MP) | | `url` | `string` | Direct download URL for the high-resolution result | Response URLs are temporary and expire after **5 minutes**. Download or process the images promptly after receiving the response. ## Frequently Asked Questions The Remove Background API is an AI-powered service that automatically detects and removes backgrounds from images. You submit an image URL, and the API uses advanced segmentation models to separate the subject from the background, returning a transparent PNG. Unlike async APIs, this endpoint processes requests synchronously and returns results immediately. The API accepts JPG and PNG images up to 20 MB in file size. Output is always a PNG with transparency. The high-resolution output supports images up to 25 megapixels, while the preview output is limited to 0.25 megapixels. All URLs in the response (original, high\_resolution, preview, url) are temporary and valid for **5 minutes** only. Make sure to download or process the images immediately after receiving the response. The Remove Background API is **synchronous**. Unlike other AI endpoints that return task IDs for polling, this API processes your request and returns the result immediately in the response. No task polling or webhooks are needed. Rate limits depend on your subscription tier. Visit the [Rate Limits](/ratelimits) page for details on request limits and quotas. Pricing varies by subscription plan. Check the [Pricing](/pricing) page for current rates and available plans. The beta designation indicates the API is production-ready but may receive updates to improve accuracy, performance, or add new features. We recommend monitoring the changelog for any changes that might affect your integration. ## Best practices * **Image quality**: Use high-resolution input images for best cutout quality * **Clear subjects**: Images with well-defined subjects produce cleaner results * **Contrast**: Higher contrast between subject and background improves accuracy * **Download promptly**: Save results immediately as URLs expire in 5 minutes * **Error handling**: Implement retry logic with exponential backoff for 5xx errors * **URL validation**: Ensure image URLs are publicly accessible before submission ## Related APIs * **[Image Upscaler](/api-reference/image-upscaler-creative/image-upscaler)**: Enhance image resolution after background removal * **[Reimagine](/api-reference/text-to-image/reimagine-flux/post-reimagine-flux)**: Transform images with AI-powered style changes * **[Image Style Transfer](/api-reference/image-style-transfer/image-styletransfer)**: Apply artistic styles to your cutout images # Remove the background of an image Source: https://docs.magnific.com/api-reference/remove-background/post-beta-remove-background post /v1/ai/beta/remove-background This endpoint removes the background from an image provided via a URL. The URLs in the response are temporary and valid for **5 minutes** only. **Supported formats:** JPG, PNG **File size limit:** up to 20 MB **Output resolutions:** Preview (up to 0.25 megapixels), Full resolution (up to 25 megapixels) # Download an resource Source: https://docs.magnific.com/api-reference/resources/download-a-resource get /v1/resources/{resource-id}/download Download a specific resource by providing the resource ID. This endpoint supports downloading various types of resources including vectors, PSDs, photos, and AI-generated content. # Get available download formats for resource Source: https://docs.magnific.com/api-reference/resources/download-resource-by-id-and-format get /v1/resources/{resource-id}/download/{resource-format} Retrieve a resource by specifying both the resource ID and the format. This endpoint allows for precise downloading of resources in the desired format. # Search and filter resources with advanced options Source: https://docs.magnific.com/api-reference/resources/get-all-resources get /v1/resources Retrieve a list of resources based on various filter criteria such as orientation, content type, license, and more. # Get detailed resource information by ID Source: https://docs.magnific.com/api-reference/resources/get-the-detail-of-a-resource-psd-vector-or-photo get /v1/resources/{resource-id} Retrieve the detailed information of a specific resource by its ID. This endpoint supports multiple resource types including PSD, vector, photo, and AI-generated content. # Images and templates API Source: https://docs.magnific.com/api-reference/resources/images-and-templates-api Search, retrieve, and download images and templates programmatically with the Magnific API. The Images and templates API lets you discover resources (photos, vectors, PSDs) with AI-powered search and download assets for use in your applications. ### What you can do * List resources with sorting and filters * Get a single resource by ID * Download a resource and specify format ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication). ### Endpoints
List resources with sorting and filters Retrieve a single resource by ID Download a resource Download a resource by ID and format
### Use cases * Build an asset picker for your CMS or design tool * AI-powered search by keyword and sort by popularity * Programmatically download assets during build/deploy ### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to resources? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). # Stock content API Source: https://docs.magnific.com/api-reference/resources/stock-content Search, retrieve, and download Magnific stock content — images, templates, icons, videos, and music — via the Magnific API. Access Magnific's catalog of images, templates, icons, videos, and music to power your apps and workflows. ### What you can do * AI-powered search and list stock content * Retrieve details by ID * Download assets in supported formats * Filter by content type and order ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication).
Photos, vectors, PSDs — list, detail, and download Browse and download icons Discover and download videos Search and download music
### Popular endpoints
List stock resources List icons List videos Search music
### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to the content? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). # Download a sound effect Source: https://docs.magnific.com/api-reference/sfx/download-sfx get /v1/sound-effects/{sfx-id}/download Get a direct download URL for a sound effect audio file. Returns the sound effect ID, title, and a CDN-hosted download link. Get a direct download URL for a sound effect audio file. See the [Sound Effects API overview](/api-reference/sfx/overview) for authentication and usage guidance. # Get detailed sound effect information by ID Source: https://docs.magnific.com/api-reference/sfx/get-sfx-by-id get /v1/sound-effects/{sfx-id} Retrieve full details for a sound effect including category hierarchy, tags, duration, popularity, and download statistics. Retrieve full details for a sound effect including category hierarchy, tags, duration, popularity, and download statistics. See the [Sound Effects API overview](/api-reference/sfx/overview) for authentication and usage guidance. # Magnific Sound Effects API Source: https://docs.magnific.com/api-reference/sfx/overview Search, filter, and download royalty-free sound effects from the Magnific catalog. Browse 42 categories including ambience, foley, transitions, and more via the Magnific API. The Magnific Sound Effects API provides programmatic access to a catalog of royalty-free sound effects. Search by keyword, filter by category and premium status, retrieve detailed metadata, and download audio files for use in your projects. ### What you can do * **Full-text search**: Query sound effects by title, tags, and categories using the `q` parameter * **Category filtering**: Browse 42 categories organized in a parent-child hierarchy (e.g., `ambience` > `forest`, `foley` > `footsteps`, `transitions` > `whooshes`) * **Premium filtering**: Control whether results include premium content via `include-premium` (default: free only) * **Flexible sorting**: Sort by `relevance`, `popularity`, `download_count`, `created_at`, `duration`, or `title` with ascending/descending support * **Pagination**: Navigate results with `limit` (1-1000, default 10) and `offset` parameters * **Detailed metadata**: Access title, tags, category hierarchy, duration, popularity score, download count, premium status, and creation date * **Direct downloads**: Get CDN-hosted download URLs for sound effect audio files ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication). ### Endpoints
Search and filter sound effects Get detailed sound effect information by ID Download a sound effect audio file
### Parameters overview The search endpoint supports the following filters: | Parameter | Type | Description | | ----------------- | --------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `q` | `string` | Full-text search across titles, tags, and categories | | `category` | `string` | Filter by category slug (e.g., `"ambience"`, `"foley"`, `"transitions"`) | | `include-premium` | `boolean` | Include premium sound effects (default: `false`) | | `order_by` | `string` | Sort order: `"relevance"`, `"popularity"`, `"-popularity"`, `"download_count"`, `"-download_count"`, `"created_at"`, `"-created_at"`, `"duration"`, `"-duration"`, `"title"`, `"-title"`. Defaults to `"relevance"` when `q` is provided, `"-popularity"` otherwise | | `limit` | `integer` | Results per page: 1-1000 (default: `10`) | | `offset` | `integer` | Pagination offset (default: `0`) | ### Available categories Sound effects are organized into 42 categories with a parent-child hierarchy. Use a parent category slug to include all its subcategories. | Parent Category | Subcategories | | ------------------------ | ----------------------------------------------------------------------------------------------------------------------------------- | | **Ambience** | `forest`, `city-ambience`, `room-tones`, `rain-and-storm`, `beach-and-ocean`, `wind` | | **Foley** | `weapons-and-warfare`, `footsteps`, `explosions-and-crashes`, `kicks-and-punches`, `clothing`, `debris` | | **Human Sounds** | `genre`, `glitch-and-noise`, `cartoon-voices-and-sounds`, `sci-fi-sounds`, `ghosts-and-horror-transitions`, `gaming-and-fantasy` | | **Interface and Alerts** | `user-interface` | | **Musical** | `drum-and-percussion-loops`, `vocal-phrases`, `one-shots`, `melodic-loops` | | **Realistic** | `wildlife`, `machines-and-tools`, `household-objects`, `cars-and-airplanes`, `water`, `electronic-devices`, `pets-and-farm-animals` | | **Transitions** | `epic-transitions`, `cinematic-impacts`, `whooshes`, `risers`, `intros-and-outros` | ### Use cases * Build a sound effects browser or audio picker into your app or CMS * Search sound effects by category for video editing and post-production workflows * Programmatically download royalty-free SFX during content production pipelines * Create curated sound libraries filtered by popularity, duration, or category * Add ambient sounds and UI feedback to games and interactive applications ### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to the sound effects? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). #### What audio format are the sound effects available in? Sound effects are delivered as audio files via CDN-hosted download URLs. #### Can I filter for only free sound effects? Yes. Set `include-premium` to `false` (the default) to return only free sound effects. #### How does category filtering work? Use the `category` parameter with a category slug. Filtering by a parent category (e.g., `ambience`) includes all its subcategories (e.g., `forest`, `city-ambience`, `rain-and-storm`). Each sound effect belongs to one category, and the response includes the full category hierarchy with parent information. #### How does sorting work? Use the `order_by` parameter. Prefix a field name with `-` for descending order (e.g., `-popularity` for most popular first). When a search query (`q`) is provided, results default to `"relevance"`. Without a query, results default to `"-popularity"`. #### What metadata is returned for each sound effect? Each sound effect includes: `id`, `title`, `tags`, `category` (with parent), `duration` (in seconds), `is_premium`, `popularity` score, `download_count`, `file_url`, and `created_at` timestamp. ## Related APIs * **[Music API](/api-reference/music/overview)**: Search and download royalty-free music tracks * **[Music Generation](/api-reference/music-generation/overview)**: Generate original music from text descriptions with AI * **[ElevenLabs Sound Effects](/api-reference/sound-effects/overview)**: Generate custom sound effects from text with AI # Search and filter sound effects Source: https://docs.magnific.com/api-reference/sfx/search-sfx get /v1/sound-effects Search the Magnific Sound Effects catalog. Filter by category, duration range, premium status, and creation date. Returns paginated results sorted by popularity by default. Use the Magnific Sound Effects API to search and filter sound effects by keyword, category, premium status, and more. See the [Sound Effects API overview](/api-reference/sfx/overview) for authentication and usage guidance. # Skin Enhancer - List tasks Source: https://docs.magnific.com/api-reference/skin-enhancer/get-skin-enhancer get /v1/ai/skin-enhancer # Skin Enhancer - Get task status Source: https://docs.magnific.com/api-reference/skin-enhancer/get-{task-id}-by-id get /v1/ai/skin-enhancer/{task-id} # Skin Enhancer - AI Portrait Enhancement API | Magnific API Source: https://docs.magnific.com/api-reference/skin-enhancer/overview Enhance portrait skin quality with AI. Three modes: Creative for artistic effects, Faithful for natural preservation, Flexible for targeted optimization. Professional results for photographers. Powered by Magnific AI, Skin Enhancer uses advanced neural networks to naturally improve skin texture while preserving facial details and identity. Skin Enhancer is an AI-powered portrait enhancement API that automatically improves skin quality in photographs. It offers three distinct processing modes: Creative for stylized artistic enhancements, Faithful for natural-looking improvements that preserve original appearance, and Flexible for targeted optimization based on specific goals. The API accepts images via URL or base64 encoding and returns high-quality enhanced portraits. ### Key capabilities * **Three enhancement modes**: Creative (artistic), Faithful (natural), and Flexible (targeted) modes for different use cases * **Sharpen control**: Adjustable sharpening intensity from 0 to 100 for precise detail enhancement * **Smart grain**: Intelligent grain application (0-100) to maintain photographic texture and avoid artificial smoothing * **Skin detail preservation**: Faithful mode includes dedicated skin detail control (0-100) to balance smoothing with texture retention * **Optimization targets**: Flexible mode offers 5 presets: `enhance_skin`, `improve_lighting`, `enhance_everything`, `transform_to_real`, `no_make_up` * **Flexible input**: Accepts both base64-encoded images and publicly accessible HTTPS URLs * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Portrait photography**: Enhance skin in professional headshots while maintaining natural appearance * **E-commerce**: Improve model photos for fashion and beauty product listings * **Social media content**: Quick skin enhancement for influencer and brand content * **Wedding photography**: Batch process wedding portraits for consistent skin quality * **Beauty industry**: Showcase cosmetic results with enhanced before/after comparisons * **Corporate headshots**: Professional skin enhancement for business profiles and LinkedIn photos ### Choose your enhancement mode Select the mode that best fits your needs: | Mode | Best For | Key Feature | | ------------ | ------------------------------------ | -------------------------------------------------------- | | **Creative** | Artistic portraits, stylized content | More pronounced artistic enhancement | | **Faithful** | Natural photography, editorial work | Preserves original appearance with `skin_detail` control | | **Flexible** | Specific optimization goals | 5 optimization presets for targeted results | ### Enhance skin with Skin Enhancer Submit a portrait image to the appropriate endpoint based on your desired enhancement mode. The service returns a task ID for async polling or webhook notification.
Artistic skin enhancement with stylized effects Natural enhancement preserving original appearance Targeted enhancement with optimization presets List all skin enhancer tasks Get task status and results by ID
### Parameters #### Common parameters (all modes) | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ------- | --------------------------------------------------------- | | `image` | `string` | Yes | - | Input image as base64 or publicly accessible HTTPS URL | | `sharpen` | `integer` | No | `0` | Sharpening intensity (0-100) | | `smart_grain` | `integer` | No | `2` | Smart grain intensity (0-100) to maintain natural texture | | `webhook_url` | `string` | No | - | URL for task completion notification | #### Faithful mode additional parameter | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ------- | ------------------------------------- | | `skin_detail` | `integer` | No | `80` | Skin detail enhancement level (0-100) | #### Flexible mode additional parameter | Parameter | Type | Required | Default | Description | | --------------- | -------- | -------- | -------------- | ---------------------------------------------------------------------------------------------------------------- | | `optimized_for` | `string` | No | `enhance_skin` | Optimization target: `enhance_skin`, `improve_lighting`, `enhance_everything`, `transform_to_real`, `no_make_up` | ## Frequently Asked Questions Skin Enhancer is an AI-powered portrait enhancement API that uses Magnific AI technology to improve skin quality in photographs. You submit an image via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The output is a high-quality enhanced portrait. **Creative** mode applies more artistic and stylized enhancements for creative projects. **Faithful** mode preserves the original appearance while improving skin quality, ideal for natural photography. **Flexible** mode allows you to select specific optimization targets like lighting improvement or makeup removal. Skin Enhancer accepts images as base64-encoded strings or publicly accessible HTTPS URLs. Common formats like JPEG and PNG are supported. For best results, use high-resolution portrait images with clear facial visibility. Flexible mode offers 5 presets: `enhance_skin` (default) focuses on skin quality, `improve_lighting` optimizes facial lighting, `enhance_everything` applies comprehensive enhancement, `transform_to_real` makes images appear more photorealistic, and `no_make_up` reduces visible makeup while enhancing skin. Use `sharpen` (0-100) to control detail enhancement and `smart_grain` (0-100) to maintain natural texture. In Faithful mode, `skin_detail` (0-100) specifically controls how much original skin texture is preserved. Lower values produce smoother results; higher values retain more natural texture. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and information about your plan. Pricing varies by subscription tier and usage volume. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Image quality**: Use high-resolution images with good lighting for optimal results * **Mode selection**: Choose Faithful for natural portraits, Creative for artistic projects, Flexible for specific optimization goals * **Parameter tuning**: Start with default values and adjust `sharpen` and `smart_grain` based on output quality * **Skin detail balance**: In Faithful mode, use `skin_detail` between 60-90 for natural results; lower values for smoother skin * **Batch processing**: For multiple images, use webhooks instead of polling for efficient production workflows * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Image Upscaler Creative](/api-reference/image-upscaler-creative/image-upscaler)**: Upscale and enhance image resolution with AI * **[Image Upscaler Precision](/api-reference/image-upscaler-precision/image-upscaler)**: High-fidelity image upscaling * **[Image Relight](/api-reference/image-relight/image-relight)**: Change lighting conditions in photos * **[Image Style Transfer](/api-reference/image-style-transfer/image-styletransfer)**: Apply artistic styles to images # Skin Enhancer Creative - Enhance skin Source: https://docs.magnific.com/api-reference/skin-enhancer/post-creative post /v1/ai/skin-enhancer/creative Enhance skin in images using AI with the Creative mode. This mode provides more artistic and stylized enhancements. # Skin Enhancer Faithful - Enhance skin Source: https://docs.magnific.com/api-reference/skin-enhancer/post-faithful post /v1/ai/skin-enhancer/faithful Enhance skin in images using AI with the Faithful mode. This mode preserves the original appearance while improving skin quality. # Skin Enhancer Flexible - Enhance skin Source: https://docs.magnific.com/api-reference/skin-enhancer/post-flexible post /v1/ai/skin-enhancer/flexible Enhance skin in images using AI with the Flexible mode. This mode allows you to choose the optimization target for the enhancement. # Sound Effects - List tasks Source: https://docs.magnific.com/api-reference/sound-effects/get-sound-effects get /v1/ai/sound-effects Get the status of all sound-effects tasks # Sound Effects - Get task status Source: https://docs.magnific.com/api-reference/sound-effects/get-{task-id}-by-id get /v1/ai/sound-effects/{task-id} Get the status of one sound-effects task # ElevenLabs Sound Effects - Text-to-Audio API Source: https://docs.magnific.com/api-reference/sound-effects/overview Generate realistic sound effects from text with ElevenLabs Sound Effects. Create 0.5-22 second audio clips with looping support for videos, games, and multimedia. Powered by ElevenLabs technology, this API generates realistic sound effects from text descriptions. ElevenLabs Sound Effects is an AI-powered text-to-audio API that creates realistic sound effects from natural language descriptions. Describe any sound you need - from animal noises to mechanical sounds, nature ambience to UI feedback - and receive a high-quality audio file. The API supports customizable durations from 0.5 to 22 seconds, seamless looping for continuous playback, and prompt influence control for fine-tuned results. ### Key capabilities * **Text-guided generation**: Describe any sound effect to generate (e.g., "A cat meowing", "Thunder rolling in the distance", "Keyboard typing") * **Customizable duration**: Generate audio from 0.5 seconds to 22 seconds * **Seamless looping**: Enable `loop` for smooth, continuous playback without audible breaks * **Prompt influence control**: Adjust `prompt_influence` (0-1) to balance creativity vs. prompt adherence * **High-quality output**: Professional-grade audio files suitable for production use * **Fast generation**: Typical processing time of 5-15 seconds * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Video production**: Create custom sound effects for films, commercials, and social media videos * **Game development**: Generate UI sounds, environmental ambience, and character audio * **Podcast production**: Add sound effects and transitions to audio content * **App development**: Create notification sounds, button clicks, and interface feedback * **Animation**: Produce foley sounds and effects for animated content * **Presentations**: Enhance slideshows and corporate videos with custom audio ### Generate sound effects with ElevenLabs Submit a text description of the sound effect you want along with the desired duration. The service returns a task ID for async polling or webhook notification.
Create a new sound effect generation task List all sound effects tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------ | --------- | -------- | ------- | ---------------------------------------------------------------------- | | `text` | `string` | Yes | - | Text description of the sound effect to generate (max 2500 characters) | | `duration_seconds` | `number` | Yes | - | Duration of the generated audio (0.5-22 seconds) | | `loop` | `boolean` | No | `false` | Create a sound effect that loops smoothly without audible breaks | | `prompt_influence` | `number` | No | `0.3` | Influence of the prompt on generation (0-1, higher = more literal) | | `webhook_url` | `string` | No | - | HTTPS URL for task completion notification | ### Prompt tips For best results, include details about: * **Sound type**: Specify the exact sound (cat meowing, door creaking, rain falling) * **Intensity**: Describe loudness or energy (soft, loud, aggressive, gentle) * **Environment**: Add context (indoor, outdoor, underwater, in a cave) * **Duration characteristics**: For loops, describe continuous sounds (steady rain, humming motor) **Example prompts:** * "A cat meowing softly, indoor setting" * "Ocean waves crashing on the beach, steady rhythm" * "Thunder rolling in the distance during a storm" * "Keyboard typing rapidly on a mechanical keyboard" * "Glass shattering on a hard floor" * "A car engine starting and idling" * "Birds chirping in a forest at dawn" * "Footsteps on gravel, slow walking pace" ## Frequently Asked Questions ElevenLabs Sound Effects is an AI-powered text-to-audio API that generates realistic sound effects from natural language descriptions. You submit a text description of the sound you want (e.g., "A cat meowing") along with the duration, receive a task ID immediately, then poll for results or receive a webhook notification. The output is a high-quality audio file ready for use in production. You can generate sound effects from 0.5 seconds (minimum) to 22 seconds (maximum). The duration is specified as a decimal number in the `duration_seconds` parameter. For longer audio, consider using the loop feature and playing the clip multiple times. When `loop` is set to `true`, the API generates audio that seamlessly connects end-to-beginning without audible breaks. This is ideal for continuous sounds like rain, wind, machinery hum, or ambient noise that need to play indefinitely. The loop feature works best with sustained, consistent sounds rather than one-time events. The `prompt_influence` parameter (0-1) controls how closely the generated audio matches your text description. Lower values (0-0.3) allow more creative interpretation, while higher values (0.7-1) produce more literal results. The default of 0.3 balances creativity with prompt adherence. Be specific about the sound type, intensity, and environment. Good examples: "A cat meowing softly indoors", "Heavy rain on a metal roof", "Car engine starting then idling". Avoid vague prompts like "noise" or "sound" - instead describe exactly what you want to hear. ElevenLabs Sound Effects outputs high-quality audio files suitable for professional production use. The generated audio is delivered via a URL that remains accessible for download after task completion. Generation typically takes 5-15 seconds depending on the requested duration and complexity. Shorter clips (under 5 seconds) usually complete in under 10 seconds. For production workflows, use webhooks for efficient notification instead of polling. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Prompt specificity**: Use detailed descriptions with context for better results * **Loop planning**: Use loop=true for ambient sounds, loop=false for one-shot effects * **Duration matching**: Match duration to your use case - shorter for UI sounds, longer for ambience * **Prompt influence tuning**: Start with default (0.3), increase for more literal results * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Music Generation](/api-reference/music-generation/overview)**: Generate original music tracks from text descriptions * **[Audio Isolation](/api-reference/audio-isolation/overview)**: Isolate specific sounds from audio or video files * **[Lip Sync](/api-reference/lip-sync/latent-sync/overview)**: Synchronize lip movements with audio tracks # Sound Effects - Generate from text Source: https://docs.magnific.com/api-reference/sound-effects/post-sound-effects post /v1/ai/sound-effects Create realistic sound effects from text descriptions using AI. Generate high-quality audio sound effects based on your text prompts. Perfect for video production, game development, podcasts, and multimedia projects. # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-flex/flux-2-flex-tasks GET /v1/ai/text-to-image/flux-2-flex Retrieve the status of all Flux 2 Flex text-to-image generation tasks for the authenticated user. # Create Image Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-flex/generate POST /v1/ai/text-to-image/flux-2-flex Create highly customizable images using FLUX.2 [flex] from Black Forest Labs, a flexible text-to-image model with an extended resolution range. **Key Features:** - Professional quality with fine-grained control - Text-to-image generation - Image-to-image editing (up to 4 input images) - Wide resolution range (256-1920px) for large-format output - Adjustable guidance and inference steps - Selectable output format (JPEG or PNG) - Reproducible results with seed **Use Cases:** - Large-format marketing materials and posters - Product photography variations - Concept art and illustrations - Image editing and enhancement # FLUX.2 Flex - Text To Image API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-flex/overview Generate professional-grade images from text with FLUX.2 Flex. Fine-grained control over guidance and steps, 256-1920px dimensions and up to 4 input images. Flexible, professional-grade image generation with fine-grained control over guidance, steps, and a wide resolution range for large-format output. FLUX.2 Flex is a text-to-image API that generates professional-grade images with fine-grained control. Developed by Black Forest Labs, it offers adjustable guidance and inference steps plus a wide resolution range from 256 to 1920 pixels per side for large-format output. It also accepts up to 4 optional input images for image-to-image generation and editing, and produces PNG or JPEG output. ### Key capabilities * **Custom resolution**: Set exact `width` and `height` from 256 to 1920 pixels per side * **Fine-grained control**: Adjustable guidance scale (1.5-10) and inference steps (1-50) for quality tuning * **Image-to-image**: Provide up to 4 reference images (`input_image`, `input_image_2`, `input_image_3`, `input_image_4`) for editing and style transfer * **Prompt upsampling**: Optional automatic prompt enhancement for more creative outputs * **Safety control**: Adjustable moderation tolerance from 0 (most strict) to 5 (least strict) * **Seed control**: Optional seed (0 to 4,294,967,295) that influences generation * **Output formats**: PNG (lossless) or JPEG (compressed) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing and advertising**: Generate product visuals and campaign imagery from descriptions * **Large-format output**: Produce high-resolution images for print and display up to 1920px per side * **Content creation**: Create unique illustrations and graphics for blogs, social media, and websites * **Image editing**: Refine and restyle existing images using up to 4 reference inputs * **E-commerce**: Generate product mockups and lifestyle imagery at scale ### Generate images with FLUX.2 Flex Create images by submitting a request to the API. The service returns a task ID for async polling or webhook notification.
Create a new image generation task List all FLUX.2 Flex tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------- | --------- | -------- | ------- | --------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image to generate | | `width` | `integer` | No | `1024` | Image width in pixels (256-1920) | | `height` | `integer` | No | `768` | Image height in pixels (256-1920) | | `seed` | `integer` | No | random | Seed that influences generation (0-4,294,967,295) | | `guidance` | `number` | No | `3.0` | Guidance scale (1.5-10), higher values follow the prompt more closely | | `steps` | `integer` | No | `50` | Inference steps (1-50), more steps produce higher quality | | `prompt_upsampling` | `boolean` | No | `false` | Enable automatic prompt enhancement for creative results | | `safety_tolerance` | `integer` | No | `5` | Content moderation: 0 (strict) to 5 (lenient) | | `output_format` | `string` | No | - | Output format: `jpeg` or `png` | | `input_image` | `string` | No | - | Base64-encoded reference image for image-to-image generation | | `input_image_2` | `string` | No | - | Second base64-encoded reference image | | `input_image_3` | `string` | No | - | Third base64-encoded reference image | | `input_image_4` | `string` | No | - | Fourth base64-encoded reference image | | `webhook_url` | `string` | No | - | URL for completion notification | `width` and `height` are floored to the nearest multiple of 16 by the provider. For example, `300x500` is generated as `288x496`. Choose multiples of 16 for exact dimensions. Generated image URLs are signed and expire quickly (around 10-15 minutes). Download or store the output promptly; do not treat the URL as a permanent link. ## Frequently Asked Questions FLUX.2 Flex is a professional-grade text-to-image model by Black Forest Labs with fine-grained control over resolution, guidance, and steps. You submit a text prompt via the API, optionally include up to 4 reference images, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. FLUX.2 Flex uses explicit `width` and `height` parameters, each accepting 256 to 1920 pixels. The default is 1024×768. Common sizes include 1024×1024 (square), 1920×1080 (16:9 landscape), and 1080×1920 (9:16 portrait). Higher total resolution increases generation cost. You can provide up to 4 Base64-encoded reference images using `input_image`, `input_image_2`, `input_image_3`, and `input_image_4`. The model uses these for image-to-image generation, editing, and style transfer while still following your text prompt. Supported input formats are JPEG, PNG, and WebP. The `guidance` parameter (1.5-10, default 3.0) controls how closely the model follows your prompt; higher values produce images more faithful to the description. The `steps` parameter (1-50, default 50) controls quality: more steps produce finer detail but take longer to process. The `seed` influences generation and helps you steer toward similar results, but identical output for an identical seed is not guaranteed. Setting `prompt_upsampling` to `true` makes runs inherently non-reproducible, since the prompt is rewritten on each call. FLUX.2 Flex offers fine-grained control with custom resolution up to 1920px per side, ideal for large-format and professional output. FLUX.2 Klein is optimized for speed with sub-second generation. Choose FLUX.2 Flex when control and resolution matter; choose FLUX.2 Klein when speed is critical. See [Rate Limits](/ratelimits) for current limits by subscription tier. See [Pricing](/pricing) for detailed rates and subscription options. ## Best practices * **Prompt writing**: Be specific about subjects, scenes, lighting, atmosphere, and art style for better results * **Resolution choice**: Set `width` and `height` for your target format; higher resolution increases cost * **Guidance tuning**: Start with the default guidance (3.0) and increase for more literal prompt adherence * **Reference images**: Use high-quality inputs when doing image-to-image or style transfer * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[FLUX.2 Klein](/api-reference/text-to-image/flux-2-klein/overview)**: Fastest FLUX.2 model for sub-second generation * **[FLUX.2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: High-quality generation for premium results * **[Flux Kontext Max](/api-reference/text-to-image/flux-kontext-max/overview)**: Context-aware image editing with maximum prompt adherence # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-flex/task-by-id GET /v1/ai/text-to-image/flux-2-flex/{task-id} Retrieve the status and results of a specific Flux 2 Flex generation task. # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-klein/flux-2-klein-tasks GET /v1/ai/text-to-image/flux-2-klein Retrieve the status of all FLUX.2 [klein] text-to-image generation tasks for the authenticated user. # Create Image Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-klein/generate POST /v1/ai/text-to-image/flux-2-klein Generate images with sub-second speed using FLUX.2 [klein], the fastest model in the FLUX.2 family by Black Forest Labs. **Key Features:** - Sub-second generation time - Up to 4 reference images for style/subject transfer - 10 preset aspect ratios with 1k or 2k resolution options - Adjustable safety tolerance (0-5) - Multiple output formats (PNG/JPEG) **Use Cases:** - Real-time applications requiring fast generation - Style transfer with reference images - Rapid prototyping and iteration - High-volume image generation # FLUX.2 Klein - Fast Text-to-Image API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-klein/overview Generate images in sub-second time with FLUX.2 Klein. Up to 4 reference images for style transfer. 256-2048px dimensions, PNG/JPEG output for real-time applications. The fastest model in the FLUX.2 family, optimized for sub-second image generation with support for up to 4 reference images. FLUX.2 Klein is a high-speed text-to-image API that generates quality images in sub-second time. Developed by Black Forest Labs, it supports up to 4 reference images for style and subject transfer, making it ideal for real-time applications and high-volume generation workflows. The model produces images in multiple aspect ratios at 1k or 2k resolution in both PNG and JPEG formats. ### Key capabilities * **Sub-second generation**: Fastest model in the FLUX.2 family for real-time applications * **Multi-image reference**: Up to 4 reference images (`input_image`, `input_image_2`, `input_image_3`, `input_image_4`) for style/subject transfer * **Flexible aspect ratios**: 10 preset ratios including square, widescreen, portrait, and social media formats * **Resolution options**: Choose between 1k (standard) or 2k (high resolution, capped at 2048px) * **Safety control**: Adjustable tolerance level from 0 (most strict) to 5 (least strict) * **Output formats**: PNG (lossless) or JPEG (compressed) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Real-time applications**: Chatbots and interactive tools requiring instant image generation * **Style transfer**: Apply visual styles from reference images to new generations * **High-volume workflows**: Generate thousands of images efficiently * **Rapid prototyping**: Quick iteration on creative concepts and designs * **E-commerce**: Generate product variations and lifestyle imagery at scale ### Generate images with FLUX.2 Klein Create images by submitting a request to the API. The service returns a task ID for async polling or webhook notification.
Create a new image generation task List all FLUX.2 Klein tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------ | --------- | -------- | ------------ | ----------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image to generate | | `aspect_ratio` | `string` | No | `square_1_1` | Image aspect ratio (see table below) | | `resolution` | `string` | No | `1k` | Resolution: `1k` (standard) or `2k` (high resolution) | | `seed` | `integer` | No | random | Seed for reproducible generation (0-4,294,967,295) | | `input_image` | `string` | No | - | Base64-encoded reference image for style transfer | | `input_image_2` | `string` | No | - | Second reference image | | `input_image_3` | `string` | No | - | Third reference image | | `input_image_4` | `string` | No | - | Fourth reference image | | `safety_tolerance` | `integer` | No | `2` | Content moderation: 0 (strict) to 5 (lenient) | | `output_format` | `string` | No | - | Output format: `png` or `jpeg` | | `webhook_url` | `string` | No | - | URL for completion notification | ### Aspect ratios | Aspect Ratio | Base Dimensions (1k) | | ------------------- | -------------------- | | `square_1_1` | 1024 × 1024 | | `widescreen_16_9` | 1344 × 768 | | `social_story_9_16` | 768 × 1344 | | `portrait_2_3` | 832 × 1216 | | `traditional_3_4` | 960 × 1280 | | `vertical_1_2` | 704 × 1408 | | `horizontal_2_1` | 1408 × 704 | | `social_post_4_5` | 896 × 1152 | | `standard_3_2` | 1216 × 832 | | `classic_4_3` | 1280 × 960 | ## Frequently Asked Questions FLUX.2 Klein is the fastest model in the FLUX.2 family by Black Forest Labs, optimized for sub-second image generation. You submit a text prompt via the API, optionally include up to 4 reference images, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. You can provide up to 4 Base64-encoded reference images using `input_image`, `input_image_2`, `input_image_3`, and `input_image_4` parameters. The model uses these images to guide style, composition, and subject elements in the generated output while still following your text prompt. FLUX.2 Klein uses preset aspect ratios (square\_1\_1, widescreen\_16\_9, portrait\_2\_3, etc.) combined with resolution settings (1k or 2k). The 2k resolution doubles the base dimensions but caps at 2048px per side. FLUX.2 Klein is the fastest model in the FLUX.2 family, achieving sub-second generation times. This makes it ideal for real-time applications where speed is critical, though FLUX.2 Pro offers higher quality for less time-sensitive use cases. See [Rate Limits](/ratelimits) for current limits by subscription tier. See [Pricing](/pricing) for detailed rates and free tier credits. FLUX.2 Klein is optimized for speed with sub-second generation, ideal for real-time applications. FLUX.2 Pro offers higher quality output for premium results. Choose FLUX.2 Klein when speed is critical; choose FLUX.2 Pro when quality is the priority. ## Best practices * **Reference images**: Use high-quality reference images with clear style elements for best transfer results * **Prompt writing**: Be specific about subjects, scenes, and visual details even when using reference images * **Aspect ratio selection**: Choose the appropriate preset aspect ratio for your use case * **Resolution choice**: Use 1k for speed, 2k for higher detail (capped at 2048px) * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[FLUX.2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: Higher quality generation for premium results * **[FLUX.2 Turbo](/api-reference/text-to-image/post-flux-2-turbo)**: Balance of speed and quality * **[Flux Kontext Pro](/api-reference/text-to-image/flux-kontext-pro/overview)**: Context-aware generation with image guidance # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/flux-2-klein/task-by-id GET /v1/ai/text-to-image/flux-2-klein/{task-id} Retrieve the status and results of a specific FLUX.2 [klein] generation task. # Flux Dev - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-dev/get-flux-dev get /v1/ai/text-to-image/flux-dev Get the status of the flux-dev task # Flux Dev - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/flux-dev/get-flux-dev-task get /v1/ai/text-to-image/flux-dev/{task-id} Get the status of the flux-dev task # Flux Dev - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/flux-dev/post-flux-dev post /v1/ai/text-to-image/flux-dev Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. # Edit Image Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-max/edit-image POST /v1/ai/text-to-image/flux-kontext-max Generate and edit images using FLUX Kontext Max, the highest-quality tier of Black Forest Labs' Kontext family. Kontext Max delivers maximum prompt adherence and detail for context-aware image editing. Provide an input image to guide the transformation, or generate from text alone. Ideal for high-fidelity edits, style changes, and professional image manipulation where output quality is the priority. # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-max/flux-kontext-max-tasks GET /v1/ai/text-to-image/flux-kontext-max Retrieve the status of all Flux Kontext Max text-to-image and image editing tasks for the authenticated user. # Flux Kontext Max - Image Editing API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-max/overview Edit and transform images from text with Flux Kontext Max. Highest-quality Kontext tier with maximum prompt adherence, 10 aspect ratios and fine-grained control. The highest-quality tier of the FLUX Kontext family, delivering maximum prompt adherence and detail for context-aware image editing. Flux Kontext Max is an image editing API that transforms a reference image based on a text prompt. Developed by Black Forest Labs, it is the highest-quality tier of the Kontext family, delivering maximum prompt adherence and fine detail for context-aware editing and generation. It accepts up to 4 reference images and produces high-quality images across 10 aspect ratios in PNG or JPEG format, suitable for professional creative workflows. ### Key capabilities * **Context-aware editing**: Uses a reference image and text prompt together for precise, prompt-driven transformations * **Multi-image reference**: Up to 4 reference image URLs (`input_image`, `input_image_2`, `input_image_3`, `input_image_4`) to guide the edit * **Maximum prompt adherence**: Highest-quality tier of the Kontext family for accurate detail and faithful edits * **Prompt upsampling**: Optional automatic prompt modification for more creative outputs * **Fine-grained control**: Adjustable guidance scale (1-10) and inference steps (1-100) for quality tuning * **10 aspect ratios**: Square (1:1), widescreen (16:9), social story (9:16), classic (4:3), portrait, and more * **Safety control**: Adjustable moderation tolerance from 0 (most strict) to 6 (least strict) * **Output formats**: PNG (lossless) or JPEG (compressed) * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Product photography**: Edit and restyle product shots while preserving key elements * **Marketing and advertising**: Adapt existing campaign imagery to new concepts from text * **Content creation**: Transform illustrations and photos for blogs, social media, and websites * **Design prototyping**: Iterate on visual concepts using a reference image as a starting point * **E-commerce**: Generate consistent lifestyle and mockup variations at scale ### Edit images with Flux Kontext Max Submit a reference image and a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new image editing task List all Flux Kontext Max tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------- | --------- | -------- | ------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image or edit to generate | | `input_image` | `string` | Yes | - | URL to the reference image that guides the generation | | `input_image_2` | `string` | No | - | Second reference image URL | | `input_image_3` | `string` | No | - | Third reference image URL | | `input_image_4` | `string` | No | - | Fourth reference image URL | | `prompt_upsampling` | `boolean` | No | `false` | Enable automatic prompt modification for creative results | | `seed` | `integer` | No | random | Seed that influences generation | | `guidance` | `number` | No | `3.0` | Guidance scale (1-10), higher values follow the prompt more closely | | `steps` | `integer` | No | `50` | Inference steps (1-100), more steps produce higher quality | | `aspect_ratio` | `string` | No | `square_1_1` | Output ratio: `square_1_1`, `classic_4_3`, `traditional_3_4`, `widescreen_16_9`, `social_story_9_16`, `standard_3_2`, `portrait_2_3`, `horizontal_2_1`, `vertical_1_2`, `social_post_4_5` | | `safety_tolerance` | `integer` | No | `2` | Content moderation: 0 (strict) to 6 (lenient) | | `output_format` | `string` | No | - | Output format: `jpeg` or `png` | | `webhook_url` | `string` | No | - | URL for completion notification | `aspect_ratio` accepts named string values only (for example `widescreen_16_9`), not numeric ratios like `"16:9"`. Passing an unlisted value returns a `400` error. ### Aspect ratios | Value | Ratio | | ------------------- | ------------- | | `square_1_1` | 1:1 (default) | | `classic_4_3` | 4:3 | | `traditional_3_4` | 3:4 | | `widescreen_16_9` | 16:9 | | `social_story_9_16` | 9:16 | | `standard_3_2` | 3:2 | | `portrait_2_3` | 2:3 | | `horizontal_2_1` | 2:1 | | `vertical_1_2` | 1:2 | | `social_post_4_5` | 4:5 | Generated image URLs are signed and expire quickly (around 10-15 minutes). Download or store the output promptly; do not treat the URL as a permanent link. ## Frequently Asked Questions Flux Kontext Max is the highest-quality tier of Black Forest Labs' FLUX Kontext family, built for context-aware image editing. You submit a reference image URL (`input_image`) and a text prompt via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. Both edit images using a reference image and text prompt. Flux Kontext Max is the highest-quality tier, offering maximum prompt adherence and detail. Flux Kontext Pro treats the reference image as optional guidance for broader text-to-image generation. Choose Flux Kontext Max when edit precision and detail are the priority. Yes. Flux Kontext Max requires both a `prompt` and an `input_image` URL. The model uses the reference image as the basis for the edit and applies the transformation described in the prompt. You can supply up to 3 additional reference images with `input_image_2`, `input_image_3`, and `input_image_4`. The `seed` influences generation and helps you steer toward similar results, but the Kontext provider does not guarantee identical output for an identical seed. Setting `prompt_upsampling` to `true` makes runs inherently non-reproducible, since the prompt is rewritten on each call. The `guidance` parameter (1-10, default 3.0) controls how closely the model follows your prompt; higher values produce edits more faithful to the description. The `steps` parameter (1-100, default 50) controls quality: more steps produce finer detail but take longer to process. Flux Kontext Max supports 10 aspect ratios, passed as named string values: `square_1_1` (default), `classic_4_3`, `traditional_3_4`, `widescreen_16_9`, `social_story_9_16`, `standard_3_2`, `portrait_2_3`, `horizontal_2_1`, `vertical_1_2`, and `social_post_4_5`. Use the exact enum value, for example `widescreen_16_9` for a 16:9 image; a numeric ratio like `"16:9"` returns a `400` error. Output is delivered as `png` (lossless) or `jpeg` (compressed). See [Rate Limits](/ratelimits) for current limits by subscription tier. See [Pricing](/pricing) for detailed rates and subscription options. ## Best practices * **Reference image**: Use a high-quality `input_image` with clear subjects for the most accurate edits * **Prompt writing**: Describe the exact edit or transformation you want, including lighting, atmosphere, and art style * **Guidance tuning**: Start with the default guidance (3.0) and increase for more literal prompt adherence * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Flux Kontext Pro](/api-reference/text-to-image/flux-kontext-pro/overview)**: Context-aware generation with optional image guidance * **[FLUX.2 Klein](/api-reference/text-to-image/flux-2-klein/overview)**: Fastest FLUX.2 model for sub-second generation * **[FLUX.2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: High-quality text-to-image generation # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-max/task-by-id GET /v1/ai/text-to-image/flux-kontext-max/{task-id} Retrieve the status and results of a specific Flux Kontext Max generation task. # Flux Kontext Pro - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-pro/get-flux-kontext-pro get /v1/ai/text-to-image/flux-kontext-pro Retrieve the status of all Flux Kontext Pro text-to-image generation tasks for the authenticated user. # Flux Kontext Pro - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-pro/get-flux-kontext-pro-task get /v1/ai/text-to-image/flux-kontext-pro/{task-id} Retrieve the status and results of a specific Flux Kontext Pro text-to-image generation task. # Flux Kontext Pro – Text-to-Image API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-pro/overview Generate high-quality images from text with Flux Kontext Pro. Advanced context understanding with optional image input support. Multiple aspect ratios and fine-grained generation control. Advanced text-to-image model that excels at understanding context and generating high-quality images with optional image input guidance. Flux Kontext Pro is a state-of-the-art text-to-image model that combines superior context understanding with flexible generation controls. It supports optional image input to guide the generation process, enabling more precise control over the output. The model produces high-quality images across multiple aspect ratios, suitable for professional creative workflows. ### Key capabilities * **Context-aware generation**: Advanced understanding of complex text descriptions for accurate image synthesis * **Optional image guidance**: Provide a reference image URL to guide the generation toward desired visual styles * **Prompt upsampling**: Automatic prompt modification for more creative and detailed outputs * **Fine-grained control**: Adjustable guidance scale (1-10) and inference steps (1-100) for quality tuning * **Multiple aspect ratios**: Support for square (1:1), widescreen (16:9), social story (9:16), classic (4:3, 3:4), and standard (3:2) * **Reproducible results**: Optional seed parameter for consistent generation across requests ### Use cases * **Marketing and advertising**: Generate product visuals and campaign imagery from descriptions * **Content creation**: Create unique illustrations and graphics for blogs, social media, and websites * **Design prototyping**: Quickly visualize concepts before detailed design work * **E-commerce**: Generate product mockups and lifestyle imagery at scale * **Creative exploration**: Experiment with visual ideas using text prompts and reference images ### Generate images with Flux Kontext Pro Create images by submitting a request to the API. The service returns a task ID for async polling or webhook notification.
Create a new image generation task List all Flux Kontext Pro tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------- | --------- | -------- | ------------ | --------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image to generate | | `input_image` | `string` | No | - | URL to a reference image for guided generation | | `prompt_upsampling` | `boolean` | No | `false` | Enable automatic prompt modification for creative results | | `seed` | `integer` | No | random | Seed for reproducible generation | | `guidance` | `number` | No | `3.0` | Guidance scale (1-10), higher values follow prompt more closely | | `steps` | `integer` | No | `50` | Inference steps (1-100), more steps produce higher quality | | `aspect_ratio` | `string` | No | `square_1_1` | Output aspect ratio: `square_1_1`, `classic_4_3`, `traditional_3_4`, `widescreen_16_9`, `social_story_9_16`, `standard_3_2` | ## Frequently Asked Questions Flux Kontext Pro is an advanced text-to-image AI model that excels at understanding context from text descriptions. You submit a prompt via the API, optionally include a reference image URL, receive a task ID, and then poll for results or receive a webhook notification when the image is ready. When you provide an `input_image` URL, the model uses that image as a reference to guide the generation process. This helps achieve specific visual styles, color palettes, or compositional elements in the output while still following your text prompt. Prompt upsampling automatically modifies your prompt to add more creative details and variations. Enable it when you want the model to expand on your description for more elaborate and artistic results. The `guidance` parameter (1-10, default 3.0) controls how closely the model follows your prompt. Higher values produce images more faithful to the description. The `steps` parameter (1-100, default 50) controls quality: more steps produce finer details but take longer to process. Flux Kontext Pro supports six aspect ratios: `square_1_1` (default), `classic_4_3`, `traditional_3_4`, `widescreen_16_9`, `social_story_9_16`, and `standard_3_2`. Choose based on your intended use case. Yes, use the `seed` parameter with the same value across requests to generate reproducible results. Combined with identical prompts and settings, you will get consistent outputs. ## Best practices * **Prompt writing**: Be specific about subjects, scenes, lighting, atmosphere, and art style for better results * **Guidance tuning**: Start with the default guidance (3.0) and increase for more literal prompt adherence * **Image guidance**: Use reference images when you need specific visual styles or want to maintain consistency across generations * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic for 503 errors during high traffic ## Related APIs * **[Flux 2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: High-quality text-to-image generation without image input support * **[Flux 2 Turbo](/api-reference/text-to-image/post-flux-2-turbo)**: Faster generation with Flux 2 for time-sensitive workflows * **[Seedream 4.5](/api-reference/text-to-image/post-seedream-v4-5)**: Alternative text-to-image model with different capabilities # Flux Kontext Pro - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/flux-kontext-pro/post-flux-kontext-pro post /v1/ai/text-to-image/flux-kontext-pro Generate images using FLUX Kontext Pro, an advanced text-to-image model with optional image input support. This model excels at understanding context and generating high-quality images from text descriptions. Optionally, you can provide an input image to guide the generation process. # Flux Pro 1.1 - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/flux-pro-v1-1/get-flux-pro-v1-1 get /v1/ai/text-to-image/flux-pro-v1-1 Get the status of the flux-pro 1.1 task # Flux Pro 1.1 - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/flux-pro-v1-1/get-flux-pro-v1-1-detail get /v1/ai/text-to-image/flux-pro-v1-1/{task-id} Get the status of the flux-pro 1.1 task # Flux Pro 1.1 - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/flux-pro-v1-1/post-flux-pro-v1-1 post /v1/ai/text-to-image/flux-pro-v1-1 Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. # Flux 2 Pro - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-flux-2-pro get /v1/ai/text-to-image/flux-2-pro Retrieve the status of all Flux 2 Pro text-to-image generation tasks for the authenticated user. # Flux 2 Pro - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-flux-2-pro-task get /v1/ai/text-to-image/flux-2-pro/{task-id} Retrieve the status and results of a specific Flux 2 Pro generation task. # Flux 2 Turbo - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-flux-2-turbo get /v1/ai/text-to-image/flux-2-turbo Retrieve the status of all Flux 2 Turbo text-to-image generation tasks for the authenticated user. # Flux 2 Turbo - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-flux-2-turbo-task get /v1/ai/text-to-image/flux-2-turbo/{task-id} Retrieve the status and results of a specific Flux 2 Turbo generation task. # Gemini 2.5 Flash - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-gemini-2-5-flash-image-preview get /v1/ai/gemini-2-5-flash-image-preview # Gemini 2.5 Flash - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-gemini-2-5-flash-image-preview-task get /v1/ai/gemini-2-5-flash-image-preview/{task-id} # HyperFlux - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-hyperflux get /v1/ai/text-to-image/hyperflux # Create image from text - Classic fast Source: https://docs.magnific.com/api-reference/text-to-image/get-image-from-text post /v1/ai/text-to-image Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. # Get the status of all Nano Banana Pro tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-nano-banana-pro get /v1/ai/text-to-image/nano-banana-pro Get the status of all Nano Banana Pro image generation tasks # Get the status of a Nano Banana Pro task Source: https://docs.magnific.com/api-reference/text-to-image/get-nano-banana-pro-task get /v1/ai/text-to-image/nano-banana-pro/{task-id} Get the status of a specific Nano Banana Pro image generation task # Get the status of all RunWay text-to-image tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-runway get /v1/ai/text-to-image/runway Get the status of all RunWay text-to-image generation tasks # Get RunWay text-to-image task by ID Source: https://docs.magnific.com/api-reference/text-to-image/get-runway-task get /v1/ai/text-to-image/runway/{task-id} Get the status and result of a specific RunWay text-to-image task # Seedream 4.5 - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v4-5 get /v1/ai/text-to-image/seedream-v4-5 Get the status of all Seedream 4.5 image generation tasks # Seedream 4.5 Edit - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v4-5-edit get /v1/ai/text-to-image/seedream-v4-5-edit Get the status of all Seedream 4.5 image editing tasks # Seedream 4.5 Edit - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v4-5-edit-task get /v1/ai/text-to-image/seedream-v4-5-edit/{task-id} Get the status of a specific Seedream 4.5 image editing task # Seedream 4.5 - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v4-5-task get /v1/ai/text-to-image/seedream-v4-5/{task-id} Get the status of a specific Seedream 4.5 image generation task # Seedream V5 Lite - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-lite get /v1/ai/text-to-image/seedream-v5-lite Get the status of all Seedream V5 Lite image generation tasks # Seedream V5 Lite Edit - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-lite-edit get /v1/ai/text-to-image/seedream-v5-lite-edit Get the status of all Seedream V5 Lite image editing tasks # Seedream V5 Lite Edit - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-lite-edit-task get /v1/ai/text-to-image/seedream-v5-lite-edit/{task-id} Get the status of a specific Seedream V5 Lite image editing task # Seedream V5 Lite - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-lite-task get /v1/ai/text-to-image/seedream-v5-lite/{task-id} Get the status of a specific Seedream V5 Lite image generation task # Seedream 5.0 Pro - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-pro get /v1/ai/text-to-image/seedream-v5-pro Get the status of all Seedream 5.0 Pro image generation tasks # Seedream 5.0 Pro Edit - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-pro-edit get /v1/ai/text-to-image/seedream-v5-pro-edit Get the status of all Seedream 5.0 Pro image editing tasks # Seedream 5.0 Pro Edit - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-pro-edit-task get /v1/ai/text-to-image/seedream-v5-pro-edit/{task-id} Get the status of a specific Seedream 5.0 Pro image editing task # Seedream 5.0 Pro - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-seedream-v5-pro-task get /v1/ai/text-to-image/seedream-v5-pro/{task-id} Get the status of a specific Seedream 5.0 Pro image generation task # Get the status of all Z-Image tasks Source: https://docs.magnific.com/api-reference/text-to-image/get-z-image get /v1/ai/text-to-image/z-image Get the status of all Z-Image image generation tasks # Get the status of a Z-Image task Source: https://docs.magnific.com/api-reference/text-to-image/get-z-image-task get /v1/ai/text-to-image/z-image/{task-id} Get the status of a specific Z-Image image generation task # HyperFlux - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/get-{task-id}-by-id get /v1/ai/text-to-image/hyperflux/{task-id} # Edit Image Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2-edit/edit POST /v1/ai/text-to-image/gpt-image-2-edit Edit or recompose one or more input images from a text prompt using OpenAI's GPT Image 2 model. **Key Features:** - Accepts 1-16 input images, used together as the visual context - Restyle a single image, or compose a new one from several - Same output controls as text-to-image (resolution, aspect ratio, format, compression, quality) **Best for:** - Product and scene variations from an existing shot - Combining a subject, a background and a style reference in one call - Adding or correcting text on an existing image **Output:** one edited image per `num_images` (up to 10), delivered as CDN URLs in the task's `generated` array. The input images are never modified. **Credits:** editing sends your images to the model and their input tokens dominate the cost, so it is priced differently from generation: flat units per output image by `quality` (`low` x1, `medium` x4, `high` x6), with `resolution` not affecting the price. The request is processed asynchronously: the endpoint returns a task id immediately. Retrieve the result by polling the task endpoint or by providing an optional `webhook_url`. # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2-edit/gpt-image-2-edit-tasks GET /v1/ai/text-to-image/gpt-image-2-edit Get the status of all GPT Image 2 image editing tasks # GPT Image 2 Edit – Image Editing API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2-edit/overview Edit existing images with GPT Image 2. Prompt-driven retouching, object and style changes, three quality tiers, up to 10 variants per call, no mask needed. Prompt-driven editing of existing images, with the same three quality tiers as GPT Image 2 generation and its own pricing. GPT Image 2 Edit is an image editing API that modifies one or more images you supply according to a text instruction. It is the reference-image counterpart of GPT Image 2: same model, same three quality tiers, same batch limit, but the input is an existing image plus a prompt rather than a prompt alone. Use it for retouching, replacing or removing objects, restyling a scene, and for correcting or replacing text that already appears in the image. ### Key capabilities * **Prompt-driven editing**: describe the change in natural language instead of masking it by hand * **Three quality tiers**: `quality` accepts `low`, `medium` and `high`. The tier changes both the render quality and the credit cost per output image — editing is priced differently from generation, see Pricing below * **Batch variants**: request up to 10 edited variants of the same input in one call with `num_images`. Each output image is billed * **In-image text edits**: legible text rendering carries over to editing, so labels and copy inside the image can be corrected or replaced * **Async processing**: every request returns a task ID immediately. Poll the task endpoint or supply `webhook_url` to be notified on completion ### Use cases * **Product photography retouching**: change backgrounds, lighting or props without a reshoot * **Localisation**: replace on-image copy for another market while keeping the layout * **Packaging revisions**: update a label, claim or price on an approved key visual * **Creative variations**: generate up to 10 alternative treatments of one approved image for A/B testing * **Clean-up**: remove unwanted objects, watermarks or distractions from an existing asset ### Edit images with GPT Image 2 Submit the source image or images plus an instruction to create an editing task. The API responds with a task ID; collect the result by polling or via webhook.
Create a new image editing task List all GPT Image 2 Edit tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | --------------- | --------- | -------- | ------------ | ---------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | The edit to apply, described in natural language | | `quality` | `string` | No | `high` | Quality and cost tier: `low`, `medium` or `high` | | `resolution` | `string` | No | `1k` | Output resolution tier: `1k`, `2k` or `4k`. Unlike generation, it does **not** change the price of an edit | | `aspect_ratio` | `string` | No | `square_1_1` | Shape of the output, e.g. `square_1_1`, `widescreen_16_9`, `banner_3_1` | | `num_images` | `integer` | No | `1` | Number of edited variants to return, up to 10. Each image is billed | | `output_format` | `string` | No | `png` | `png`, `jpeg` or `webp`. `output_compression` applies to the last two | | `background` | `string` | No | - | `opaque` or `auto`. `auto` may return transparency, which only `png` and `webp` can keep | | `moderation` | `string` | No | `auto` | Strictness of the model provider's own filter: `auto` or `low` | | `webhook_url` | `string` | No | - | URL called when the task completes | The source images are supplied on the request body alongside the prompt. The endpoint reference pages above are generated from the OpenAPI specification and are the authoritative, complete parameter list, including the exact field and accepted formats for the input images. ### Pricing Editing sends your source images to the model, and their input tokens dominate the cost, so it is not priced like a generation of the same size: | `quality` | relative cost per output image | | --------- | ------------------------------ | | `low` | x1 | | `medium` | x4 | | `high` | x6 | `resolution` does not affect the price of an edit — the input images set the cost and the output size barely moves it. Every image in `num_images` is billed. ## Frequently Asked Questions GPT Image 2 Edit is an image editing API powered by OpenAI's GPT Image 2 model. You submit one or more source images together with a text instruction and receive a task ID immediately. Poll the task endpoint or configure a webhook to receive the edited images when processing completes. No. The edit is described in the prompt rather than painted as a mask. Be specific about what should change and what must stay the same, since the model decides the region itself. `quality` selects one of three tiers — `low`, `medium` or `high`. A higher tier spends more compute per output image, which improves detail and text fidelity, and costs more credits per image: a `high` edit costs six times a `low` one, and `medium` four times. Up to 10, via `num_images`. Every output image is billed at the tier you selected, so a request for 10 variants at `high` costs ten times a single `high` image. GPT Image 2 generates a new image from a text prompt alone. GPT Image 2 Edit starts from images you provide and modifies them. Both use the same model, the same three quality tiers and the same batch limit, and both bill per output image. Rate limits depend on your subscription tier. See [Rate Limits](/ratelimits) for current limits. Cost per output image depends on the `quality` tier you request. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Name what stays**: state explicitly which parts of the image must be preserved, not only what should change * **One change at a time**: a single, clearly scoped edit per request is more reliable than a list of unrelated changes * **Tier the workflow**: explore variants at `quality: "low"`, then re-run the chosen instruction at `quality: "high"` * **Batch deliberately**: `num_images` is billed per output image, so batch at `low` and finalise at `high` * **Production integration**: use `webhook_url` instead of polling for scalable applications * **Error handling**: implement retry logic with exponential backoff for 503 errors ## Related APIs * **[GPT Image 2](/api-reference/text-to-image/gpt-image-2/overview)**: generate a new image from a text prompt with the same model * **[Seedream 5 Pro Edit](/api-reference/text-to-image/post-seedream-v5-pro-edit)**: reference-image editing with resolution-based tiers * **[Reimagine Flux](/api-reference/text-to-image/reimagine-flux/post-reimagine-flux)**: re-generate variations of an existing image * **[Skin Enhancer](/api-reference/skin-enhancer/overview)**: specialised portrait and skin retouching # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2-edit/task-by-id GET /v1/ai/text-to-image/gpt-image-2-edit/{task-id} Get the status of a specific GPT Image 2 image editing task # Create Image Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2/generate POST /v1/ai/text-to-image/gpt-image-2 Generate images from a text prompt using OpenAI's GPT Image 2 model. **Key Features:** - Strong instruction following on long, detailed prompts - Reliable text rendering inside the image - Explicit resolution tier, aspect ratio, format and compression control - Selectable quality tier (`low` / `medium` / `high`) **Best for:** - Marketing and editorial images that must contain legible text - Prompts with many precise constraints - Cases where transparency or an exact output format is required **Output:** one image per `num_images` (up to 10), at the size selected by `resolution` + `aspect_ratio`, delivered as CDN URLs in the task's `generated` array. **Credits:** the `quality` tier sets the price per image and `resolution` multiplies it (`1k` x1, `2k` x2, `4k` x3). Every image in `num_images` is billed. The request is processed asynchronously: the endpoint returns a task id immediately. Retrieve the result by polling the task endpoint or by providing an optional `webhook_url`. # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2/gpt-image-2-tasks GET /v1/ai/text-to-image/gpt-image-2 Get the status of all GPT Image 2 image generation tasks # GPT Image 2 – Text-to-Image API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2/overview Generate images from text with GPT Image 2. Three quality tiers, up to 10 images per request, strong prompt following and legible in-image text rendering. Text-to-image generation with three cost/quality tiers and unusually reliable prompt following, including legible text rendered inside the image. GPT Image 2 is a text-to-image API that turns a written prompt into one or more images. Built on OpenAI's GPT Image 2 model, it is chosen mainly for two things: it follows long, specific prompts closely, and it renders readable text inside the image — labels, signage, UI copy, packaging — which most diffusion models struggle with. A `quality` parameter selects the tier, so you can iterate cheaply and then re-run the same prompt at higher quality. To edit an existing image instead of generating from scratch, see GPT Image 2 Edit. ### Key capabilities * **Three quality tiers**: `quality` accepts `low`, `medium` and `high`. The tier changes both the render quality and the credit cost of the request, so the same prompt can be drafted cheaply and finalised at `high` * **Batch generation**: request up to 10 images in a single call with `num_images`. Each image in the batch is billed * **Prompt adherence**: follows long, multi-clause prompts including counts, spatial relationships and negations more closely than typical diffusion models * **In-image text**: renders legible words and short phrases inside the image, which makes it suitable for mockups, posters and packaging * **Async processing**: every request returns a task ID immediately. Poll the task endpoint or supply `webhook_url` to be notified on completion ### Use cases * **Marketing and advertising**: on-brand key visuals where the headline or product name has to be readable inside the image * **Packaging and print mockups**: labels, boxes and signage that need real text rather than text-shaped noise * **Product and UI concepting**: interface screens, dashboards and device mockups with legible captions * **Editorial illustration**: article and newsletter art generated from a detailed brief * **Content pipelines at scale**: batch up to 10 variants per prompt at `low`, review, then re-render the winner at `high` ### Generate images with GPT Image 2 Submit a prompt to create a generation task. The API responds with a task ID; collect the result by polling or via webhook.
Create a new image generation task List all GPT Image 2 tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | --------------- | --------- | -------- | ------------ | ----------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image to generate | | `quality` | `string` | No | `high` | Quality and cost tier: `low`, `medium` or `high` | | `resolution` | `string` | No | `1k` | Output resolution tier: `1k`, `2k` or `4k`. Also multiplies the credit cost — see Pricing below | | `aspect_ratio` | `string` | No | `square_1_1` | Shape of the output, e.g. `square_1_1`, `widescreen_16_9`, `banner_3_1`. Combined with `resolution` it selects the exact pixel size | | `num_images` | `integer` | No | `1` | Number of images to generate in one request, up to 10. Each image is billed | | `output_format` | `string` | No | `png` | `png`, `jpeg` or `webp`. `output_compression` applies to the last two | | `background` | `string` | No | - | `opaque` or `auto`. `auto` may return transparency, which only `png` and `webp` can keep | | `moderation` | `string` | No | `auto` | Strictness of the model provider's own filter: `auto` or `low`. Applied on top of Freepik's platform moderation, which cannot be disabled | | `webhook_url` | `string` | No | - | URL called when the task completes | ### Pricing The `quality` tier sets the price per image and `resolution` multiplies it: | | `1k` | `2k` | `4k` | | ----------------- | ---- | ---- | ---- | | credits per image | x1 | x2 | x3 | Every image in `num_images` is billed, so a `4k` request for 4 images costs twelve times a single `1k` image of the same quality tier. The endpoint reference pages above are generated from the OpenAPI specification and are the authoritative, complete parameter list. ## Frequently Asked Questions GPT Image 2 is a text-to-image API powered by OpenAI's GPT Image 2 model. You submit a text prompt and receive a task ID immediately. Poll the task endpoint or configure a webhook to receive the generated images when processing completes. `quality` selects one of three tiers — `low`, `medium` or `high`. A higher tier spends more compute per image, which improves detail and text fidelity, and costs more credits per image. `low` is intended for fast iteration, `high` for final assets. Up to 10, via `num_images`. Every image in the batch is billed at the tier you selected, so a request for 10 images at `high` costs ten times a single `high` image. Yes — reliable in-image text is its main differentiator over most diffusion models. Short, specific strings (a product name, a headline, a button label) work best. Long paragraphs of body copy are still unreliable in any current image model. GPT Image 2 generates a new image from a text prompt alone. GPT Image 2 Edit takes one or more existing images plus a prompt and returns a modified version, for retouching, object changes and style edits. Both share the same three quality tiers and the same batch limit. Rate limits depend on your subscription tier. See [Rate Limits](/ratelimits) for current limits. Cost per image depends on the `quality` tier you request. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Tier the workflow**: draft at `quality: "low"`, review, then re-run the winning prompt at `quality: "high"`. The prompt does not need to change * **Batch deliberately**: `num_images` is billed per image. Batch at `low` for exploration, generate single images at `high` * **Write text explicitly**: put the exact string you want rendered in quotes inside the prompt, and keep it short * **Be specific**: GPT Image 2 rewards long, concrete prompts — subject, composition, lighting, camera, style — more than keyword lists * **Production integration**: use `webhook_url` instead of polling for scalable applications * **Error handling**: implement retry logic with exponential backoff for 503 errors ## Related APIs * **[GPT Image 2 Edit](/api-reference/text-to-image/gpt-image-2-edit/overview)**: edit existing images with the same model and quality tiers * **[Nano Banana Pro](/api-reference/text-to-image/post-nano-banana-pro)**: Google Gemini image generation, also strong at in-image text * **[Seedream 5 Pro](/api-reference/text-to-image/post-seedream-v5-pro)**: high-resolution text-to-image with a generate and edit split * **[FLUX.2 Pro](/api-reference/text-to-image/post-flux-2-pro)**: photorealistic generation with fine-grained resolution control # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/gpt-image-2/task-by-id GET /v1/ai/text-to-image/gpt-image-2/{task-id} Get the status of a specific GPT Image 2 image generation task # Create Image Source: https://docs.magnific.com/api-reference/text-to-image/nano-banana-pro-flash/generate POST /v1/ai/text-to-image/nano-banana-pro-flash Generate images from text descriptions using Google's Nano Banana Pro Flash model (Gemini 3.1 Flash), a faster variant of Nano Banana Pro optimized for quick image generation. **Key Features:** - Fast image generation with Gemini 3.1 Flash - Google Search grounding for real-world accuracy - Support for reference images for guided generation - Multiple aspect ratios and resolutions up to 4K **Best for:** - Rapid image generation with shorter wait times - Grounded image generation using Google Search - Reference-guided generation - Iterative creative workflows where speed matters # List Tasks Source: https://docs.magnific.com/api-reference/text-to-image/nano-banana-pro-flash/nano-banana-pro-flash-tasks GET /v1/ai/text-to-image/nano-banana-pro-flash Get the status of all Nano Banana Pro Flash image generation tasks # Nano Banana Pro Flash API Source: https://docs.magnific.com/api-reference/text-to-image/nano-banana-pro-flash/overview Generate images from text with Nano Banana Pro Flash. Powered by Gemini 3.1 Flash for fast generation with Google Search grounding, reference images, and resolutions up to 4K. The faster variant of Nano Banana Pro, powered by Google's Gemini 3.1 Flash model for rapid image generation with optional web search grounding. Nano Banana Pro Flash (also known as **Nano Banana 2** on fal.ai) is a text-to-image API powered by Google's Gemini 3.1 Flash model. It generates images from natural language prompts with faster processing times than the standard Nano Banana Pro, making it suited for iterative creative workflows and high-volume applications. The model supports Google Search grounding, reference-guided generation with up to 3 images, 10 aspect ratios, and resolutions up to 4K. ### Key capabilities * **Fast generation**: Powered by Gemini 3.1 Flash for shorter processing times compared to Nano Banana Pro * **Google Search grounding**: Enable `use_google_search_tool` for images that need real-world accuracy (places, brands, current events) * **Reference image support**: Up to 3 reference images (PNG, JPEG, WebP) for style and subject guidance * **Flexible aspect ratios**: 10 ratios including `1:1`, `16:9`, `9:16`, `21:9`, `2:3`, `3:2`, `4:3`, `3:4`, `5:4`, `4:5` * **Resolution options**: `1K` (fast), `2K` (balanced), `4K` (best quality) * **Prompt range**: 2 to 3,000 characters for detailed scene descriptions * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Rapid prototyping**: Fast iteration on visual concepts where speed matters more than maximum quality * **Grounded generation**: Images of real places, products, or events using Google Search for accuracy * **Reference-guided creation**: Style transfer and subject guidance from existing images * **Marketing and social media**: Quick generation of campaign visuals in various aspect ratios * **High-volume pipelines**: Batch generation where throughput is a priority ### Generate images with Nano Banana Pro Flash Create images by submitting a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new image generation task List all Nano Banana Pro Flash tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------------ | --------- | -------- | ------- | ------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the image to generate (2-3,000 characters) | | `reference_images` | `array` | No | - | Up to 3 reference images with optional `text` and `mime_type` per image | | `aspect_ratio` | `string` | No | `1:1` | Output ratio: `1:1`, `2:3`, `3:2`, `4:3`, `3:4`, `5:4`, `4:5`, `16:9`, `9:16`, `21:9` | | `resolution` | `string` | No | `1K` | Quality level: `1K`, `2K`, `4K` | | `use_google_search_tool` | `boolean` | No | `false` | Enable Google Search grounding for real-world accuracy | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Nano Banana Pro Flash is a text-to-image API powered by Google's Gemini 3.1 Flash model. You submit a text prompt (2-3,000 characters) via the API, optionally include up to 3 reference images, and receive a task ID immediately. Poll the task endpoint or configure a webhook to receive the generated image URL when processing completes. When `use_google_search_tool` is set to `true`, the model uses Google Search to ground its generation with real-world information. This improves accuracy for prompts referencing real places, people, brands, or current events. The response includes `groundingMetadata` with source information when this feature is active. Nano Banana Pro Flash accepts PNG, JPEG, and WebP reference images. Each reference image must be a publicly accessible URL. You can include up to 3 reference images, each with optional descriptive text and mime type. Three resolution levels: `1K` (fastest generation), `2K` (balanced quality and speed), and `4K` (best quality). Rate limits depend on your subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. Nano Banana Pro Flash uses Google's Gemini 3.1 Flash model optimized for speed, while Nano Banana Pro uses the standard model for maximum quality. Both share the same parameters, aspect ratios, and resolution options. Choose Flash when generation speed matters; choose Pro when you need the highest quality output. Flash defaults to `1K` resolution while Pro defaults to `2K`. ## Best practices * **Google Search grounding**: Enable `use_google_search_tool` when your prompt references real-world entities (landmarks, brands, public figures) for more accurate results * **Reference images**: Use high-quality reference images with clear visual elements for best guided generation * **Prompt writing**: Be specific about subjects, composition, lighting, and style. The 3,000-character limit allows for detailed scene descriptions * **Resolution selection**: Use `1K` for fastest iteration, `2K` for balanced workflows, `4K` for final production assets * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Mystic](/api-reference/mystic/mystic)**: Magnific's proprietary image generation with LoRA style control * **[Flux Kontext Pro](/api-reference/text-to-image/flux-kontext-pro/overview)**: Context-aware generation with image-guided editing * **[Seedream 4](/api-reference/text-to-image/seedream-4/overview)**: High-quality text-to-image generation with advanced prompt understanding # Get Task by ID Source: https://docs.magnific.com/api-reference/text-to-image/nano-banana-pro-flash/task-by-id GET /v1/ai/text-to-image/nano-banana-pro-flash/{task-id} Get the status of a specific Nano Banana Pro Flash image generation task # Flux 2 Pro - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-flux-2-pro post /v1/ai/text-to-image/flux-2-pro Create professional-grade images using FLUX.2 [pro], the next generation of Black Forest Labs' image models. **Key Features:** - Professional quality without complex tuning - Text-to-image generation - Image-to-image editing (up to 4 input images) - Customizable dimensions (256-1440px) - Optional prompt enhancement - Reproducible results with seed **Use Cases:** - Marketing materials and advertisements - Product photography variations - Concept art and illustrations - Image editing and enhancement # Flux 2 Turbo - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-flux-2-turbo post /v1/ai/text-to-image/flux-2-turbo Create high-quality images quickly using FLUX.2 [turbo], the speed-optimized version of Flux 2. **Key Features:** - Fast generation (optimized for speed) - Lower cost than Pro version - Adjustable guidance scale for prompt adherence - Custom image dimensions (512-2048px) - Safety checker for content filtering - Multiple output formats (PNG/JPEG) **Use Cases:** - Rapid prototyping and iteration - Content exploration - High-volume generation - Testing prompts and concepts # Gemini 2.5 Flash - Create or edit image Source: https://docs.magnific.com/api-reference/text-to-image/post-gemini-2-5-flash-image-preview post /v1/ai/gemini-2-5-flash-image-preview This endpoint allows you to generate or edit images using the Gemini 2.5 Flash model. You can provide a prompt for image generation, or include reference images for image editing and style transfer. The model supports both text-to-image generation and image-to-image editing with up to 3 reference images. # HyperFlux - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-hyperflux post /v1/ai/text-to-image/hyperflux This endpoint allows you to generate an image using the HyperFlux model, the fastest Flux model available # Create image from text - Nano Banana Pro Source: https://docs.magnific.com/api-reference/text-to-image/post-nano-banana-pro post /v1/ai/text-to-image/nano-banana-pro Generate high-quality images from text descriptions using Google's Nano Banana Pro model (Gemini 3). **Key Features:** - Advanced image generation with complex compositions - Support for reference images for guided generation - Multiple aspect ratios and resolutions - High-quality output up to 4K resolution **Best for:** - High-quality image generation - Complex scene compositions - Reference-guided generation - Professional visual content # Create image from text - RunWay Source: https://docs.magnific.com/api-reference/text-to-image/post-runway post /v1/ai/text-to-image/runway Generate high-quality images from text descriptions using RunWay's Gen4 Image model. **Key Features:** - Photorealistic and artistic image generation - Multiple aspect ratios supported - Reference image support with @tag syntax - High-resolution output **Best for:** - Photorealistic images - Artistic and creative visuals - Marketing and promotional content # Seedream 4.5 - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v4-5 post /v1/ai/text-to-image/seedream-v4-5 Generate high-quality images from text descriptions using ByteDance's Seedream 4.5 model. **Key Features:** - Superior typography and text rendering - Excellent poster composition and branded visuals - Up to 4MP resolution support (2048x2048) - Enhanced editing consistency **Best for:** - Marketing materials with text - Professional posters and banners - Branded visual content - High-resolution image generation # Seedream 4.5 - Edit image Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v4-5-edit post /v1/ai/text-to-image/seedream-v4-5-edit Edit images using ByteDance's Seedream 4.5 model with text guidance. **Key Features:** - Preserves subject details, lighting, and color tone - Supports up to 5 reference images - Enhanced editing consistency - Up to 4MP output resolution **Best for:** - Image-to-image editing - Style transfer with consistency - Multi-image reference editing # Seedream V5 Lite - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v5-lite post /v1/ai/text-to-image/seedream-v5-lite Generate high-quality images from text descriptions using ByteDance's Seedream V5 Lite model. **Key Features:** - Improved detail and composition over previous versions - Up to 4MP resolution support - Various aspect ratios available - Optional seed for reproducibility **Best for:** - High-quality image generation - Detailed scenes and compositions - Cinematic and artistic imagery # Seedream V5 Lite - Edit image Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v5-lite-edit post /v1/ai/text-to-image/seedream-v5-lite-edit Edit images using ByteDance's Seedream V5 Lite model with text guidance. **Key Features:** - Preserves subject details, lighting, and color tone - Supports up to 5 reference images - Enhanced editing consistency - Up to 4MP output resolution **Best for:** - Image-to-image editing - Style transfer with consistency - Multi-image reference editing # Seedream 5.0 Pro - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v5-pro post /v1/ai/text-to-image/seedream-v5-pro Generate high-quality images from text descriptions using ByteDance's Seedream 5.0 Pro model. **Key Features:** - Flagship image quality with high-fidelity detail - Strong prompt adherence and superior composition - Selectable resolution tiers (1.5K, 2K) - Various aspect ratios available - Optional seed for reproducibility **Best for:** - Professional and large-format imagery - Detailed scenes and complex compositions - Cinematic and artistic visuals # Seedream 5.0 Pro - Edit image Source: https://docs.magnific.com/api-reference/text-to-image/post-seedream-v5-pro-edit post /v1/ai/text-to-image/seedream-v5-pro-edit Edit images using ByteDance's Seedream 5.0 Pro model with text guidance. **Key Features:** - Preserves subject details, lighting, and color tone - Supports up to 10 reference images - Enhanced editing consistency - Selectable resolution tiers (1.5K, 2K) **Best for:** - Image-to-image editing - Style transfer with consistency - Multi-image reference editing # Create image from text - Z-Image Source: https://docs.magnific.com/api-reference/text-to-image/post-z-image post /v1/ai/text-to-image/z-image Generate high-quality images from text descriptions using the Z-Image turbo model. **Key Features:** - Superior speed with turbo architecture - High-quality image generation - Flexible image size configuration - Supports LoRA and ControlNet variants **Best for:** - Fast prototyping and iteration - High-volume image generation - Cost-effective production workloads # Flux Reimagine - Transform image Source: https://docs.magnific.com/api-reference/text-to-image/reimagine-flux/post-reimagine-flux post /v1/ai/beta/text-to-image/reimagine-flux (Beta, synchronous) Reimagine Flux is a new AI model that allows you to generate images from text prompts. # Seedream 4 Edit - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4-edit/get-seedream-v4-edit get /v1/ai/text-to-image/seedream-v4-edit Get the status of all Seedream v4 edit tasks # Seedream 4 Edit - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4-edit/get-seedream-v4-edit-detail get /v1/ai/text-to-image/seedream-v4-edit/{task-id} Get the status of the Seedream v4 edit task # Seedream 4 Edit – Image Editing API Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4-edit/overview Next‑generation editing with precise instruction following, high‑fidelity feature preservation, and multi‑image input/output for production‑ready results. This page introduces the editing‑focused variant of Seedream 4. Endpoints follow the Seedream naming with an editing suffix for clarity. Seedream 4 Edit brings precise, instruction‑driven image editing. It supports addition, deletion, modification, replacement, and reference operations in a single step, while preserving identity, texture, and style. Multi‑image input/output enables complex compositions and storyboard‑like workflows. Looking for high‑quality text‑to‑image generation? See Seedream 4. ### Key capabilities * Precise instruction editing: addition, change, replace, and reference * High‑degree feature preservation to avoid "AI artifacts" * Deep intent understanding for better prompt adherence * Multi‑image input and output for interconnected results * Ultra‑fast generation and up to 4K outputs ### Use cases * Commercial/product edits with brand‑safe preservation * Character/style‑consistent edits across variations and scenes * Multi‑shot creation for campaigns and storyboards * Text replacement and layout‑aware modifications
Create a Seedream 4 Edit task Check Seedream 4 Edit task status List Seedream 4 Edit tasks
## Frequently Asked Questions By default, Seedream 4 Edit prioritizes feature preservation and identity consistency. You can steer larger edits via explicit instructions. Yes. The model supports multi‑image input to guide composition, style, and subject relationships across outputs. Yes. Outputs target production‑ready fidelity. For extreme resolutions, pair with an upscaler. # Seedream 4 - Edit image Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4-edit/post-seedream-v4-edit post /v1/ai/text-to-image/seedream-v4-edit Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. # Seedream 4 - List tasks Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4/get-seedream-v4 get /v1/ai/text-to-image/seedream-v4 Get the status of all Seedream v4 tasks # Seedream 4 - Get task status Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4/get-seedream-v4-detail get /v1/ai/text-to-image/seedream-v4/{task-id} Get the status of the Seedream v4 task # Seedream 4 - Text To Image API Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4/overview Next-generation text-to-image model for high-quality, consistent generations with flexible aspect ratios and production-ready outputs. Seedream 4 is a text‑to‑image model focused on reliable, production‑ready outputs. It delivers high‑quality imagery from natural language prompts, with strong detail retention, coherent composition, and controllable styles across seven aspect ratios. Need precise instruction‑driven editing? Try Seedream 4 Edit. ### Key capabilities * **High‑quality generations** with strong subject and style consistency * **Seven aspect ratios**: `square_1_1`, `widescreen_16_9`, `social_story_9_16`, `portrait_2_3`, `traditional_3_4`, `standard_3_2`, `classic_4_3` * **Guidance scale control** (0–20, default 2.5) to tune prompt adherence vs. creative freedom * **Reproducible outputs** via optional seed parameter (0–2,147,483,647) * **Async workflow** with task polling and optional webhook notifications ### Use cases * Product and lifestyle imagery for e‑commerce and marketing * Concept art and visual ideation with consistent styles * Editorial and social assets tailored to target aspect ratios * Brand‑aligned visuals with coherent color and composition * Automated content pipelines integrated via webhooks ### Generate images with Seedream 4 Create images by submitting a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a Seedream 4 generation task Get task status and results by ID List all Seedream 4 tasks
### Parameters | Parameter | Type | Required | Default | Description | | ---------------- | --------- | -------- | ------------ | ------------------------------------------------------------------------------------------------------------------------------------ | | `prompt` | `string` | Yes | — | Text description of the image to generate | | `aspect_ratio` | `string` | No | `square_1_1` | Output ratio: `square_1_1`, `widescreen_16_9`, `social_story_9_16`, `portrait_2_3`, `traditional_3_4`, `standard_3_2`, `classic_4_3` | | `guidance_scale` | `number` | No | `2.5` | Controls prompt adherence (0–20). Higher values follow the prompt more closely | | `seed` | `integer` | No | — | Random seed for reproducible results (0–2,147,483,647) | | `webhook_url` | `string` | No | — | Callback URL for async task status notifications | ## Frequently Asked Questions Seedream 4 is a text‑to‑image generation model. You submit a text prompt via the API, receive a task ID, and poll for results or receive a webhook notification when the image is ready. Seedream 4 supports seven aspect ratios: `square_1_1`, `widescreen_16_9`, `social_story_9_16`, `portrait_2_3`, `traditional_3_4`, `standard_3_2`, and `classic_4_3`. The default is `square_1_1`. The `guidance_scale` parameter (0–20, default 2.5) controls how closely the output follows the prompt. Higher values produce images that adhere more strictly to the text; lower values allow more creative freedom. Yes. Seedream 4 targets production‑ready image quality and prompt reliability. For extreme resolutions, pair it with an upscaler. Seedream 4 generates images from text prompts. Seedream 4 Edit performs instruction‑driven editing on existing images, supporting addition, deletion, modification, and replacement operations with reference images. # Seedream 4 - Create image from text Source: https://docs.magnific.com/api-reference/text-to-image/seedream-4/post-seedream-v4 post /v1/ai/text-to-image/seedream-v4 Convert descriptive text input into images using AI. This endpoint accepts a variety of parameters to customize the generated images. # Get all Veo 3.1 T2V tasks Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1 get /v1/ai/text-to-video/veo-3-1 # Get all Veo 3.1 T2V Fast tasks Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1-fast get /v1/ai/text-to-video/veo-3-1-fast # Get Veo 3.1 T2V Fast task by ID Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1-fast-task get /v1/ai/text-to-video/veo-3-1-fast/{task-id} # Get all Veo 3.1 Lite T2V tasks Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1-lite get /v1/ai/text-to-video/veo-3-1-lite # Get Veo 3.1 Lite T2V task by ID Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1-lite-task get /v1/ai/text-to-video/veo-3-1-lite/{task-id} # Get Veo 3.1 T2V task by ID Source: https://docs.magnific.com/api-reference/text-to-video/get-veo-3-1-task get /v1/ai/text-to-video/veo-3-1/{task-id} # Happy Horse 1.1 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1-1/generate post /v1/ai/text-to-video/happy-horse-1-1 Generate a high-quality AI video from a text prompt using Happy Horse 1.1, Alibaba's text-to-video model. Supports 9 configurable aspect ratios (16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21, 5:4, 4:5), resolutions (720P, 1080P), and duration from 3 to 15 seconds. **Key features:** - 720P and 1080P resolution support - 9 aspect ratio options - Duration: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 seconds - Prompt up to 2500 characters # Happy Horse 1.1 T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1-1/happy-horse-1-1-t2v-tasks get /v1/ai/text-to-video/happy-horse-1-1 Retrieve the list of all Happy Horse 1.1 text-to-video tasks for the authenticated user. # Happy Horse 1.1 Text To Video API Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1-1/overview Generate AI videos from text prompts with Happy Horse 1.1. 720P and 1080P resolution, 9 aspect ratios, and 3-15 second durations, from Alibaba ATH. Happy Horse 1.1 is the latest iteration of Alibaba ATH's flagship video model, delivering state-of-the-art motion fidelity, prompt adherence, and visual quality, now with 9 aspect ratios for text-to-video. Happy Horse 1.1 Text-to-Video is an AI video generation API that creates MP4 videos from text descriptions. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), it produces high-fidelity video with cinematic motion at 720P (1280x720) or 1080P (1920x1080) resolution, supports 9 configurable aspect ratios, and durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: Latest iteration of the Alibaba ATH Happy Horse video family * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **9 aspect ratios**: `16:9`, `9:16`, `1:1`, `4:3`, `3:4`, `21:9`, `9:21`, `5:4`, `4:5` * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Multilingual prompts**: Up to 2500 characters describing scene, motion, and style * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing videos**: Create product showcases and brand content from text descriptions * **Social media content**: Generate short-form videos for TikTok, Instagram, and YouTube in portrait or landscape * **Concept visualization**: Transform ideas and scripts into motion for rapid prototyping * **Educational content**: Illustrate concepts with AI-generated video explanations * **Storyboarding**: Pre-visualize scenes from a script before live production * **Creative exploration**: Experiment with text prompts and aspect ratios for unique visual content ### API operations Generate videos by submitting a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new text-to-video generation task List all Happy Horse 1.1 T2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | -------------- | --------- | -------- | --------- | ------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video. Max 2500 characters | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"`, `"21:9"`, `"9:21"`, `"5:4"`, `"4:5"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | When the task completes, the generated MP4 video URL is returned in `data.generated`. The URL is valid for 24 hours — download and persist it to permanent storage immediately. ## Frequently Asked Questions Happy Horse 1.1 is an AI video generation model developed by Alibaba ATH (Taotian Innovation Unit), led by Zhang Di. It is the latest iteration of the Happy Horse video family and generates high-fidelity MP4 video from text prompts. Submit a text prompt describing your desired video, receive a task ID immediately, then poll the GET endpoint or wait for a webhook notification when processing completes. The model generates an MP4 video at 720P or 1080P resolution in durations from 3 to 15 seconds. Happy Horse 1.1 T2V supports 9 aspect ratios: `16:9`, `9:16`, `1:1`, `4:3`, `3:4`, `21:9`, `9:21`, `5:4`, and `4:5`. The default is `16:9`. Happy Horse 1.1 T2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Allowed values: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15. Processing time depends on resolution, duration, and server load. Higher resolution (1080P) and longer durations take more time. For production workflows, use webhooks instead of polling for scalable integration. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. Both are Alibaba ATH video models. Happy Horse 1.1 is the newer iteration and expands text-to-video and reference-to-video to 9 aspect ratios (`16:9`, `9:16`, `1:1`, `4:3`, `3:4`, `21:9`, `9:21`, `5:4`, `4:5`) versus 5 in 1.0. Choose Happy Horse 1.1 for the latest quality and wider aspect ratio coverage. ## Best practices * **Prompt writing**: Be specific about scenes, characters, camera movements (zoom, pan, tilt), lighting, and atmosphere. Detailed prompts produce better results than vague descriptions. * **Duration selection**: Start with shorter durations (3-5 seconds) for quick iterations, then increase for final outputs. * **Reproducibility**: Save the `seed` value from successful generations to recreate similar results with the same parameters. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.1 Image-to-Video](/api-reference/image-to-video/happy-horse-1-1/overview)**: Animate a single image into video with Happy Horse 1.1 * **[Happy Horse 1.1 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1-1/overview)**: Generate videos from up to 9 character reference images * **[Happy Horse 1.0 Text-to-Video](/api-reference/text-to-video/happy-horse-1/overview)**: Previous Happy Horse iteration, ranked #1 on the Artificial Analysis Video Arena (April 2026) # Happy Horse 1.1 T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1-1/task-by-id get /v1/ai/text-to-video/happy-horse-1-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.1 text-to-video task by its ID. # Happy Horse 1.0 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1/generate post /v1/ai/text-to-video/happy-horse-1 Generate a high-quality AI video from a text prompt using Happy Horse 1.0, Alibaba's text-to-video model ranked #1 on the Artificial Analysis Video Arena (April 2026). Supports configurable aspect ratios (16:9, 9:16, 1:1, 4:3, 3:4), resolutions (720P, 1080P), and duration from 3 to 15 seconds. **Key features:** - 720P and 1080P resolution support - 5 aspect ratio options - Duration: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 seconds - Prompt up to 2500 characters # Happy Horse 1.0 T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1/happy-horse-1-t2v-tasks get /v1/ai/text-to-video/happy-horse-1 Retrieve the list of all Happy Horse 1.0 text-to-video tasks for the authenticated user. # Happy Horse 1.0 Text To Video API Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1/overview Generate AI videos from text prompts with Happy Horse 1.0, ranked #1 on the Artificial Analysis Video Arena (April 2026). 720P and 1080P resolution, 5 aspect ratios, and 3-15 second durations. Happy Horse 1.0 is Alibaba ATH's flagship video model, ranked #1 on the Artificial Analysis Video Arena (April 2026) for both text-to-video and image-to-video, delivering state-of-the-art motion fidelity, prompt adherence, and visual quality. Happy Horse 1.0 Text-to-Video is an AI video generation API that creates MP4 videos from text descriptions. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), it currently holds the #1 position on the Artificial Analysis Video Arena leaderboard (April 2026) for both T2V and I2V. The model produces high-fidelity video with cinematic motion at 720P (1280x720) or 1080P (1920x1080) resolution, supports 5 configurable aspect ratios, and durations from 3 to 15 seconds. ### Key capabilities * **State-of-the-art quality**: #1 on Artificial Analysis Video Arena (April 2026) for both T2V and I2V * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **5 aspect ratios**: `16:9` landscape, `9:16` portrait, `1:1` square, `4:3` standard, `3:4` portrait * **Flexible durations**: 3 to 15 seconds of video output (default 5 seconds) * **Multilingual prompts**: Up to 2500 characters describing scene, motion, and style * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing videos**: Create product showcases and brand content from text descriptions * **Social media content**: Generate short-form videos for TikTok, Instagram, and YouTube in portrait or landscape * **Concept visualization**: Transform ideas and scripts into motion for rapid prototyping * **Educational content**: Illustrate concepts with AI-generated video explanations * **Storyboarding**: Pre-visualize scenes from a script before live production * **Creative exploration**: Experiment with text prompts and aspect ratios for unique visual content ### API operations Generate videos by submitting a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new text-to-video generation task List all Happy Horse 1.0 T2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | -------------- | --------- | -------- | --------- | -------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video. Max 2500 characters | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Happy Horse 1.0 is an AI video generation model developed by Alibaba ATH (Taotian Innovation Unit), led by Zhang Di. As of April 2026, it ranks #1 on the Artificial Analysis Video Arena leaderboard for both text-to-video (T2V) and image-to-video (I2V) categories, ahead of other state-of-the-art video models. Submit a text prompt describing your desired video, receive a task ID immediately, then poll the GET endpoint or wait for a webhook notification when processing completes. The model generates an MP4 video at 720P or 1080P resolution in durations from 3 to 15 seconds. Happy Horse 1.0 T2V supports 5 aspect ratios: `16:9` (landscape widescreen), `9:16` (portrait/mobile), `1:1` (square), `4:3` (standard landscape), and `3:4` (standard portrait). The default is `16:9`. Happy Horse 1.0 T2V supports integer durations from 3 to 15 seconds. The default is 5 seconds. Allowed values: 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15. Processing time depends on resolution, duration, and server load. Higher resolution (1080P) and longer durations take more time. For production workflows, use webhooks instead of polling for scalable integration. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. Both are Alibaba video models. Happy Horse 1.0 is built by the ATH (Taotian) Innovation Unit and currently ranks #1 on the Artificial Analysis Video Arena (April 2026). WAN 2.7 is built by the WAN team and offers audio-guided generation and longer minimum durations (2 seconds vs 3 seconds for Happy Horse). Choose Happy Horse 1.0 for benchmark-leading quality and Reference-to-Video with up to 9 character references; choose WAN 2.7 for audio-guided generation. ## Best practices * **Prompt writing**: Be specific about scenes, characters, camera movements (zoom, pan, tilt), lighting, and atmosphere. Detailed prompts produce better results than vague descriptions. * **Duration selection**: Start with shorter durations (3-5 seconds) for quick iterations, then increase for final outputs. * **Reproducibility**: Save the `seed` value from successful generations to recreate similar results with the same parameters. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.0 Image-to-Video](/api-reference/image-to-video/happy-horse-1/overview)**: Animate a single image into video with Happy Horse 1.0 * **[Happy Horse 1.0 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1/overview)**: Generate videos from up to 9 character reference images * **[Happy Horse 1.0 Video Edit](/api-reference/video-edit/happy-horse-1/overview)**: Edit videos with natural-language instructions * **[WAN 2.7 Text-to-Video](/api-reference/text-to-video/wan-2-7/overview)**: Alibaba WAN T2V with audio-guided generation # Happy Horse 1.0 T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/happy-horse-1/task-by-id get /v1/ai/text-to-video/happy-horse-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.0 text-to-video task by its ID. # Kling 3.0 Turbo T2V 1080p - Generate video Source: https://docs.magnific.com/api-reference/text-to-video/kling-v3-turbo/generate-1080p post /v1/ai/text-to-video/kling-v3-turbo-1080p Generate AI video from a text prompt using Kling 3.0 Turbo at 1080p resolution, a fast text-to-video model from the Kling 3.0 family. **Features:** - **Text-to-video**: Generate videos from detailed text prompts - **Resolution**: Fixed 1080p output - **Aspect ratio**: Landscape (16:9), portrait (9:16), or square (1:1) - **Duration**: 3-15 seconds **How it works:** 1. Submit a text prompt describing your desired video 2. Receive a task ID to track progress 3. Poll the status endpoint or use a `webhook_url` to get notified when the task completes 4. Download the generated video from the result URL (valid for 24 hours) # Kling 3.0 Turbo T2V 720p - Generate video Source: https://docs.magnific.com/api-reference/text-to-video/kling-v3-turbo/generate-720p post /v1/ai/text-to-video/kling-v3-turbo-720p Generate AI video from a text prompt using Kling 3.0 Turbo at 720p resolution, a fast text-to-video model from the Kling 3.0 family. **Features:** - **Text-to-video**: Generate videos from detailed text prompts - **Resolution**: Fixed 720p output - **Aspect ratio**: Landscape (16:9), portrait (9:16), or square (1:1) - **Duration**: 3-15 seconds **How it works:** 1. Submit a text prompt describing your desired video 2. Receive a task ID to track progress 3. Poll the status endpoint or use a `webhook_url` to get notified when the task completes 4. Download the generated video from the result URL (valid for 24 hours) # Kling 3.0 Turbo API Source: https://docs.magnific.com/api-reference/text-to-video/kling-v3-turbo/overview Generate AI videos fast with Kling 3.0 Turbo. Text-to-video and image-to-video with dedicated 720p and 1080p generate endpoints, 3-15 second durations, and 16:9, 9:16, or 1:1 aspect ratios. Kling 3.0 Turbo is the fast generation model from the Kling 3.0 family, delivering smooth motion and high visual fidelity with shorter processing times for rapid iteration. Kling 3.0 Turbo is an AI video generation API optimized for speed. It produces high-fidelity videos from a text prompt (text-to-video) or from a single image (image-to-video), with durations from 3 to 15 seconds. As part of the Kling 3.0 family, Turbo prioritizes faster generation, making it well suited for high-volume workflows and quick creative iteration. Resolution is selected at generation time: each mode has a dedicated **720p** generate endpoint and a dedicated **1080p** generate endpoint. Task status and listing are handled by a single generic endpoint per mode, regardless of the resolution used to create the task. ### Key capabilities * **Dual input modes**: Generate video from a text prompt (t2v) or animate a first-frame image (i2v) * **Resolution by generate endpoint**: Dedicated `720p` (standard high definition) and `1080p` (full high definition) generate endpoints per mode * **Unified task tracking**: One generic list endpoint and one generic status endpoint per mode, for tasks of any resolution * **Fast generation**: Turbo-optimized pipeline for shorter processing times than standard Kling 3.0 models * **Flexible durations**: Any whole number from 3 to 15 seconds, default `5` * **Aspect ratios (t2v)**: `16:9` landscape, `9:16` portrait, and `1:1` square * **Prompt control**: Up to 3072 characters in text-to-video, up to 2500 characters of motion guidance in image-to-video * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated MP4 video is returned via a URL valid for 24 hours ### Use cases * **Social media content**: Produce vertical `9:16` clips for TikTok, Instagram Reels, and YouTube Shorts * **Marketing videos**: Animate product images or generate brand content from prompts at scale * **Rapid prototyping**: Iterate quickly on creative concepts thanks to fast Turbo generation * **Concept visualization**: Turn static designs into motion for pitches and previews * **Ad variations**: Generate multiple resolutions, aspect ratios, and durations from a single concept ### Generation modes | Mode | Input | Resolution | Aspect ratio control | Best for | | ------------------------ | ---------------------------------------------- | ----------------------------------- | --------------------------- | ------------------------------------------------- | | **Text-to-Video (t2v)** | Text prompt (required) | `720p` or `1080p` generate endpoint | `16:9`, `9:16`, `1:1` | Creating videos from scratch, concept exploration | | **Image-to-Video (i2v)** | First-frame image (required) + optional prompt | `720p` or `1080p` generate endpoint | Inherited from source image | Animating existing images and product shots | ## API Operations ### Text-to-Video (t2v) Generate a video from a text description using the 720p or 1080p endpoint, then track it with the generic list and status endpoints. Generation returns a task ID for async polling or webhook notification.
Generate a 720p video from a text prompt Generate a 1080p video from a text prompt List all Kling 3.0 Turbo T2V tasks (any resolution) Get T2V task status and result by ID
### Image-to-Video (i2v) Animate a first-frame image using the 720p or 1080p endpoint, then track it with the generic list and status endpoints. Generation returns a task ID for async polling or webhook notification.
Generate a 720p video from a first-frame image Generate a 1080p video from a first-frame image List all Kling 3.0 Turbo I2V tasks (any resolution) Get I2V task status and result by ID
### Endpoint structure | Operation | Text-to-Video | Image-to-Video | | ------------------ | --------------------------------------------------- | ---------------------------------------------------- | | **Generate 720p** | `POST /v1/ai/text-to-video/kling-v3-turbo-720p` | `POST /v1/ai/image-to-video/kling-v3-turbo-720p` | | **Generate 1080p** | `POST /v1/ai/text-to-video/kling-v3-turbo-1080p` | `POST /v1/ai/image-to-video/kling-v3-turbo-1080p` | | **List tasks** | `GET /v1/ai/text-to-video/kling-v3-turbo` | `GET /v1/ai/image-to-video/kling-v3-turbo` | | **Get task** | `GET /v1/ai/text-to-video/kling-v3-turbo/{task-id}` | `GET /v1/ai/image-to-video/kling-v3-turbo/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-720p` or `-1080p`) and is not a body parameter. #### Text-to-Video | Parameter | Type | Required | Default | Description | | -------------- | -------- | -------- | ------- | ---------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video. 1-3072 characters (2500 or fewer recommended) | | `aspect_ratio` | `string` | No | `16:9` | Output ratio: `16:9` (landscape), `9:16` (portrait), `1:1` (square) | | `duration` | `string` | No | `5` | Video length in seconds, any whole number from `3` to `15` | | `webhook_url` | `string` | No | - | URL for async status notifications | #### Image-to-Video | Parameter | Type | Required | Default | Description | | ------------- | -------- | -------- | ------- | --------------------------------------------------------------------------------------------------------------------------- | | `image` | `string` | Yes | - | First-frame image as a public URL or Base64 string. JPG, JPEG, or PNG; min 300x300px; max 50MB; aspect ratio 1:2.5 to 2.5:1 | | `prompt` | `string` | No | - | Motion and style guidance. Up to 2500 characters. If omitted, motion is inferred from the image | | `duration` | `string` | No | `5` | Video length in seconds, any whole number from `3` to `15` | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Kling 3.0 Turbo is the fast-generation model in the Kling 3.0 family. You submit a text prompt (text-to-video) or a first-frame image (image-to-video) to a resolution-specific generate endpoint and receive a task ID immediately. Poll the generic GET status endpoint or supply a `webhook_url` to be notified when the task completes, then download the MP4 video from the returned URL (valid for 24 hours). Resolution is selected by the generate endpoint, not by a body parameter. Call the `-720p` endpoint for standard high definition or the `-1080p` endpoint for full high definition. Task listing and status use a single generic endpoint per mode, regardless of the resolution. No. Status and listing are generic per mode: `GET /v1/ai/text-to-video/kling-v3-turbo` (and `/{task-id}`) covers all text-to-video tasks, and `GET /v1/ai/image-to-video/kling-v3-turbo` (and `/{task-id}`) covers all image-to-video tasks, whether they were generated at 720p or 1080p. Text-to-video generates a video purely from a text prompt and lets you choose the aspect ratio (`16:9`, `9:16`, or `1:1`). Image-to-video animates a first-frame image you provide; the aspect ratio is inherited from that image and the prompt is optional, used only to guide motion and style. Kling 3.0 Turbo supports any whole-number duration from 3 to 15 seconds (default `5`). Shorter durations generate faster. Text-to-video supports `16:9` (landscape), `9:16` (portrait), and `1:1` (square). Image-to-video does not accept an aspect ratio parameter; the output ratio is inherited from the source image. Image-to-video accepts JPG, JPEG, and PNG, provided as a publicly accessible URL or a Base64-encoded string (without the data URI prefix). Requirements: minimum 300x300 pixels, maximum 50MB, and an aspect ratio between 1:2.5 and 2.5:1. Only the first frame is supported. Generation time depends on resolution and duration, but Turbo is optimized for faster processing than standard Kling 3.0 models. Poll the GET task endpoint or provide a `webhook_url` to be notified on completion. The result MP4 is delivered via a URL that remains valid for 24 hours. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Pricing varies by resolution and duration. See the [Pricing](/pricing) page for current rates and credit costs. ## Best practices * **Resolution choice**: Use the `-720p` generate endpoint for faster generation and the `-1080p` generate endpoint when you need maximum detail * **Prompt writing**: Be specific about the scene, subjects, motion, camera movement, lighting, and visual style for better results * **Negative descriptions inline**: Steer the model away from artifacts by adding phrases like "no text, no blur" directly in the prompt * **Image selection (i2v)**: Use clear, high-resolution images within the 300x300px to 50MB range and the 1:2.5 to 2.5:1 aspect ratio bounds * **Duration selection**: Start at 3-5 seconds for fast iterations, then increase up to 15 seconds for final outputs * **Production integration**: Use webhooks instead of polling for scalable applications * **Result retrieval**: Download the output promptly, as the delivery URL is valid for 24 hours ## Related APIs * **[Kling 3](/api-reference/video/kling-v3/overview)**: Standard Kling 3.0 model for higher-fidelity output * **[Kling 2.6](/api-reference/image-to-video/kling-v2-6/task-by-id)**: Previous-generation Kling with motion control * **[WAN 2.7](/api-reference/text-to-video/wan-2-7/overview)**: Alternative t2v and i2v model with multi-shot sequences * **[Runway Gen-4 Turbo](/api-reference/image-to-video/runway-gen4-turbo)**: Fast i2v alternative with cinematic quality # Kling 3.0 Turbo T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/kling-v3-turbo/task-by-id get /v1/ai/text-to-video/kling-v3-turbo/{task-id} Retrieve the status and result of a specific Kling 3.0 Turbo T2V task by its task ID, regardless of the resolution it was created at. # Kling 3.0 Turbo T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/kling-v3-turbo/tasks get /v1/ai/text-to-video/kling-v3-turbo Retrieve the list of all Kling 3.0 Turbo T2V tasks for the authenticated user, across all resolutions. # LTX Video 2.0 Fast - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-fast post /v1/ai/text-to-video/ltx-2-fast Generate a video from text prompt using the LTX Video 2.0 Fast model. **Features:** - Fast video generation with resolutions up to 4K (2160p) - Extended duration options: 6-20 seconds in 2-second increments - Optional synchronized audio generation - **Note:** Durations longer than 10 seconds require 25 FPS and 1080p resolution # LTX Video 2.0 Fast T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-fast-tasks get /v1/ai/text-to-video/ltx-2-fast # LTX Video 2.0 Fast T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-fast/task-by-id get /v1/ai/text-to-video/ltx-2-fast/{task-id} # LTX Video 2.0 Pro - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-pro post /v1/ai/text-to-video/ltx-2-pro Generate a video from text prompt using the LTX Video 2.0 Pro model. **Features:** - High-quality video generation with resolutions up to 4K (2160p) - Duration options: 6, 8, or 10 seconds - Optional synchronized audio generation - Reproducible results with seed parameter # LTX Video 2.0 Pro T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-pro-tasks get /v1/ai/text-to-video/ltx-2-pro # LTX Video 2.0 Pro T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/ltx-2-pro/task-by-id get /v1/ai/text-to-video/ltx-2-pro/{task-id} # PixVerse V5.5 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5-5/create post /v1/ai/text-to-video/pixverse-v5-5 Generate a video from a text prompt using the PixVerse-V5.5 text-to-video model. Compared to PixVerse-V5, version 5.5 adds: - 10-second duration (in addition to 5s and 8s; 10s capped at 720p) - Native audio generation via `generate_audio_switch` - Multi-clip output with dynamic camera changes via `generate_multi_clip_switch` - Prompt reasoning enhancement via `thinking_type` # PixVerse V5.5 Text to Video API Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5-5/overview Generate videos from text prompts with PixVerse V5.5. Native synchronized audio, multi-clip output, prompt reasoning, and 5, 8, or 10 second durations. ## Overview PixVerse V5.5 Text to Video is a text-to-video API that generates a video from a text prompt with native synchronized audio and multi-clip output. Compared to PixVerse V5, version 5.5 adds `generate_audio_switch` for background music, sound effects, or dialogue, `generate_multi_clip_switch` for dynamic camera changes within one generation, `thinking_type` prompt reasoning, and a 10-second duration option. ### Key capabilities * **Native synchronized audio**: enable `generate_audio_switch` to produce audio (music, SFX, or dialogue) together with the video * **Multi-clip with dynamic cameras**: enable `generate_multi_clip_switch` for multi-clip output with camera changes in a single generation * **Prompt reasoning (`thinking_type`)**: choose `enabled` (default), `disabled`, or `auto` to control whether the model rewrites the prompt before rendering * **Duration (5, 8, or 10 seconds)**: `5` (default), `8`, or `10`; 8s costs double, 10s is available up to 720p, and 1080p is limited to 5 or 8 seconds * **Resolutions**: `360p`, `540p`, `720p`, `1080p` * **Aspect ratios**: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` * **Visual styles**: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` * **Async processing**: poll the task endpoint or receive a webhook notification on completion ### Use cases * **Marketing and ads**: short promotional clips with synchronized audio generated from a brief * **Social content**: vertical `social_story_9_16` clips with sound for TikTok, Instagram Reels, and YouTube Shorts * **Story sequences**: multi-clip output with camera changes for teasers and narrative shorts * **Stylized shorts**: `anime`, `cyberpunk`, or `clay` looks for art and concept work * **Prototyping**: iterate on prompt and audio variations before production
Generate a video from a text prompt with PixVerse V5.5 Get task status and result by ID List all PixVerse V5.5 text-to-video tasks
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | -------------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text prompt describing the video to generate | | `aspect_ratio` | `string` | No | `widescreen_16_9` | `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` | | `resolution` | `string` | No | - | Output resolution: `360p`, `540p`, `720p`, `1080p` | | `duration` | `integer` | No | `5` | Video length in seconds: `5`, `8`, or `10`. 8s costs double; 10s up to 720p; 1080p limited to 5 or 8s | | `negative_prompt` | `string` | No | `""` | Negative prompt for the generation | | `style` | `string` | No | - | Visual style: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` | | `seed` | `integer` | No | - | Reproducibility seed; same seed + same prompt yields the same video | | `generate_audio_switch` | `boolean` | No | `false` | When `true`, generates synchronized audio (music, SFX, or dialogue) with the video | | `generate_multi_clip_switch` | `boolean` | No | `false` | When `true`, produces multi-clip output with dynamic camera changes in a single generation | | `thinking_type` | `string` | No | `enabled` | Prompt reasoning mode: `enabled`, `disabled`, or `auto` | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ## Frequently Asked Questions PixVerse V5.5 adds native synchronized audio (`generate_audio_switch`), multi-clip output (`generate_multi_clip_switch`), prompt reasoning (`thinking_type`), and a 10-second duration option. PixVerse V5 supports only 5 or 8 second durations with no audio or multi-clip. Set `generate_audio_switch` to `true` and PixVerse V5.5 produces synchronized audio (background music, sound effects, or dialogue) together with the video in a single request. No separate audio call is required. When set to `true`, PixVerse V5.5 produces multi-clip output with dynamic camera changes inside a single generation, simulating cuts and camera moves without stitching multiple requests. PixVerse V5.5 supports `5` (default), `8`, or `10` second durations and resolutions of `360p`, `540p`, `720p`, and `1080p`. 8-second videos cost double, 10-second videos are available up to 720p, and 1080p is limited to 5 or 8 seconds. `thinking_type` controls prompt reasoning. `enabled` (default) rewrites the prompt automatically for better results, `disabled` uses the prompt exactly as written, and `auto` lets the model decide based on the input. Rate limits and pricing depend on your subscription tier. See [Rate Limits](/ratelimits) and the [Pricing page](/pricing) for current values. ## Best practices * **Audio**: enable `generate_audio_switch` only when you want the model to author audio; if you have your own track, leave it `false` and mix externally * **Duration selection**: use `5` seconds for most shots to reduce cost; request 10s only at 720p or lower * **Prompt reasoning**: leave `thinking_type` as `enabled` for general prompts; switch to `disabled` when you need literal prompt adherence * **Multi-clip**: enable `generate_multi_clip_switch` for narrative sequences that benefit from camera changes * **Production integration**: use `webhook_url` instead of polling for scalable workflows * **Error handling**: implement retry with exponential backoff for 503 responses ## Related APIs * **PixVerse V6 Text to Video**: adds flexible 1-15 second durations and extra aspect ratios (`portrait_2_3`, `landscape_3_2`, `cinematic_21_9`). See the [PixVerse V6 Text to Video overview](/api-reference/text-to-video/pixverse-v6/overview). * **PixVerse V5 Text to Video**: earlier version without audio, multi-clip, or 10-second support. See the [PixVerse V5 Text to Video overview](/api-reference/text-to-video/pixverse-v5/overview). * **PixVerse V5.5 (image-to-video)**: animate an existing image instead of generating from text. See the [PixVerse V5.5 overview](/api-reference/image-to-video/pixverse-v5-5/overview). # PixVerse V5.5 Text-to-Video - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5-5/pixverse-v5-5-tasks get /v1/ai/text-to-video/pixverse-v5-5 List PixVerse-V5.5 text-to-video tasks. # PixVerse V5.5 Text-to-Video - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5-5/task-by-id get /v1/ai/text-to-video/pixverse-v5-5/{task-id} # PixVerse V5 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5/create post /v1/ai/text-to-video/pixverse-v5 Generate a video from a text prompt using the PixVerse-V5 text-to-video model. Control aspect ratio, resolution, duration and visual style directly from the request body. # PixVerse V5 Text to Video API Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5/overview Generate videos from text prompts with PixVerse V5. Control aspect ratio, resolution, duration, and visual style. 5 or 8 second output for social and marketing. ## Overview PixVerse V5 Text to Video is a text-to-video API that generates a video directly from a text prompt, with no input image required. Describe the scene in `prompt` and control the output with aspect ratio, resolution, duration, and visual style. It produces 5 or 8 second clips at up to 1080p, making it a fast option for social content, ads, and concept work. ### Key capabilities * **Text-only generation**: create a video from a `prompt` alone, no source image needed * **Duration (5 or 8 seconds)**: `5` (default) or `8`; 8-second videos cost double, and 1080p output is limited to 5 seconds * **Resolutions**: `360p`, `540p`, `720p`, `1080p` * **Aspect ratios**: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` * **Visual styles**: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` * **Negative prompt and seed**: refine output with `negative_prompt` and reproduce results with `seed` * **Async processing**: poll the task endpoint or receive a webhook notification on completion ### Use cases * **Marketing and ads**: short promotional clips generated straight from a copy line or brief * **Social content**: vertical `social_story_9_16` clips for TikTok, Instagram Reels, and YouTube Shorts * **Concept and pitching**: quick visual drafts from a written idea before committing to production * **Stylized shorts**: `anime`, `cyberpunk`, or `clay` looks for art and design work * **Prototyping**: fast iteration on prompt and style variations with a fixed `seed`
Generate a video from a text prompt with PixVerse V5 Get task status and result by ID List all PixVerse V5 text-to-video tasks
### Parameters | Parameter | Type | Required | Default | Description | | ----------------- | -------------- | -------- | ----------------- | -------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text prompt describing the video to generate | | `aspect_ratio` | `string` | No | `widescreen_16_9` | `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16` | | `resolution` | `string` | No | - | Output resolution: `360p`, `540p`, `720p`, `1080p` | | `duration` | `integer` | No | `5` | Video length in seconds: `5` or `8`. 8s costs double; 1080p is limited to 5s | | `negative_prompt` | `string` | No | `""` | Negative prompt for the generation | | `style` | `string` | No | - | Visual style: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` | | `seed` | `integer` | No | - | Reproducibility seed; same seed + same prompt yields the same video | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ## Frequently Asked Questions PixVerse V5 Text to Video generates a video from a text `prompt` without any input image. You submit the prompt and optional controls, receive a task ID immediately, then poll the task endpoint or receive a webhook notification when the video is ready. No. This is a text-only endpoint: the video is generated entirely from your `prompt`. To animate from an existing image, use an image-to-video PixVerse endpoint instead. PixVerse V5 supports `5` (default) or `8` second durations and resolutions of `360p`, `540p`, `720p`, and `1080p`. 8-second videos cost double, and 1080p output is limited to 5 seconds. Five ratios: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, and `social_story_9_16`. Use `social_story_9_16` for vertical social formats. Pass the same `seed` together with the same `prompt` and model version. Identical inputs produce the same video every time, which is useful for controlled iteration. Rate limits and pricing depend on your subscription tier. See [Rate Limits](/ratelimits) and the [Pricing page](/pricing) for current values. ## Best practices * **Prompt detail**: describe subject, setting, motion, and mood explicitly for stronger, more controllable results * **Duration selection**: use `5` seconds for most shots to reduce cost; reserve `8` seconds when the extra length is needed * **Resolution vs duration**: request 1080p only for 5-second clips; use 720p or lower for 8-second output * **Style consistency**: set a single `style` per project and keep `seed` fixed to compare prompt variations * **Production integration**: use `webhook_url` instead of polling for scalable workflows * **Error handling**: implement retry with exponential backoff for 503 responses ## Related APIs * **PixVerse V5.5 Text to Video**: adds native audio, multi-clip output, prompt reasoning, and 10-second duration. See the [PixVerse V5.5 Text to Video overview](/api-reference/text-to-video/pixverse-v5-5/overview). * **PixVerse V6 Text to Video**: adds flexible 1-15 second durations and extra aspect ratios. See the [PixVerse V6 Text to Video overview](/api-reference/text-to-video/pixverse-v6/overview). * **PixVerse V5 (image-to-video)**: animate an existing image instead of generating from text. See the [PixVerse V5 overview](/api-reference/image-to-video/pixverse/overview). # PixVerse V5 Text-to-Video - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5/pixverse-v5-tasks get /v1/ai/text-to-video/pixverse-v5 List PixVerse-V5 text-to-video tasks. # PixVerse V5 Text-to-Video - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v5/task-by-id get /v1/ai/text-to-video/pixverse-v5/{task-id} # PixVerse V6 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v6/create post /v1/ai/text-to-video/pixverse-v6 Generate a video from a text prompt using the PixVerse-V6 text-to-video model. Compared to PixVerse-V5.5, version 6 adds: - Flexible duration: any integer from `1` to `15` seconds (previously fixed to `5`, `8` or `10`). - Extended aspect ratios: `2:3`, `3:2` and `21:9` in addition to `16:9`, `4:3`, `1:1`, `3:4` and `9:16`. Existing PixVerse-V5.5 features are kept: native audio (`generate_audio_switch`), multi-clip output (`generate_multi_clip_switch`) and prompt reasoning enhancement (`thinking_type`). # PixVerse V6 Text to Video API Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v6/overview Generate videos from text prompts with PixVerse V6. Flexible 1-15 second durations, native audio, multi-clip output, and cinematic 21:9 aspect ratios. ## Overview PixVerse V6 Text to Video is a text-to-video API that generates a video from a text prompt with flexible duration and cinematic aspect ratios. Compared to PixVerse V5.5, version 6 unlocks any integer duration from `1` to `15` seconds (instead of fixed `5`, `8`, or `10`) and adds three aspect ratios: `portrait_2_3`, `landscape_3_2`, and `cinematic_21_9`. Native synchronized audio, multi-clip output, and prompt reasoning are all preserved. ### Key capabilities * **Flexible duration (1-15 seconds)**: any integer length from `1` to `15`, with default `5` (previously fixed to `5`, `8`, or `10`) * **Extended aspect ratios**: `widescreen_16_9` (default), `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, plus `portrait_2_3`, `landscape_3_2`, and `cinematic_21_9` * **Native synchronized audio**: enable `generate_audio_switch` to produce music, sound effects, or dialogue together with the video * **Multi-clip with dynamic cameras**: enable `generate_multi_clip_switch` for multi-clip output with camera changes in a single generation * **Prompt reasoning (`thinking_type`)**: choose `enabled` (default), `disabled`, or `auto` to control whether the model rewrites the prompt before rendering * **Resolutions**: `360p`, `540p`, `720p`, `1080p` * **Visual styles**: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` * **Async processing**: poll the task endpoint or receive a webhook notification on completion ### Use cases * **Marketing and ads**: promotional clips in any length that fits the channel, with synchronized audio * **Cinematic content**: widescreen `cinematic_21_9` sequences for teasers and film-style shots * **Social content**: vertical `social_story_9_16` or `portrait_2_3` clips for TikTok, Instagram Reels, and YouTube Shorts * **Short films**: longer 10-15 second sequences with multi-clip camera transitions * **Stylized shorts**: `anime`, `cyberpunk`, or `clay` looks for art and concept work
Generate a video from a text prompt with PixVerse V6 Get task status and result by ID List all PixVerse V6 text-to-video tasks
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | -------------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text prompt describing the video to generate | | `aspect_ratio` | `string` | No | `widescreen_16_9` | `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, `portrait_2_3`, `landscape_3_2`, `cinematic_21_9` | | `resolution` | `string` | No | - | Output resolution: `360p`, `540p`, `720p`, `1080p` | | `duration` | `integer` | No | `5` | Video length in seconds. Any integer from `1` to `15` | | `negative_prompt` | `string` | No | `""` | Negative prompt for the generation | | `style` | `string` | No | - | Visual style: `anime`, `3d_animation`, `clay`, `cyberpunk`, `comic` | | `seed` | `integer` | No | - | Reproducibility seed; same seed + same prompt yields the same video | | `generate_audio_switch` | `boolean` | No | `false` | When `true`, generates synchronized audio (music, SFX, or dialogue) with the video | | `generate_multi_clip_switch` | `boolean` | No | `false` | When `true`, produces multi-clip output with dynamic camera changes in a single generation | | `thinking_type` | `string` | No | `enabled` | Prompt reasoning mode: `enabled`, `disabled`, or `auto` | | `webhook_url` | `string` (uri) | No | - | Optional callback URL that receives task status notifications | ## Frequently Asked Questions PixVerse V6 introduces flexible duration (any integer from `1` to `15` seconds instead of fixed `5`, `8`, or `10`) and three additional aspect ratios: `portrait_2_3`, `landscape_3_2`, and `cinematic_21_9`. Native synchronized audio, multi-clip output, and prompt reasoning work the same as in V5.5. PixVerse V6 accepts any integer duration from `1` to `15` seconds, with a default of `5`. You are no longer limited to fixed steps, so you can match the exact shot length you need. On top of `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, and `social_story_9_16`, PixVerse V6 adds `portrait_2_3`, `landscape_3_2`, and `cinematic_21_9` for portrait, landscape, and cinematic framing. Set `generate_audio_switch` to `true` and PixVerse V6 produces synchronized audio (background music, sound effects, or dialogue) together with the video in a single request. No separate audio call is required. `thinking_type` controls prompt reasoning. `enabled` (default) rewrites the prompt automatically for better results, `disabled` uses the prompt exactly as written, and `auto` lets the model decide based on the input. Rate limits and pricing depend on your subscription tier. See [Rate Limits](/ratelimits) and the [Pricing page](/pricing) for current values. ## Best practices * **Duration selection**: pick the shortest duration that conveys your shot to reduce processing time and cost; reserve 10-15s for narrative sequences * **Aspect ratio**: use `cinematic_21_9` for film-style shots and `social_story_9_16` or `portrait_2_3` for vertical social formats * **Audio**: enable `generate_audio_switch` only when you want the model to author audio; otherwise leave it `false` and mix your own track externally * **Prompt reasoning**: leave `thinking_type` as `enabled` for general prompts; switch to `disabled` when you need literal prompt adherence * **Production integration**: use `webhook_url` instead of polling for scalable workflows * **Error handling**: implement retry with exponential backoff for 503 responses ## Related APIs * **PixVerse V5.5 Text to Video**: prior version with fixed `5`/`8`/`10` second durations and five aspect ratios. See the [PixVerse V5.5 Text to Video overview](/api-reference/text-to-video/pixverse-v5-5/overview). * **PixVerse V5 Text to Video**: earlier version without audio, multi-clip, or 10-second support. See the [PixVerse V5 Text to Video overview](/api-reference/text-to-video/pixverse-v5/overview). * **PixVerse V6 (image-to-video)**: animate an existing image or interpolate a first/last frame transition. See the [PixVerse V6 overview](/api-reference/image-to-video/pixverse-v6/overview). # PixVerse V6 Text-to-Video - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v6/pixverse-v6-tasks get /v1/ai/text-to-video/pixverse-v6 List PixVerse-V6 text-to-video tasks. # PixVerse V6 Text-to-Video - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/pixverse-v6/task-by-id get /v1/ai/text-to-video/pixverse-v6/{task-id} # Create video from text - Veo 3.1 Source: https://docs.magnific.com/api-reference/text-to-video/post-veo-3-1 post /v1/ai/text-to-video/veo-3-1 Generate a video from text prompt using Google Veo 3.1 model. Supports multiple resolutions (720p, 1080p, 4K) and optional audio generation. # Create video from text - Veo 3.1 Fast Source: https://docs.magnific.com/api-reference/text-to-video/post-veo-3-1-fast post /v1/ai/text-to-video/veo-3-1-fast Generate a video from text prompt using Google Veo 3.1 Fast model. Faster generation at a lower cost. # Create video from text - Veo 3.1 Lite Source: https://docs.magnific.com/api-reference/text-to-video/post-veo-3-1-lite post /v1/ai/text-to-video/veo-3-1-lite Generate a video from a text prompt using Google Veo 3.1 Lite. A cost-efficient variant of Veo 3.1 supporting 720p and 1080p resolutions with optional native audio generation. # Veo 3.1 Lite – Text-to-Video API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-video/veo-3-1-lite/overview Generate cinematic videos from text prompts with Google Veo 3.1 Lite. A cost-efficient variant supporting 720p and 1080p resolutions with optional native audio. Cost-efficient text-to-video generation powered by Google's Veo 3.1 Lite model with 720p and 1080p output and optional audio synthesis. Veo 3.1 Lite is the cost-efficient variant of Google's Veo 3.1 text-to-video model. It transforms text prompts into high-quality cinematic videos at 720p or 1080p resolution, with durations of 4-8 seconds and optional native audio generation. Choose Veo 3.1 Lite when you want the Veo 3.1 look at a lower cost for prototyping, iteration, and high-volume workflows. ### Key capabilities * **Cost-efficient generation**: The lightweight Veo 3.1 variant for prototyping and high-volume workflows * **Multi-resolution output**: Generate videos in 720p or 1080p resolution * **Audio generation**: Optional synchronized audio for complete multimedia content * **Flexible durations**: 4, 6, or 8 second video outputs * **Aspect ratio control**: 16:9 (landscape) or 9:16 (portrait) formats * **Negative prompts**: Specify elements to avoid in generation * **Long prompts**: Up to 20,000 characters for detailed descriptions * **Reproducible results**: Seed parameter for consistent outputs ### Use cases * **Marketing content**: Create promotional videos and product showcases from descriptions * **Social media**: Generate engaging short-form content for TikTok, Reels, and Shorts * **Storytelling**: Visualize narratives, concepts, and creative ideas * **Advertising**: Produce video ads from text briefs * **Education**: Create visual explanations and demonstrations * **Prototyping**: Quickly visualize video concepts before committing to a higher-fidelity model ### Generate with Veo 3.1 Lite Submit a text prompt describing the video you want to create. The service returns a task ID for async polling or webhook notification.
Create a new text-to-video task List all Veo 3.1 Lite tasks Get task status by ID
### Parameters | Parameter | Type | Required | Description | | ----------------- | --------- | -------- | ------------------------------------------------------------ | | `prompt` | `string` | Yes | Text describing the video to generate (max 20,000 chars) | | `negative_prompt` | `string` | No | Text describing what to avoid in the video (max 2,000 chars) | | `duration` | `integer` | No | Video duration: `4`, `6`, or `8` seconds (default: `8`) | | `resolution` | `string` | No | Output resolution: `"720p"` or `"1080p"` (default: `"720p"`) | | `aspect_ratio` | `string` | No | Video format: `"16:9"` or `"9:16"` (default: `"16:9"`) | | `generate_audio` | `boolean` | No | Generate synchronized audio (default: `true`) | | `seed` | `integer` | No | Random seed for reproducibility (0-4294967295) | | `webhook_url` | `string` | No | URL for task completion notification | ## Frequently Asked Questions Veo 3.1 Lite is the cost-efficient variant of Veo 3.1. It supports 720p and 1080p resolutions (Veo 3.1 also offers 4K) and is optimized for lower-cost, high-volume generation. Use Veo 3.1 Lite for prototyping and iteration, and Veo 3.1 Standard when you need maximum fidelity or 4K output. Veo 3.1 Lite supports **720p** (HD) and **1080p** (Full HD). For 4K output, use the standard Veo 3.1 model. When `generate_audio` is set to `true` (the default), Veo 3.1 Lite synthesizes synchronized audio that matches the visual content. This includes ambient sounds, music, and effects appropriate to the scene described in your prompt. Generation time varies based on resolution, duration, and server load. Use webhooks for production workflows to avoid polling. Effective prompts describe the scene, subjects, motion, camera movement, lighting, and atmosphere. Be specific about visual details. Example: *"A serene mountain landscape at sunset with clouds moving slowly across an orange sky, camera slowly panning right"* Yes, set `aspect_ratio` to `"9:16"` to generate vertical videos optimized for TikTok, Instagram Reels, and YouTube Shorts. ## Best practices * **Prompt detail**: Provide specific descriptions of scenes, subjects, motion, and atmosphere for best results * **Negative prompts**: Use `negative_prompt` to exclude unwanted elements like "blurry, low quality, distorted" * **Resolution selection**: Choose 720p for fast iteration and 1080p for social media and final delivery * **Audio control**: Disable `generate_audio` if you plan to add custom audio in post-production * **Webhook integration**: Use webhooks instead of polling for scalable production applications * **Seed for consistency**: Use the same seed with identical parameters to reproduce similar results ## Related APIs * **[Veo 3.1 Text-to-Video](/api-reference/text-to-video/veo-3-1/overview)**: The standard Veo 3.1 model with 4K support and Fast mode * **[Veo 3.1 Lite Image-to-Video](/api-reference/image-to-video/veo-3-1-lite/overview)**: Generate videos from an image with Veo 3.1 Lite * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Alternative image-to-video with motion control * **[WAN 2.6](/api-reference/text-to-video/wan-v2-6-720p)**: Text-to-video at 720p and 1080p resolutions # Veo 3.1 – Text-to-Video API | Magnific API Source: https://docs.magnific.com/api-reference/text-to-video/veo-3-1/overview Generate cinematic videos from text prompts with Google Veo 3.1. Supports 720p, 1080p, and 4K resolutions with optional audio. Standard and Fast modes available. State-of-the-art text-to-video generation powered by Google's Veo 3.1 model with multi-resolution output and optional audio synthesis. Veo 3.1 is Google's advanced text-to-video model that transforms text prompts into high-quality cinematic videos. It supports resolutions up to 4K, durations of 4-8 seconds, and optional audio generation for complete multimedia output. Choose between **Standard** mode for maximum quality or **Fast** mode for quicker generation at lower cost. ### Key capabilities * **Multi-resolution output**: Generate videos in 720p, 1080p, or 4K resolution * **Audio generation**: Optional synchronized audio for complete multimedia content * **Flexible durations**: 4, 6, or 8 second video outputs * **Aspect ratio control**: 16:9 (landscape) or 9:16 (portrait) formats * **Negative prompts**: Specify elements to avoid in generation * **Two generation modes**: Standard (highest quality) and Fast (cost-efficient) * **Long prompts**: Up to 20,000 characters for detailed descriptions * **Reproducible results**: Seed parameter for consistent outputs ### Use cases * **Marketing content**: Create promotional videos and product showcases from descriptions * **Social media**: Generate engaging short-form content for TikTok, Reels, and Shorts * **Storytelling**: Visualize narratives, concepts, and creative ideas * **Advertising**: Produce video ads from text briefs * **Education**: Create visual explanations and demonstrations * **Prototyping**: Quickly visualize video concepts before production ### Generate with Veo 3.1 Standard High-quality video generation with maximum fidelity and detail.
Create a new text-to-video task List all Veo 3.1 tasks Get task status by ID
### Generate with Veo 3.1 Fast Faster generation at reduced cost, ideal for prototyping and high-volume workflows.
Create a fast text-to-video task List all Veo 3.1 Fast tasks Get task status by ID
### Parameters | Parameter | Type | Required | Description | | ----------------- | --------- | -------- | --------------------------------------------------------------------- | | `prompt` | `string` | Yes | Text describing the video to generate (max 20,000 chars) | | `negative_prompt` | `string` | No | Text describing what to avoid in the video | | `duration` | `integer` | No | Video duration: `4`, `6`, or `8` seconds (default: `8`) | | `resolution` | `string` | No | Output resolution: `"720p"`, `"1080p"`, or `"4k"` (default: `"720p"`) | | `aspect_ratio` | `string` | No | Video format: `"16:9"` or `"9:16"` (default: `"16:9"`) | | `generate_audio` | `boolean` | No | Generate synchronized audio (default: `true`) | | `seed` | `integer` | No | Random seed for reproducibility | | `webhook_url` | `string` | No | URL for task completion notification | ## Frequently Asked Questions **Standard mode** produces the highest quality output with maximum detail and fidelity, ideal for final production content. **Fast mode** generates videos more quickly at a lower cost, making it perfect for prototyping, testing prompts, and high-volume workflows where speed matters more than maximum quality. Veo 3.1 supports three resolution options: **720p** (HD), **1080p** (Full HD), and **4K** (Ultra HD). Higher resolutions provide more detail but may take longer to generate. When `generate_audio` is set to `true` (the default), Veo 3.1 synthesizes synchronized audio that matches the visual content. This includes ambient sounds, music, and effects appropriate to the scene described in your prompt. Generation time varies based on resolution, duration, and server load. Fast mode typically completes in less time than Standard mode. Use webhooks for production workflows to avoid polling. Effective prompts describe the scene, subjects, motion, camera movement, lighting, and atmosphere. Be specific about visual details. Example: *"A serene mountain landscape at sunset with clouds moving slowly across an orange sky, camera slowly panning right"* Yes, set `aspect_ratio` to `"9:16"` to generate vertical videos optimized for TikTok, Instagram Reels, and YouTube Shorts. ## Best practices * **Prompt detail**: Provide specific descriptions of scenes, subjects, motion, and atmosphere for best results * **Negative prompts**: Use `negative_prompt` to exclude unwanted elements like "blurry, low quality, distorted" * **Resolution selection**: Choose 720p for fast iteration, 1080p for social media, 4K for high-end production * **Audio control**: Disable `generate_audio` if you plan to add custom audio in post-production * **Webhook integration**: Use webhooks instead of polling for scalable production applications * **Seed for consistency**: Use the same seed with identical parameters to reproduce similar results ## Related APIs * **[Veo 3.1 Image-to-Video](/api-reference/image-to-video/veo-3-1/overview)**: Generate videos from images with Veo 3.1 * **[Veo 3.1 Reference-to-Video](/api-reference/reference-to-video/veo-3-1/overview)**: Generate videos with consistent characters using reference images * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Alternative image-to-video with motion control * **[WAN 2.6](/api-reference/text-to-video/wan-v2-6-720p)**: Text-to-video at 720p and 1080p resolutions # WAN 2.5 1080p - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-1080p post /v1/ai/text-to-video/wan-2-5-t2v-1080p Generate a 1080p video from text prompt using the WAN 2.5 model. # WAN 2.5 T2V 1080p - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-1080p-tasks get /v1/ai/text-to-video/wan-2-5-t2v-1080p Get the list of WAN 2.5 Text-to-Video 1080p tasks # WAN 2.5 480p - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-480p post /v1/ai/text-to-video/wan-2-5-t2v-480p Generate a 480p video from text prompt using the WAN 2.5 model. # WAN 2.5 T2V 480p - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-480p-tasks get /v1/ai/text-to-video/wan-2-5-t2v-480p Get the list of WAN 2.5 Text-to-Video 480p tasks # WAN 2.5 720p - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-720p post /v1/ai/text-to-video/wan-2-5-t2v-720p Generate a 720p video from text prompt using the WAN 2.5 model. # WAN 2.5 T2V 720p - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-5-t2v-720p-tasks get /v1/ai/text-to-video/wan-2-5-t2v-720p Get the list of WAN 2.5 Text-to-Video 720p tasks # WAN 2.7 - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-7/generate post /v1/ai/text-to-video/wan-2-7 Generate a video from a text prompt using WAN 2.7. Supports configurable aspect ratios (16:9, 9:16, 1:1, 4:3, 3:4), resolutions (720P, 1080P), optional audio input, and duration from 2 to 15 seconds. **Key features:** - 720P and 1080P resolution support - 5 aspect ratio options - Optional audio-guided generation - Automatic prompt expansion for richer output - Duration range: 2-15 seconds # WAN 2.7 Text To Video API Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-7/overview Generate AI videos from text prompts with WAN 2.7. 720P and 1080P resolution, 5 aspect ratios, 2-15 second durations, audio-guided generation, and prompt expansion for cinematic results. WAN 2.7 is Alibaba's latest video generation model, delivering cinematic motion, high visual fidelity, audio-guided generation, and automatic prompt expansion across 5 aspect ratios. WAN 2.7 Text-to-Video is an AI video generation API that creates MP4 videos from text descriptions. It produces smooth, high-fidelity video with cinematic motion at 720P (1280x720) or 1080P (1920x1080) resolution. The model supports durations from 2 to 15 seconds, 5 configurable aspect ratios, optional audio input for sound-guided generation, and automatic prompt expansion for richer output. ### Key capabilities * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **5 aspect ratios**: `16:9` landscape, `9:16` portrait, `1:1` square, `4:3` standard, `3:4` standard portrait * **Flexible durations**: 2 to 15 seconds of video output * **Audio-guided generation**: Provide a WAV or MP3 audio file (2-30 seconds, max 15MB) to guide video creation * **Prompt expansion**: AI optimizer expands short prompts into detailed scripts for richer, more cinematic output * **Negative prompts**: Exclude unwanted elements like watermarks, blur, or distortion (max 500 characters) * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing videos**: Create product showcases and brand content from text descriptions * **Social media content**: Generate short-form videos for TikTok, Instagram, and YouTube in portrait or landscape * **Music visualization**: Use audio-guided generation to create videos synchronized with a soundtrack * **Concept visualization**: Transform ideas and scripts into motion for rapid prototyping * **Educational content**: Illustrate concepts with AI-generated video explanations * **Creative exploration**: Experiment with text prompts and aspect ratios for unique visual content ### API operations Generate videos by submitting a text prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new text-to-video generation task List all WAN 2.7 T2V tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------------------- | --------- | -------- | --------- | ----------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video to generate. Max 5000 characters | | `negative_prompt` | `string` | No | - | Elements to avoid (e.g., "blurry, watermark"). Max 500 characters | | `audio_url` | `string` | No | - | URL of audio file (WAV/MP3, 2-30s, max 15MB) to guide generation | | `aspect_ratio` | `string` | No | `"16:9"` | Output ratio: `"16:9"`, `"9:16"`, `"1:1"`, `"4:3"`, `"3:4"` | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `duration` | `integer` | No | `5` | Video length in seconds: 2 to 15 | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `additional_settings.prompt_extend` | `boolean` | No | `true` | Enable AI prompt expansion for richer output | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions WAN 2.7 Text-to-Video is an AI video generation API developed by Alibaba. You submit a text prompt describing your desired video, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The model generates MP4 video at 720P or 1080P resolution in durations from 2 to 15 seconds. WAN 2.7 supports 5 aspect ratios: `16:9` (landscape widescreen), `9:16` (portrait/mobile), `1:1` (square), `4:3` (standard landscape), and `3:4` (standard portrait). The default is `16:9`. Provide a WAV or MP3 audio file URL via the `audio_url` parameter. The audio must be 2-30 seconds long and under 15MB. WAN 2.7 uses the audio to guide the visual content and motion of the generated video. If no audio is provided, the model may auto-generate audio. Prompt expansion (`additional_settings.prompt_extend`) uses AI to transform short prompts into detailed video scripts before generation. It is enabled by default. Disable it when you need precise control over exactly what the model generates. Processing time depends on resolution, duration, and server load. Higher resolution (1080P) and longer durations take more time. For production workflows, use webhooks instead of polling for scalable integration. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. WAN 2.7 adds audio-guided generation, 5 aspect ratios (vs limited options in 2.6), extended duration range of 2-15 seconds, and higher prompt limits (5000 characters). WAN 2.6 offers multi-shot sequences. Choose WAN 2.7 for the latest capabilities and audio input support. ## Best practices * **Prompt writing**: Be specific about scenes, camera movements (zoom, pan, tilt), lighting, and atmosphere. Detailed prompts produce better results than vague descriptions. * **Audio input**: Use clean audio files with clear rhythm or speech for best audio-guided results. Ensure audio duration aligns with your target video duration. * **Negative prompts**: Always include common artifacts to avoid: "blurry, low quality, watermark, text, distortion, extra limbs" * **Duration selection**: Start with shorter durations (2-5 seconds) for quick iterations, then increase for final outputs. * **Prompt expansion**: Leave enabled (default) for short prompts. Disable for precise control over generation. * **Reproducibility**: Save the `seed` value from successful generations to recreate similar results. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[WAN 2.7 Image-to-Video](/api-reference/image-to-video/wan-2-7/overview)**: Animate images or extend existing videos with WAN 2.7 * **[WAN 2.7 Reference-to-Video](/api-reference/reference-to-video/wan-2-7/overview)**: Generate videos featuring characters from reference images or videos * **[WAN 2.6 Text-to-Video](/api-reference/wan-v2-6-i2v/overview)**: Previous WAN generation with multi-shot sequences * **[WAN 2.5 Text-to-Video](/api-reference/wan-2-5-t2v/overview)**: WAN 2.5 with 480p, 720p, and 1080p options # WAN 2.7 T2V - Get task status Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-7/task-by-id get /v1/ai/text-to-video/wan-2-7/{task-id} Retrieve the status and result of a specific WAN 2.7 text-to-video task by its ID. # WAN 2.7 T2V - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-2-7/wan-2-7-t2v-tasks get /v1/ai/text-to-video/wan-2-7 Retrieve the list of all WAN 2.7 text-to-video tasks for the authenticated user. # WAN 2.6 1080p - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-v2-6-1080p post /v1/ai/text-to-video/wan-v2-6-1080p Generate a 1080p video from text prompt using the WAN 2.6 model. # WAN 2.6 T2V 1080p - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-v2-6-1080p-tasks get /v1/ai/text-to-video/wan-v2-6-1080p Get the list of WAN 2.6 Text-to-Video 1080p tasks # WAN 2.6 720p - Create video from text Source: https://docs.magnific.com/api-reference/text-to-video/wan-v2-6-720p post /v1/ai/text-to-video/wan-v2-6-720p Generate a 720p video from text prompt using the WAN 2.6 model. # WAN 2.6 T2V 720p - List tasks Source: https://docs.magnific.com/api-reference/text-to-video/wan-v2-6-720p-tasks get /v1/ai/text-to-video/wan-v2-6-720p Get the list of WAN 2.6 Text-to-Video 720p tasks # Happy Horse 1.0 - Edit video Source: https://docs.magnific.com/api-reference/video-edit/happy-horse-1/generate post /v1/ai/video-edit/happy-horse-1 Edit an existing video with natural-language instructions using Happy Horse 1.0, Alibaba's AI video editing model, optionally guided by up to 5 reference images. **Editing modes:** - **Instruction-based editing**: Provide `prompt` with editing instructions (e.g., "change the sky to sunset") - **Reference-image editing**: Provide `prompt` and `image_urls`, then reference images from the prompt using `@Image1`..`@Image5` (1-based, in array order) **Key features:** - 720P and 1080P resolution support - Up to 5 reference images - Output is capped at 15 seconds - Configurable audio handling: `auto` (regenerate when relevant) or `origin` (preserve input audio) # Happy Horse 1.0 Video Edit - List tasks Source: https://docs.magnific.com/api-reference/video-edit/happy-horse-1/happy-horse-1-video-edit-tasks get /v1/ai/video-edit/happy-horse-1 Retrieve the list of all Happy Horse 1.0 video edit tasks for the authenticated user. # Happy Horse 1.0 Video Edit API Source: https://docs.magnific.com/api-reference/video-edit/happy-horse-1/overview Edit videos with natural-language instructions using Happy Horse 1.0. Up to 5 reference images via @Image1..@Image5 tokens. Origin or auto audio handling, 720P/1080P output. Happy Horse 1.0 Video Edit modifies an existing video using natural-language instructions, optionally guided by up to 5 reference images. Output is capped at 15 seconds. Happy Horse 1.0 Video Edit is an AI video editing API that transforms an existing video with natural-language instructions. Built by Alibaba ATH (Taotian Innovation Unit, led by Zhang Di), the model edits the input MP4/MOV video based on a text prompt and up to 5 optional reference images. Output is delivered at 720P (1280x720) or 1080P (1920x1080) and capped at 15 seconds. Audio handling is configurable: `auto` lets the model decide whether to regenerate audio based on the prompt; `origin` preserves the input video's original audio. ### Key capabilities * **Natural-language editing**: Describe the desired edit in a prompt up to 2500 characters * **Up to 5 reference images**: Guide the edit with reference images via `@Image1`..`@Image5` tokens * **Audio control**: `auto` (model decides) or `origin` (preserve input audio) via `audio_setting` * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **Output cap**: Edited video is at most 15 seconds long, regardless of input length * **Wide input support**: MP4 or MOV with H.264 codec, 3 to 60 seconds, max 100 MB * **Reproducible results**: Fixed seed support (0 to 2147483647) for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### How reference images work Provide up to 5 reference images in the `image_urls` array. Reference each one in the `prompt` using `@Image1`..`@Image5` tokens (1-based, in array order): **Example prompt:** > "Replace the background with the cityscape in @Image1 and change the protagonist's outfit to match @Image2" ### Use cases * **Background replacement**: Swap backgrounds for product, fashion, or marketing videos * **Style transfer**: Change the visual style or color grading of a clip * **Object replacement**: Replace specific elements while preserving motion and composition * **Costume/wardrobe edits**: Modify outfits or props using reference images * **Visual effects**: Add or remove elements with natural-language instructions * **Quick iterations**: Rapidly produce variants of a video without re-shooting ### API operations Edit videos by submitting a video URL, prompt, and optional reference images. The service returns a task ID for async polling or webhook notification.
Create a new video edit task List all Happy Horse 1.0 Video Edit tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | --------------- | --------- | -------- | --------- | ---------------------------------------------------------------------------------------------------------------------------------------------- | | `video_url` | `string` | Yes | - | URL of the source video (MP4/MOV with H.264, 3-60s, longer side max 2160 px, shorter side min 320 px, aspect ratio 1:2.5 to 2.5:1, max 100 MB) | | `prompt` | `string` | Yes | - | Editing instructions. Reference images via `@Image1`..`@Image5`. Max 2500 characters | | `image_urls` | `array` | No | - | Up to 5 reference images (JPEG/JPG/PNG/WEBP, min 300 px, max 10 MB each) | | `resolution` | `string` | No | `"1080P"` | Output resolution: `"720P"` or `"1080P"` | | `audio_setting` | `string` | No | `"auto"` | Audio handling: `"auto"` (model decides) or `"origin"` (preserve input audio) | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions Happy Horse 1.0 Video Edit is an AI model from Alibaba ATH that edits an existing video based on natural-language instructions. You provide a video URL, a prompt describing the desired changes, and optionally up to 5 reference images. The model returns a new video at 720P or 1080P, capped at 15 seconds. Happy Horse 1.0 Video Edit accepts MP4 and MOV files with H.264 codec via publicly accessible URLs. Input duration must be between 3 and 60 seconds, frame rate greater than 8 fps, longer side at most 2160 px, shorter side at least 320 px, aspect ratio between 1:2.5 and 2.5:1, and max file size 100 MB. The model produces edits up to 15 seconds long regardless of input duration. If your input video is longer than 15 seconds, the output is truncated. Plan accordingly when selecting input clips. Use position-based tokens that match the order of the `image_urls` array: the 1st image is `@Image1`, the 2nd is `@Image2`, and so on up to `@Image5`. Example: "Replace the background with the scene in @Image1 and add the object from @Image2 in the foreground." `audio_setting` controls how the original audio track is handled. `"auto"` lets the model decide based on the prompt — if the prompt involves audio descriptions, the model may regenerate audio; otherwise the original audio is preserved. `"origin"` forces the original audio track to be preserved without regeneration. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Input video selection**: Use clips at or below 15 seconds to avoid truncation. Higher-quality inputs produce better edits. * **Prompt clarity**: Be explicit about which elements to change and which to keep ("Replace the background but keep the subject's pose and clothing"). * **Reference images**: Use clean reference images with the target element clearly visible. Match aspect ratio and lighting where possible. * **Audio choice**: Use `"origin"` when the input has dialogue or music you want preserved. Use `"auto"` when the prompt involves audio changes. * **Iterate with seed**: Save the `seed` from successful edits to recreate similar results. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[Happy Horse 1.0 Text-to-Video](/api-reference/text-to-video/happy-horse-1/overview)**: Generate videos from text prompts * **[Happy Horse 1.0 Image-to-Video](/api-reference/image-to-video/happy-horse-1/overview)**: Animate a single image into video * **[Happy Horse 1.0 Reference-to-Video](/api-reference/reference-to-video/happy-horse-1/overview)**: Generate videos from up to 9 character reference images # Happy Horse 1.0 Video Edit - Get task status Source: https://docs.magnific.com/api-reference/video-edit/happy-horse-1/task-by-id get /v1/ai/video-edit/happy-horse-1/{task-id} Retrieve the status and result of a specific Happy Horse 1.0 video edit task by its ID. # WAN 2.7 - Edit video Source: https://docs.magnific.com/api-reference/video-edit/wan-2-7/generate post /v1/ai/video-edit/wan-2-7 Edit an existing video using WAN 2.7. Transform video content with text instructions, apply visual references, or transfer artistic styles. **Three editing modes:** - **Instruction-based editing**: Provide `prompt` with editing instructions (e.g., "change the sky to sunset") - **Reference-image editing**: Provide `prompt` and `image_urls` to apply visual references to the video - **Style transfer**: Provide `prompt` and `image_urls` with style reference images to transfer artistic styles **Key features:** - 720P and 1080P resolution support - Up to 3 reference images for editing or style transfer - Set `duration` to `0` to preserve the full input video duration - Configurable audio handling: auto-generate or preserve original # WAN 2.7 Video Edit API Source: https://docs.magnific.com/api-reference/video-edit/wan-2-7/overview Edit existing videos with WAN 2.7 using text instructions, reference images, or style transfer. 720P and 1080P output, up to 3 reference images, and configurable audio handling. WAN 2.7 Video Edit transforms an existing video with natural-language instructions, reference images, or artistic style transfer while preserving the source footage. WAN 2.7 Video Edit is an AI video editing API that transforms an existing video based on a text prompt. You provide an input video via `video_url` and describe the desired changes, optionally guiding the edit with up to 3 reference images. The model supports three editing modes: instruction-based editing, reference-image editing, and style transfer. Output is available at 720P (1280x720) or 1080P (1920x1080) resolution, and you can preserve the full input duration or cap it at up to 10 seconds. ### Editing modes | Mode | Parameters | Use case | | ----------------------------- | -------------------------------------------------- | ------------------------------------------------------------------ | | **Instruction-based editing** | `video_url` + `prompt` | Change scene elements with text (e.g., "change the sky to sunset") | | **Reference-image editing** | `video_url` + `prompt` + `image_urls` | Apply visual references from images to the video | | **Style transfer** | `video_url` + `prompt` + `image_urls` (style refs) | Transfer artistic styles from reference images | ### Key capabilities * **Instruction-based editing**: Transform video content with natural-language prompts up to 5,000 characters * **Reference-image editing**: Guide edits with up to 3 reference images (JPEG, PNG, BMP, WEBP) * **Style transfer**: Apply artistic styles from reference images to the source video * **Resolution options**: 720P (1280x720) and 1080P (1920x1080) output * **5 aspect ratios**: `16:9`, `9:16`, `1:1`, `4:3`, `3:4` (or preserve the input video's ratio) * **Duration control**: Set `duration` to `0` to preserve the full input video, or cap output at up to 10 seconds * **Audio handling**: `auto` to generate/adapt audio, or `origin` to preserve the original track * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Scene retiming and relighting**: Change time of day, weather, or lighting with a text instruction * **Background replacement**: Swap or restyle backgrounds while keeping subjects intact * **Brand styling**: Apply a consistent artistic look across video assets with style transfer * **Reference-guided edits**: Match a target look supplied through reference images * **Content repurposing**: Reframe or restyle existing footage for different channels ### Edit videos with WAN 2.7 Edit a video by submitting the input video URL and a prompt to the API. The service returns a task ID for async polling or webhook notification.
Create a new video edit task List all WAN 2.7 video edit tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------------------- | --------- | -------- | ----------- | ------------------------------------------------------------------------------------------------------ | | `video_url` | `string` | Yes | - | URL of the input video to edit. MP4/MOV, 2-10s, 240-4096px per side, max 100MB | | `prompt` | `string` | No | - | Editing instructions describing the desired changes. Max 5,000 characters | | `negative_prompt` | `string` | No | - | Elements to avoid (e.g., "watermark, text, distortion, blurry"). Max 500 characters | | `image_urls` | `array` | No | - | Reference images for editing or style transfer (up to 3). JPEG/PNG/BMP/WEBP, 240-8000px, max 20MB each | | `aspect_ratio` | `string` | No | Input ratio | Output ratio: `16:9`, `9:16`, `1:1`, `4:3`, `3:4` | | `resolution` | `string` | No | `1080P` | Output resolution: `720P` or `1080P` | | `duration` | `integer` | No | `0` | Output length in seconds (0-10). Set to `0` to preserve the full input video duration | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647) | | `additional_settings.audio_setting` | `string` | No | `auto` | Audio handling: `auto` (generate/adapt) or `origin` (preserve original) | | `additional_settings.prompt_extend` | `boolean` | No | `true` | Enable AI prompt expansion for richer edits | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions WAN 2.7 Video Edit is an AI video editing API developed by Alibaba. You provide an input video URL and a text prompt describing the changes you want. The model applies those edits and returns a task ID immediately, then you poll for results or receive a webhook notification when processing completes. Three modes: instruction-based editing (`prompt` only), reference-image editing (`prompt` + `image_urls`), and style transfer (`prompt` + style reference `image_urls`). All three require an input `video_url`. Input videos must be MP4 or MOV at a publicly accessible URL, 2-10 seconds long, 240-4096px per side with an aspect ratio between 1:8 and 8:1, and a maximum file size of 100MB. Set `duration` to `0` to preserve the full duration of the input video. To cap the output, set `duration` to any value from 1 to 10 seconds. Yes. Set `additional_settings.audio_setting` to `origin` to preserve the input video's audio, or use `auto` (the default) to automatically generate or adapt audio for the edited video. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. See the [Pricing page](/pricing) for current rates and subscription options. ## Best practices * **Clear instructions**: Describe the exact change you want ("replace the background with a snowy mountain range") for predictable edits. * **Reference images**: Provide clean, high-resolution reference images when using reference-image editing or style transfer. * **Duration**: Start with `duration: 0` to preserve the source, then cap the length only when needed. * **Audio**: Use `origin` when the original audio must be retained (e.g., dialogue); use `auto` when the edit changes the scene significantly. * **Negative prompts**: Include: "watermark, text, distortion, blurry, extra limbs" * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[WAN 2.7 Text-to-Video](/api-reference/text-to-video/wan-2-7/overview)**: Generate videos from text prompts * **[WAN 2.7 Image-to-Video](/api-reference/image-to-video/wan-2-7/overview)**: Animate images or extend existing videos * **[WAN 2.7 Reference-to-Video](/api-reference/reference-to-video/wan-2-7/overview)**: Generate videos featuring characters from reference images or videos # WAN 2.7 Video Edit - Get task status Source: https://docs.magnific.com/api-reference/video-edit/wan-2-7/task-by-id get /v1/ai/video-edit/wan-2-7/{task-id} Retrieve the status and result of a specific WAN 2.7 video edit task by its ID. # WAN 2.7 Video Edit - List tasks Source: https://docs.magnific.com/api-reference/video-edit/wan-2-7/wan-2-7-video-edit-tasks get /v1/ai/video-edit/wan-2-7 Retrieve the list of all WAN 2.7 video edit tasks for the authenticated user. # Kling 4K I2V - List tasks Source: https://docs.magnific.com/api-reference/video/get-kling-4k-i2v get /v1/ai/video/kling-4k-i2v Retrieve the list of all Kling 4K Image-to-Video tasks for the authenticated user. # Kling 4K I2V - Get task status Source: https://docs.magnific.com/api-reference/video/get-kling-4k-i2v-task get /v1/ai/video/kling-4k-i2v/{task-id} Retrieve the status and result of a specific Kling 4K Image-to-Video task by its task ID. # Kling 4K T2V - List tasks Source: https://docs.magnific.com/api-reference/video/get-kling-4k-t2v get /v1/ai/video/kling-4k-t2v Retrieve the list of all Kling 4K Text-to-Video tasks for the authenticated user. # Kling 4K T2V - Get task status Source: https://docs.magnific.com/api-reference/video/get-kling-4k-t2v-task get /v1/ai/video/kling-4k-t2v/{task-id} Retrieve the status and result of a specific Kling 4K Text-to-Video task by its task ID. # Kling 2.6 Pro - Motion control video Source: https://docs.magnific.com/api-reference/video/kling-v2-6-motion-control-pro post /v1/ai/video/kling-v2-6-motion-control-pro Transfer motion from a reference video to a character image using Kling 2.6 Pro. The model preserves the character's appearance while applying motion patterns from the reference video. # Kling 2.6 Standard - Motion control video Source: https://docs.magnific.com/api-reference/video/kling-v2-6-motion-control-std post /v1/ai/video/kling-v2-6-motion-control-std Transfer motion from a reference video to a character image using Kling 2.6 Standard. The model preserves the character's appearance while applying motion patterns from the reference video. # Kling 2.6 Pro - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v2-6-motion-control-task-by-id get /v1/ai/image-to-video/kling-v2-6/{task-id} # Kling 2.6 Pro - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v2-6-motion-control-tasks get /v1/ai/image-to-video/kling-v2-6 # Kling 3 Pro - Motion control video Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/generate-pro post /v1/ai/video/kling-v3-motion-control-pro Transfer motion from a reference video to a character image using Kling 3 Pro. The model preserves the character's appearance while applying motion patterns from the reference video. # Kling 3 Standard - Motion control video Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/generate-std post /v1/ai/video/kling-v3-motion-control-std Transfer motion from a reference video to a character image using Kling 3 Standard. The model preserves the character's appearance while applying motion patterns from the reference video. # Kling 3 Motion Control API Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/overview Transfer motion from reference videos to character images with Kling 3 Motion Control. Preserves character appearance while applying motion patterns from 3-30 second reference videos. Transfer motion patterns from a reference video to a character image, preserving the character's appearance while applying realistic movement. Kling 3 Motion Control is a video generation API that transfers motion from a reference video to a character image. It preserves the character's appearance while applying the motion patterns from the reference video, producing realistic character animation. Available in Pro and Standard tiers, it supports reference videos from 3-30 seconds with optional text prompts for guiding the motion transfer. ### Key capabilities * **Motion transfer**: Extract motion from a reference video and apply it to a character image * **Character preservation**: Maintains the character's visual identity, face, and clothing throughout the video * **Orientation modes**: `video` mode for complex motions (up to 30 seconds output) or `image` mode for camera movements (up to 10 seconds output) * **Text guidance**: Optional prompt up to 2500 characters to refine the motion transfer * **CFG scale control**: Adjust prompt adherence from 0 (creative) to 1 (strict), default 0.5 * **Flexible input**: Reference videos from 3-30 seconds in MP4, MOV, WEBM, or M4V format * **Image support**: Character images in JPG, JPEG, PNG, or WEBP format (min 300x300px, max 10MB) * **Async processing**: Webhook notifications or polling for task completion ### Pro vs Standard | Feature | Kling 3 Motion Control Pro | Kling 3 Motion Control Standard | | -------- | ------------------------------ | ------------------------------- | | Quality | Higher fidelity, richer detail | Good quality, cost-effective | | Speed | Standard processing | Faster processing | | Best for | Premium content, marketing | High-volume, testing | ### Use cases * **Character animation**: Animate product mascots or brand characters with realistic dance or gesture movements * **Social media content**: Create engaging short videos by transferring trending motions to custom characters * **E-commerce**: Showcase apparel on virtual models by transferring real model movements * **Gaming and entertainment**: Generate character motion previews from reference footage * **Marketing campaigns**: Produce personalized character videos at scale with consistent branding ### Generate with Kling 3 Motion Control Create motion control videos by submitting a character image and reference video to the API. The service returns a task ID for async polling or webhook notification.
Generate motion control video with Pro quality Generate motion control video with Standard quality List all Motion Control Pro tasks Get Pro task status by ID List all Motion Control Standard tasks Get Standard task status by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------------- | -------- | -------- | ------- | ------------------------------------------------------------------------------------------------------------- | | `image_url` | `string` | Yes | - | Character/reference image URL. Min 300x300px, max 10MB. Formats: JPG, JPEG, PNG, WEBP. | | `video_url` | `string` | Yes | - | Reference video URL with the motion to transfer. Duration: 3-30 seconds. Formats: MP4, MOV, WEBM, M4V. | | `prompt` | `string` | No | - | Text prompt to guide motion transfer (max 2500 characters) | | `character_orientation` | `string` | No | `video` | Orientation mode: `video` (complex motions, up to 30s output) or `image` (camera movements, up to 10s output) | | `cfg_scale` | `number` | No | `0.5` | Prompt adherence: 0 (creative) to 1 (strict) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Kling 3 Motion Control is an AI video generation API that transfers motion from a reference video to a character image. You provide a character image and a reference video containing the desired motion. The model extracts the motion patterns and applies them to the character while preserving its visual appearance. Processing is asynchronous: you receive a task ID immediately, then poll for results or receive a webhook notification. The `character_orientation` parameter controls how the model interprets spatial information. In `video` mode (default), orientation matches the reference video, which works better for complex motions and supports output up to 30 seconds. In `image` mode, orientation matches the character image, which is better for following camera movements but limits output to 10 seconds maximum. Reference videos must be publicly accessible URLs in MP4, MOV, WEBM, or M4V format with a duration between 3 and 30 seconds. Character images support JPG, JPEG, PNG, and WEBP formats with minimum 300x300 pixel resolution and maximum 10MB file size. Motion control tasks have dedicated retrieval endpoints for each tier. Use `GET /v1/ai/video/kling-v3-motion-control-pro` or `GET /v1/ai/video/kling-v3-motion-control-std` to list tasks, and append `/{task-id}` to get a specific task's status and results. CFG scale controls how closely the model follows your text prompt during motion transfer. Use 0 for maximum creative freedom, 0.5 (default) for balanced results, or 1 for strict adherence to the prompt. When no prompt is provided, cfg\_scale has minimal effect. Pro delivers higher fidelity with richer detail, ideal for premium content and marketing videos. Standard offers good quality with faster processing, suitable for high-volume generation and testing. Both tiers share the same parameters and capabilities. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and quotas. Pricing varies based on model tier (Pro vs Standard) and output duration. See the [Pricing](/pricing) page for current rates. ## Best practices * **Character image quality**: Use high-resolution images with clear subjects and good lighting for best character preservation * **Reference video selection**: Choose reference videos with clear, well-defined motions. Avoid heavily occluded or blurry footage. * **Orientation mode**: Use `video` mode for dance or full-body motions; use `image` mode when camera movement matters more than body motion * **Prompt guidance**: Add a text prompt to refine the motion transfer, especially when the reference video has ambiguous movements * **CFG tuning**: Start with 0.5, decrease for more creative interpretation, increase for strict prompt following * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Kling 3](/api-reference/video/kling-v3/overview)**: Text-to-video and image-to-video generation with multi-shot support * **[Kling 3 Omni](/api-reference/video/kling-v3-omni/overview)**: Multi-modal video generation with reference images and video references * **[Kling 2.6 Motion Control](/api-reference/video/kling-v2-6-motion-control-pro)**: Previous generation motion control * **[VFX](/api-reference/video/vfx/overview)**: Apply visual effects to generated videos # Kling 3 Pro Motion Control - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/pro-task-by-id get /v1/ai/video/kling-v3-motion-control-pro/{task-id} Retrieve the status and result of a specific Kling 3 Pro Motion Control video generation task by its task ID. # Kling 3 Pro Motion Control - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/pro-tasks get /v1/ai/video/kling-v3-motion-control-pro Retrieve the list of all Kling 3 Pro Motion Control video generation tasks for the authenticated user. # Kling 3 Standard Motion Control - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/std-task-by-id get /v1/ai/video/kling-v3-motion-control-std/{task-id} Retrieve the status and result of a specific Kling 3 Standard Motion Control video generation task by its task ID. # Kling 3 Standard Motion Control - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v3-motion-control/std-tasks get /v1/ai/video/kling-v3-motion-control-std Retrieve the list of all Kling 3 Standard Motion Control video generation tasks for the authenticated user. # Kling 3 Omni Pro - Generate video from text or image Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/generate-pro post /v1/ai/video/kling-v3-omni-pro Generate AI video using Kling 3 Omni Pro with advanced multi-modal capabilities. **Features:** - **Text-to-video**: Generate videos from text prompts - **Image-to-video**: Use start and/or end frame images to guide generation - **Multi-shot**: Create videos with up to 6 shots (max 15s total) - **Element control**: Include reference images for consistent character/style **Duration:** 3-15 seconds **Quality:** Pro mode offers highest quality output. **Note:** For video-to-video generation using a reference video, use the `/ai/reference-to-video/kling-v3-omni-pro` endpoint instead. # Kling 3 Omni Pro - Video-to-video generation Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/generate-pro-video-reference post /v1/ai/reference-to-video/kling-v3-omni-pro Generate AI video using Kling 3 Omni Pro with a reference video for motion and style guidance. **Video-to-video mode:** This endpoint requires a `video_url` parameter. Reference the video in your prompt using `@Video1`. **Features:** - Use a reference video (3-10s) to guide motion and style - Combine with an image for start frame control - High-quality pro output **Use case:** Create videos that follow motion patterns from a reference video while applying your creative prompt. **Duration:** 3-15 seconds **Quality:** Pro mode offers highest quality output. **Tip:** For text-to-video or image-to-video without a reference video, use the `/ai/video/kling-v3-omni-pro` endpoint instead. # Kling 3 Omni Standard - Generate video from text or image Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/generate-std post /v1/ai/video/kling-v3-omni-std Generate AI video using Kling 3 Omni Standard with advanced multi-modal capabilities. **Features:** - **Text-to-video**: Generate videos from text prompts - **Image-to-video**: Use start and/or end frame images to guide generation - **Multi-shot**: Create videos with up to 6 shots (max 15s total) - **Element control**: Include reference images for consistent character/style **Duration:** 3-15 seconds **Quality:** Standard mode offers faster generation at slightly lower quality. **Note:** For video-to-video generation using a reference video, use the `/ai/reference-to-video/kling-v3-omni-std` endpoint instead. # Kling 3 Omni Standard - Video-to-video generation Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/generate-std-video-reference post /v1/ai/reference-to-video/kling-v3-omni-std Generate AI video using Kling 3 Omni Standard with a reference video for motion and style guidance. **Video-to-video mode:** This endpoint requires a `video_url` parameter. Reference the video in your prompt using `@Video1`. **Features:** - Use a reference video (3-10s) to guide motion and style - Combine with an image for start frame control - Faster generation at slightly lower quality **Use case:** Create videos that follow motion patterns from a reference video while applying your creative prompt. **Duration:** 3-15 seconds **Quality:** Standard mode offers faster generation at slightly lower quality. **Tip:** For text-to-video or image-to-video without a reference video, use the `/ai/video/kling-v3-omni-std` endpoint instead. # Kling 3 Omni - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/kling-v3-omni-tasks get /v1/ai/video/kling-v3-omni Retrieve the list of all Kling 3 Omni video generation tasks for the authenticated user. # Kling 3 Omni API Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/overview Generate AI videos with multi-modal capabilities using Kling 3 Omni. Supports text-to-video and image-to-video with Pro and Standard tiers. Generate AI videos with multi-modal inputs: text prompts and reference images. Kling 3 Omni is a versatile video generation API that supports multiple input modes: text-to-video and image-to-video with reference images. It offers advanced features like multi-shot mode for scene-by-scene control and reference images for style guidance. **Looking for video-to-video?** Use the dedicated [Reference-to-Video endpoints](#video-to-video-with-reference-video) to generate videos from a reference video using `video_url`. ### Key capabilities * **Text-to-video**: Generate videos from text prompts up to 2500 characters * **Image-to-video**: Use `image_url` for start frame, `end_image_url` for end frame control * **Reference images**: Add style guidance with `image_urls`, reference as `@Image1`, `@Image2` in prompts * **Multi-shot mode**: Create multi-scene videos with `multi_prompt` for shot-by-shot control * **Duration control**: Generate videos from 3-15 seconds * **Audio options**: Generate native audio or use voice IDs for narration * **Async processing**: Webhook notifications or polling for task completion ### Generation modes | Mode | Parameters | Use case | | ---------------------- | ----------------------- | ------------------------------------ | | **Text-to-video** | `prompt` (required) | Generate video from text description | | **Image-to-video** | `image_url` + `prompt` | Animate a starting image | | **Reference-to-video** | `image_urls` + `prompt` | Maintain style consistency | ### Pro vs Standard | Feature | Kling 3 Omni Pro | Kling 3 Omni Standard | | -------- | ---------------------- | ---------------------------- | | Quality | Higher fidelity output | Good quality, cost-effective | | Speed | Standard processing | Faster processing | | Best for | Premium productions | Testing, high-volume | ### Use cases * **Character animation**: Animate a character image with controlled, consistent motion * **Product visualization**: Animate product images with controlled motion * **Storyboarding**: Create multi-scene videos with shot-by-shot prompts * **Style transfer**: Apply visual style from reference images to generated content * **Marketing content**: Generate promotional videos from brand imagery ### Generate videos with Kling 3 Omni Create videos by submitting prompts with optional images. The service returns a task ID for async polling or webhook notification.
Generate video with Kling 3 Omni Pro Generate video with Kling 3 Omni Standard List all Kling 3 Omni tasks Get task status by ID
### Video-to-video with reference video For video-to-video generation using a reference video, use the dedicated Reference-to-Video endpoints. These endpoints accept `video_url` and let you reference the video in your prompt as `@Video1`.
Generate video from reference video (Pro) Generate video from reference video (Standard) List all reference-to-video tasks Get reference-to-video task by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------- | --------- | ----------- | ----------- | -------------------------------------------------------------- | | `prompt` | `string` | Conditional | - | Text prompt (max 2500 chars). Required for text-to-video. | | `image_url` | `string` | No | - | Start frame image URL for image-to-video | | `start_image_url` | `string` | No | - | Alternative start frame image | | `end_image_url` | `string` | No | - | End frame image URL | | `image_urls` | `array` | No | - | Reference images for style. Use `@Image1`, `@Image2` in prompt | | `multi_prompt` | `array` | No | - | Shot-by-shot prompts for multi-scene videos (max 6 shots) | | `shot_type` | `string` | No | `customize` | Multi-shot type (only `customize` supported) | | `aspect_ratio` | `string` | No | `16:9` | Video ratio: `16:9`, `9:16`, `1:1` | | `duration` | `integer` | No | `5` | Duration in seconds: 3-15 | | `generate_audio` | `boolean` | No | - | Generate native audio for the video | | `voice_ids` | `array` | No | - | Voice IDs for narration. Use `<<>>` in prompt | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Kling 3 Omni is a multi-modal video generation API that creates videos from text prompts, images, or a combination. It supports reference images for style guidance and multi-shot mode for scene-by-scene control. Video durations range from 3-15 seconds. For video-to-video generation (using a reference video), use the dedicated [Reference-to-Video endpoints](/api-reference/video/kling-v3-omni/generate-pro-video-reference). These endpoints accept a `video_url` parameter and let you reference the video in your prompt as `@Video1`. Multi-shot mode lets you create videos with multiple scenes, each with its own prompt. Provide an array of prompts via `multi_prompt` (max 6 shots). Each shot must be at least 3 seconds. The total duration is the sum of all shots. Images must be publicly accessible URLs in JPG, JPEG, or PNG format. Requirements: minimum 300x300 pixels, maximum 10MB file size. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and quotas. Pricing varies based on model tier (Pro vs Standard) and video duration. See the [Pricing](/pricing) page for current rates. ## Best practices * **Reference images**: Use clear, well-lit reference images. Reference them as `@Image1`, `@Image2` in your prompt for best results. * **Duration planning**: Start with 5-second videos to test, then increase duration for final output. * **Multi-shot flow**: Plan shot transitions carefully; each shot should have a coherent prompt. * **Audio options**: Use `generate_audio: true` for ambient sound, or `voice_ids` for narration. * **Production integration**: Use webhooks instead of polling for scalable applications. ## Related APIs * **[Kling 3](/api-reference/video/kling-v3/overview)**: Standard Kling 3 without Omni multi-modal features * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Previous generation with motion control * **[Runway Gen 4.5](/api-reference/video/runway-gen-4-5/overview)**: Alternative video generation model * **[VFX](/api-reference/video/vfx/overview)**: Apply visual effects to generated videos # Kling 3 Omni - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/task-by-id get /v1/ai/video/kling-v3-omni/{task-id} Retrieve the status and result of a specific Kling 3 Omni video generation task by its task ID. # Kling 3 Omni Reference-to-Video - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/video-reference-task-by-id get /v1/ai/reference-to-video/kling-v3-omni/{task-id} Retrieve the status and result of a specific Kling 3 Omni reference-to-video task (Pro or Standard) by its task ID. # Kling 3 Omni Reference-to-Video - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v3-omni/video-reference-tasks get /v1/ai/reference-to-video/kling-v3-omni Retrieve the list of all Kling 3 Omni reference-to-video tasks (both Pro and Standard) for the authenticated user. # Kling 3 Pro - Generate video Source: https://docs.magnific.com/api-reference/video/kling-v3/generate-pro post /v1/ai/video/kling-v3-pro Generate AI video using Kling 3 Pro with text-to-video or image-to-video capabilities. **Features:** - **Text-to-video**: Generate videos from text prompts - **Image-to-video**: Use start and/or end frame images to guide generation - **Multi-shot**: Create videos with up to 6 shots (max 15s total) - **Element control**: Include reference images for consistent character/style **Duration:** 3-15 seconds **Quality:** Pro mode offers highest quality output with longer processing time. # Kling 3 Standard - Generate video Source: https://docs.magnific.com/api-reference/video/kling-v3/generate-std post /v1/ai/video/kling-v3-std Generate AI video using Kling 3 Standard with text-to-video or image-to-video capabilities. **Features:** - **Text-to-video**: Generate videos from text prompts - **Image-to-video**: Use start and/or end frame images to guide generation - **Multi-shot**: Create videos with up to 6 shots (max 15s total) - **Element control**: Include reference images for consistent character/style **Duration:** 3-15 seconds **Quality:** Standard mode offers faster generation at slightly lower quality compared to Pro. # Kling 3 - List tasks Source: https://docs.magnific.com/api-reference/video/kling-v3/kling-v3-tasks get /v1/ai/video/kling-v3 Retrieve the list of all Kling 3 video generation tasks for the authenticated user. # Kling 3 API Source: https://docs.magnific.com/api-reference/video/kling-v3/overview Generate AI videos from text or images with Kling 3. Multi-shot support, first/end frame control, and durations from 3-15 seconds. Pro and Standard tiers for creative video production. Generate high-quality videos from text prompts or images using Kling's latest V3 model with multi-shot support and advanced frame control. Kling 3 is a dual-mode video generation API that creates professional-grade videos from either text descriptions or source images. It supports multi-shot mode for creating complex narratives with up to 6 scenes, first and end frame image control, and flexible durations from 3 to 15 seconds. Available in Pro and Standard tiers to balance quality and cost. ### Key capabilities * **Text-to-Video (T2V)**: Generate videos from text prompts up to 2500 characters * **Image-to-Video (I2V)**: Use first\_frame and/or end\_frame images to control video start and end points * **Multi-shot mode**: Create videos with up to 6 scenes, each with custom prompts and durations (max 15 seconds total) * **Flexible durations**: 3-15 seconds with per-shot duration control in multi-shot mode * **CFG scale control**: Adjust prompt adherence from 0 (creative) to 1 (strict), default 0.5 * **Negative prompts**: Exclude unwanted elements, styles, or artifacts * **Async processing**: Webhook notifications or polling for task completion ### Pro vs Standard | Feature | Kling 3 Pro | Kling 3 Standard | | -------- | ------------------------------ | ---------------------------- | | Quality | Higher fidelity, richer detail | Good quality, cost-effective | | Speed | Standard processing | Faster processing | | Best for | Premium content, marketing | High-volume, testing | ### Use cases * **Marketing and advertising**: Create multi-scene product narratives with consistent branding * **Social media content**: Generate vertical videos for TikTok, Instagram Reels, and YouTube Shorts * **E-commerce**: Animate product images with controlled start and end frames * **Storyboarding**: Turn scripts into multi-shot video sequences * **Creative storytelling**: Build narratives with scene-by-scene control ### Generate videos with Kling 3 Create videos by submitting a text prompt (T2V) or images with prompt (I2V) to the API. The service returns a task ID for async polling or webhook notification.
Generate video with Kling 3 Pro Generate video with Kling 3 Standard List all Kling 3 tasks Get task status by ID
### Parameters | Parameter | Type | Required | Default | Description | | ----------------- | --------- | -------- | ----------- | ------------------------------------------------------------------------------------------------- | | `prompt` | `string` | No | - | Text prompt describing the video (max 2500 chars). Required for T2V. | | `negative_prompt` | `string` | No | - | Text describing what to avoid (max 2500 chars) | | `image_list` | `array` | No | - | Reference images with `image_url` and `type` (first\_frame/end\_frame) | | `multi_shot` | `boolean` | No | `false` | Enable multi-shot mode for multi-scene videos | | `shot_type` | `string` | No | `customize` | Shot segmentation: `customize` (manual per-shot prompts) or `intelligence` (AI auto-segmentation) | | `multi_prompt` | `array` | No | - | Shot definitions: `index` (0-5), `prompt`, `duration` per scene | | `aspect_ratio` | `string` | No | `16:9` | Video ratio: `16:9`, `9:16`, `1:1` | | `duration` | `integer` | No | `5` | Duration in seconds: 3-15 (default 5) | | `cfg_scale` | `number` | No | `0.5` | Prompt adherence: 0 (creative) to 1 (strict) | | `webhook_url` | `string` | No | - | URL for task completion notification | ### Image list item | Field | Type | Description | | ----------- | -------- | ------------------------------------------------------------------- | | `image_url` | `string` | Publicly accessible image URL (300x300 min, 10MB max, JPG/JPEG/PNG) | | `type` | `string` | Image role: `first_frame` or `end_frame` | ### Multi-prompt item | Field | Type | Description | | ---------- | --------- | ------------------------------------------ | | `index` | `integer` | Shot order (0-5) | | `prompt` | `string` | Text prompt for this shot (max 2500 chars) | | `duration` | `number` | Shot duration in seconds | ## Frequently Asked Questions Kling 3 is an AI video generation model that creates videos from text prompts (T2V) or images (I2V). You submit your request via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. Typical generation takes 30-120 seconds depending on duration and complexity. Multi-shot mode lets you create videos with up to 6 distinct scenes. Each scene can have its own prompt and duration. The total duration across all shots cannot exceed 15 seconds, and each shot must be at least 3 seconds. Enable with `multi_shot: true` and define scenes in `multi_prompt`. Use the `image_list` parameter to provide reference images. Set `type: "first_frame"` to use an image as the video's starting point, or `type: "end_frame"` for the ending point. You can use both to create a transition from one image to another. Kling 3 accepts JPG, JPEG, and PNG images via publicly accessible URLs. Requirements: minimum 300x300 pixels, maximum 10MB file size, aspect ratio between 1:2.5 and 2.5:1. CFG scale controls how closely the model follows your prompt. Use 0 for maximum creativity and artistic interpretation, 0.5 (default) for balanced results, or 1 for strict adherence to your prompt with less creative variation. Pro delivers higher fidelity with richer detail, ideal for premium content and marketing. Standard offers good quality with faster processing, suitable for high-volume generation and testing. Both share the same parameters and capabilities. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and quotas. Pricing varies based on model tier (Pro vs Standard) and video duration. See the [Pricing](/pricing) page for current rates. ## Best practices * **Prompt clarity**: Write detailed prompts specifying subject, action, camera movement, and atmosphere * **Start simple**: Begin with single-shot mode before attempting multi-shot sequences * **Image quality**: For I2V, use high-resolution source images with clear subjects (min 300x300) * **Duration planning**: For multi-shot, plan scene durations to stay within 15-second total limit * **CFG tuning**: Start with 0.5, decrease for more creativity, increase for prompt precision * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Kling 3 Omni](/api-reference/video/kling-v3-omni/overview)**: Kling 3 with video reference support for motion/style guidance * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Previous generation with motion control capabilities * **[Kling O1](/api-reference/image-to-video/kling-o1/overview)**: High-performance video generation * **[Runway Gen 4.5](/api-reference/video/runway-gen-4-5/overview)**: Alternative video generation model # Kling 3 - Get task status Source: https://docs.magnific.com/api-reference/video/kling-v3/task-by-id get /v1/ai/video/kling-v3/{task-id} Retrieve the status and result of a specific Kling 3 video generation task by its task ID. # OmniHuman 1.5 - Create human animation Source: https://docs.magnific.com/api-reference/video/omni-human-1-5 post /v1/ai/video/omni-human-1-5 Generate animated video of a human figure driven by audio using OmniHuman 1.5 model (ByteDance). Supports natural head movements, facial expressions, and body motion synchronized to audio input. # OmniHuman 1.5 - List tasks Source: https://docs.magnific.com/api-reference/video/omni-human-1-5-tasks get /v1/ai/video/omni-human-1-5 # OmniHuman 1.5 - Get task status Source: https://docs.magnific.com/api-reference/video/omni-human-1-5/task-by-id get /v1/ai/video/omni-human-1-5/{task-id} # Kling 4K I2V - Generate video from image Source: https://docs.magnific.com/api-reference/video/post-kling-4k-i2v post /v1/ai/video/kling-4k-i2v Generate AI video in 4K resolution from an image using Kling 4K Image-to-Video. **Features:** - **Image-to-video**: Use a reference image to guide 4K video generation - **End frame control**: Optionally specify an end frame image - **Motion brush**: Use static and dynamic masks for precise motion control **Duration:** 3-15 seconds **Resolution:** 4K output quality. # Kling 4K T2V - Generate video from text Source: https://docs.magnific.com/api-reference/video/post-kling-4k-t2v post /v1/ai/video/kling-4k-t2v Generate AI video in 4K resolution from a text prompt using Kling 4K Text-to-Video. **Features:** - **Text-to-video**: Generate 4K videos from text prompts - **Camera control**: Optionally specify camera movement type and configuration **Duration:** 3-15 seconds **Resolution:** 4K output quality. # RunWay Act Two Character Performance Source: https://docs.magnific.com/api-reference/video/runway-act-two post /v1/ai/video/runway-act-two Generate a character performance video using RunWay Act Two model. Transfer facial expressions and body movements from a reference video to a character image or video. # RunWay Act Two - List tasks Source: https://docs.magnific.com/api-reference/video/runway-act-two-tasks get /v1/ai/video/runway-act-two # RunWay Act Two - Get task status Source: https://docs.magnific.com/api-reference/video/runway-act-two/task-by-id get /v1/ai/video/runway-act-two/{task-id} # Create video from image - RunWay Gen 4.5 Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/generate-i2v post /v1/ai/image-to-video/runway-4-5 Generate high-quality videos from images using RunWay Gen 4.5 model. **Features:** - Transform static images into dynamic videos - Precise motion control via text prompts - Multiple aspect ratios for different use cases - Duration options: 5, 8, or 10 seconds - Reproducible results with seed parameter **Supported aspect ratios:** - `1280:720`: Landscape (16:9) - ideal for YouTube, presentations - `720:1280`: Portrait (9:16) - ideal for TikTok, Instagram Reels - `1104:832`: Landscape (4:3) - classic format - `960:960`: Square (1:1) - ideal for Instagram posts - `832:1104`: Portrait (3:4) - ideal for Pinterest **Use cases:** Product animations, social media content, marketing videos, and bringing photos to life. # Create video from text - RunWay Gen 4.5 Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/generate-t2v post /v1/ai/text-to-video/runway-4-5 Generate high-quality videos from text descriptions using RunWay Gen 4.5 model. **Features:** - State-of-the-art text-to-video generation - Multiple aspect ratios for different use cases - Duration options: 5, 8, or 10 seconds - High visual fidelity and motion quality **Supported aspect ratios:** - `1280:720`: Landscape (16:9) - ideal for YouTube, presentations - `720:1280`: Portrait (9:16) - ideal for TikTok, Instagram Reels - `1104:832`: Landscape (4:3) - classic format - `960:960`: Square (1:1) - ideal for Instagram posts - `832:1104`: Portrait (3:4) - ideal for Pinterest **Use cases:** Social media content, marketing videos, creative projects, and visual storytelling. # Runway Gen 4.5 API Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/overview Generate cinematic AI videos from text or images with Runway Gen 4.5. Supports 5, 8, and 10 second durations with multiple aspect ratios. Professional-grade video generation for creative projects. Generate high-quality videos from text prompts or images using Runway's latest Gen 4.5 model with cinematic quality and precise motion control. Runway Gen 4.5 is a dual-mode video generation API that creates professional-grade videos from either text descriptions or source images. It delivers cinematic quality with natural motion, supporting durations of 5, 8, or 10 seconds and multiple aspect ratios including landscape (16:9), portrait (9:16), and square (1:1) formats. ### Key capabilities * **Text-to-Video (T2V)**: Generate videos from text descriptions up to 2000 characters * **Image-to-Video (I2V)**: Animate source images with motion guided by text prompts * **Flexible durations**: 5, 8, or 10 second video generation (default: 5 seconds) * **Multiple aspect ratios**: `1280:720` (16:9), `720:1280` (9:16), `1104:832` (4:3), `960:960` (1:1), `832:1104` (3:4) * **Reproducible results**: Seed parameter for consistent video generation with identical inputs * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing and advertising**: Create product showcase videos and promotional content from concepts * **Social media content**: Generate vertical videos for TikTok, Instagram Reels, and YouTube Shorts * **E-commerce**: Animate product images to create engaging video presentations * **Creative storytelling**: Transform storyboards and illustrations into video sequences * **Music videos**: Generate visual content from artistic concepts and mood descriptions ### Generate videos with Runway Gen 4.5 Create videos by submitting a text prompt (T2V) or image with prompt (I2V) to the API. The service returns a task ID for async polling or webhook notification.
Create video from text description Create video from source image List all text-to-video tasks List all image-to-video tasks Get T2V task status by ID Get I2V task status by ID
### Text-to-Video Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ---------- | ----------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video to generate. Maximum 2000 characters. | | `ratio` | `string` | No | `1280:720` | Aspect ratio: `1280:720`, `720:1280`, `1104:832`, `960:960`, `832:1104` | | `duration` | `integer` | No | `5` | Video duration in seconds: `5`, `8`, or `10` | | `webhook_url` | `string` | No | - | URL for task completion notification | ### Image-to-Video Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ---------- | ----------------------------------------------------------------------- | | `image` | `string` | Yes | - | Source image URL (HTTPS, publicly accessible) or Base64 encoded image | | `prompt` | `string` | Yes | - | Text description of the desired motion. Maximum 2000 characters. | | `ratio` | `string` | No | `1280:720` | Aspect ratio: `1280:720`, `720:1280`, `1104:832`, `960:960`, `832:1104` | | `duration` | `integer` | No | `5` | Video duration in seconds: `5`, `8`, or `10` | | `seed` | `integer` | No | - | Random seed for reproducible results (0-4294967295) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions Runway Gen 4.5 is an AI video generation model that creates high-quality videos from text prompts (T2V) or source images (I2V). You submit your request via the API, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. Typical generation takes 30-90 seconds depending on duration. Text-to-Video (T2V) generates videos entirely from text descriptions, creating both the visual content and motion. Image-to-Video (I2V) takes a source image and animates it based on your text prompt, preserving the original image's style and subject while adding motion. Runway Gen 4.5 accepts images via publicly accessible HTTPS URLs or Base64 encoded strings. Common formats like JPEG, PNG, and WebP are supported. Ensure your image URL is accessible without authentication. Processing time depends on the selected duration and current server load. Typical generation takes 30-90 seconds. Longer durations (10 seconds) may take slightly longer. Use webhooks for production workflows to receive notifications instead of polling. Runway Gen 4.5 supports five aspect ratios: `1280:720` (16:9 landscape), `720:1280` (9:16 portrait), `1104:832` (4:3 landscape), `960:960` (1:1 square), and `832:1104` (3:4 portrait). The default is `1280:720`. Yes, for Image-to-Video requests. Use the `seed` parameter with a value between 0 and 4294967295. Using the same seed with identical prompt and image will produce the same video. Note: seed is only available for I2V, not T2V. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits and quotas. Pricing varies based on video duration and model tier. See the [Pricing](/pricing) page for current rates and subscription options. ## Best practices * **Prompt clarity**: Write detailed, descriptive prompts specifying subject, action, lighting, and style for best results * **Image quality**: For I2V, use high-resolution source images with clear subjects for better animation * **Aspect ratio matching**: Match the aspect ratio to your target platform (9:16 for TikTok/Reels, 16:9 for YouTube) * **Duration selection**: Start with 5 seconds for testing, then increase to 8 or 10 for final production * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Runway Gen4 Turbo](/api-reference/image-to-video/runway-gen4-turbo)**: Faster video generation with the Gen4 Turbo model * **[Runway Act Two](/api-reference/video/runway-act-two)**: Character animation and acting performance videos * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Alternative video generation with advanced motion control * **[WAN 2.6](/api-reference/wan-v2-6-i2v/overview)**: High-resolution video generation up to 1080p # RunWay Gen 4.5 I2V - List tasks Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/runway-4-5-i2v-tasks get /v1/ai/image-to-video/runway-4-5 Retrieve the status of all RunWay Gen 4.5 image-to-video generation tasks for the authenticated user. # RunWay Gen 4.5 T2V - List tasks Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/runway-4-5-t2v-tasks get /v1/ai/text-to-video/runway-4-5 Retrieve the status of all RunWay Gen 4.5 text-to-video generation tasks for the authenticated user. # RunWay Gen 4.5 I2V - Get task status Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/task-by-id-i2v get /v1/ai/image-to-video/runway-4-5/{task-id} Retrieve the status and result of a specific RunWay Gen 4.5 image-to-video task by its task ID. # RunWay Gen 4.5 T2V - Get task status Source: https://docs.magnific.com/api-reference/video/runway-gen-4-5/task-by-id-t2v get /v1/ai/text-to-video/runway-4-5/{task-id} Retrieve the status and result of a specific RunWay Gen 4.5 text-to-video task by its task ID. # Seedance 2.5 Pro 1080p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/generate-1080p post /v1/ai/video/seedance-2-5-pro-1080p Generate 1080p video using Seedance 2.5 Pro. Supports text-to-video, image-to-video (first/last frame), and reference-to-video with up to 30 reference images (@Image1..@Image30) and 10 reference audios (@Audio1..@Audio10). Also supports camera motion, multishot sequences (up to 6 shots) and native audio (sound effects). # Seedance 2.5 Pro 480p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/generate-480p post /v1/ai/video/seedance-2-5-pro-480p Generate 480p video using Seedance 2.5 Pro. Supports text-to-video, image-to-video (first/last frame), and reference-to-video with up to 30 reference images (@Image1..@Image30) and 10 reference audios (@Audio1..@Audio10). Also supports camera motion, multishot sequences (up to 6 shots) and native audio (sound effects). # Seedance 2.5 Pro 720p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/generate-720p post /v1/ai/video/seedance-2-5-pro-720p Generate 720p video using Seedance 2.5 Pro. Supports text-to-video, image-to-video (first/last frame), and reference-to-video with up to 30 reference images (@Image1..@Image30) and 10 reference audios (@Audio1..@Audio10). Also supports camera motion, multishot sequences (up to 6 shots) and native audio (sound effects). # Seedance 2.5 Pro API Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/overview Generate high-quality AI videos with Seedance 2.5 Pro. Text-to-video, image-to-video, reference-driven and multishot generation at 480p, 720p, or 1080p, 4 to 30 second durations, native audio, camera motion control, and seven aspect ratios. Seedance 2.5 Pro is the latest Seedance video model, generating videos from text, images, reference media, or a multishot storyboard at 480p, 720p, and 1080p — with native audio, camera-motion control, and durations from 4 to 30 seconds. Seedance 2.5 Pro is an AI video generation API that produces videos from a text prompt (text-to-video), from a first-frame `image` and optional last-frame `image_end` (image-to-video and transitions), from a set of `reference_images` and `reference_audios` (reference-to-video), or from a `multishot` storyboard of up to six shots. A single text `prompt` is always required. Compared with Seedance 2.0 Pro, the 2.5 model adds native audio generation, camera-motion presets, subject/style/audio references, multishot sequences, flexible durations from 4 to 30 seconds, and a selectable output container. Resolution is selected at generation time: there is a dedicated **480p**, **720p**, and **1080p** generate endpoint. Task listing and status are per resolution — each resolution has its own list and status endpoint. ### Key capabilities * **Text-to-video and image-to-video**: Generate from a text prompt alone, or guide generation with a first-frame `image` and optional last-frame `image_end` * **First-to-last-frame transitions**: Provide both `image` and `image_end` to generate a video that transitions between the two frames * **Reference-to-video**: Supply up to 30 `reference_images` (subjects, characters, products, styles) and up to 10 `reference_audios`, and cite them in the prompt with `@Image1`, `@Audio1`, … * **Multishot storyboards**: Provide up to 6 `multishot` shots to drive a multi-scene sequence in a single request * **Native audio**: Generate sound effects and ambience with `sound_effects`, and optionally suppress background music with `no_music` * **Camera motion control**: Choose from more than 50 `camera_motion` presets, or keep a static tripod shot with `camera_fixed` * **Resolution by generate endpoint**: Dedicated `480p`, `720p`, and `1080p` generate endpoints; resolution is not a body parameter * **Flexible durations**: Any integer from `4` to `30` seconds (default `5`) * **Seven aspect ratios**: From ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21`, default `widescreen_16_9` * **Selectable output format**: `mp4` (default) or `mov` * **Reproducible output**: Provide a `seed` (0 to 4294967295) to repeat results; `-1` selects a random seed * **Safety filtering**: Content safety checker is enabled by default via `enable_safety_checker` * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated video is returned via a URL valid for 24 hours ### Use cases * **Premium marketing content**: Produce high-resolution 1080p clips with native audio for ads, launches, and brand campaigns * **Cinematic storytelling**: Combine `camera_motion` presets, ultra-wide `film_horizontal_21_9`, and multishot sequences for film-style scenes * **Character and product consistency**: Keep the same subject or product across shots with `reference_images` * **Longer-form clips**: Generate up to 30 seconds in a single request * **Social media content**: Generate vertical `social_story_9_16` clips for TikTok, Instagram Reels, and YouTube Shorts ### Generation modes | Mode | Required input | Optional input | Output behavior | | ---------------------- | ------------------------------------------------------- | ------------------------------------------------------------------------------------ | ------------------------------------------------------------------------- | | **Text-to-Video** | `prompt` | `aspect_ratio`, `duration`, `seed`, `camera_fixed`, `camera_motion`, `sound_effects` | Video generated entirely from the text prompt | | **Image-to-Video** | `prompt` + `image` | `image_end`, `duration`, `seed`, `camera_fixed` | Video starts from the first-frame image; add `image_end` for a transition | | **Reference-to-Video** | `prompt` + `reference_images` and/or `reference_audios` | `duration`, `seed`, `camera_fixed`, `camera_motion` | Prompt cites references with `@Image1`, `@Audio1`, … | | **Multishot** | `prompt` + `multishot` | `duration` (per shot and total), `seed` | Up to 6 shots drive a multi-scene sequence | ## API Operations Generate a video using the `480p`, `720p`, or `1080p` endpoint, then track it with the list and status endpoints for that resolution. Generation returns a task ID for async polling or webhook notification.
Generate a 480p video from text, image, references, or a multishot storyboard Generate a 720p video from text, image, references, or a multishot storyboard Generate a 1080p video from text, image, references, or a multishot storyboard List tasks for a given resolution Get task status and result by ID
### Endpoint structure | Operation | Endpoint | | ------------------ | --------------------------------------------------------------- | | **Generate 480p** | `POST /v1/ai/video/seedance-2-5-pro-480p` | | **Generate 720p** | `POST /v1/ai/video/seedance-2-5-pro-720p` | | **Generate 1080p** | `POST /v1/ai/video/seedance-2-5-pro-1080p` | | **List tasks** | `GET /v1/ai/video/seedance-2-5-pro-{480p,720p,1080p}` | | **Get task** | `GET /v1/ai/video/seedance-2-5-pro-{480p,720p,1080p}/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-480p`, `-720p`, or `-1080p`) and is not a body parameter. The request body is identical for all three endpoints. | Parameter | Type | Required | Default | Description | | ----------------------- | ---------- | -------- | ----------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video. Cite references as `@Image1`, `@Audio1`, … in 1-indexed array order. Up to 40000 characters (over 5,000 the results may become less predictable) | | `image` | `string` | No | - | First-frame image for image-to-video, as a public URL or Base64 string. Mutually exclusive with `reference_images` | | `image_end` | `string` | No | - | Last-frame image, as a public URL or Base64 string. Requires `image`; produces a first-to-last-frame transition | | `reference_images` | `string[]` | No | - | Up to 30 reference images (subjects, characters, products, styles). Cite them as `@Image1` … `@Image30`. Mutually exclusive with `image` | | `reference_audios` | `string[]` | No | - | Up to 10 reference audios (2–60s each, mp3 or wav URL). Cite them as `@Audio1` … `@Audio10` | | `multishot` | `object[]` | No | - | Up to 6 shots, each `{ prompt, duration? }` (per-shot `prompt` up to 40000 characters, per-shot `duration` from `1` to `12` seconds). When provided, the shots drive generation (top-level `prompt` still required). If any shot has a `duration`, all must, and they must sum to `duration` | | `duration` | `integer` | No | `5` | Video length in seconds: any integer from `4` to `30` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | Output ratio: `film_horizontal_21_9`, `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, `film_vertical_9_21`. Ignored when `image` is provided | | `camera_fixed` | `boolean` | No | `false` | When `true`, the camera stays fixed (tripod shot); when `false`, natural camera movement may be applied | | `camera_motion` | `string` | No | - | One of 50+ camera-motion presets (e.g. `orbitLeft`, `pushIn`, `craneUp`, `fpvDrone`). Expanded into a descriptive instruction appended to the prompt | | `sound_effects` | `boolean` | No | `true` | Generate native audio (sound effects and ambience) for the video | | `no_music` | `boolean` | No | `false` | Suppress background music in the native audio. Only applies when `sound_effects` is enabled | | `output_format` | `string` | No | `mp4` | Container format: `mp4` (default) or `mov`. `mov` trades file size for colour fidelity | | `seed` | `integer` | No | `-1` | Random seed for reproducible results (`0` to `4294967295`). Use `-1` for a random seed | | `enable_safety_checker` | `boolean` | No | `true` | Enables content safety filtering on the generated video | | `webhook_url` | `string` | No | - | URL for async status notifications when the task completes | ## Frequently Asked Questions Seedance 2.5 Pro is the latest Seedance video model. You submit a text `prompt` (and optionally an `image`, `reference_images`, `reference_audios`, or a `multishot` storyboard) to a resolution-specific generate endpoint and receive a task ID immediately. Poll the GET status endpoint for that resolution or supply a `webhook_url` to be notified when the task completes, then download the video from the returned URL (valid for 24 hours). Resolution is selected by the generate endpoint, not by a body parameter. Call `POST /v1/ai/video/seedance-2-5-pro-480p` for 480p, `POST /v1/ai/video/seedance-2-5-pro-720p` for 720p, or `POST /v1/ai/video/seedance-2-5-pro-1080p` for 1080p. Task listing and status are per resolution. Use `GET /v1/ai/video/seedance-2-5-pro-{resolution}` to list tasks and `GET /v1/ai/video/seedance-2-5-pro-{resolution}/{task-id}` to get a single task, using the same resolution you generated with. Provide `reference_images` (up to 30) and/or `reference_audios` (up to 10), then cite them in the `prompt` by index: `@Image1` is the first item of `reference_images`, `@Audio1` is the first item of `reference_audios`, and so on. `reference_images` is mutually exclusive with the first-frame `image`; `reference_audios` accepts 2–60 second mp3 or wav URLs. `multishot` is an ordered list of up to 6 shots, each with its own `prompt` (up to 40000 characters) and optional `duration` (1 to 12 seconds). When provided, the shots drive the generation and the top-level `prompt` is still required. If any shot specifies a `duration`, every shot must, and the per-shot durations must sum to the top-level `duration`. Yes. Native audio (sound effects and ambience) is generated by default (`sound_effects: true`). Set `sound_effects: false` for a silent video, or keep audio and set `no_music: true` to suppress background music while keeping other sound. Use `camera_fixed: true` for a static tripod shot, or set `camera_motion` to one of 50+ presets (for example `orbitLeft`, `pushIn`, `craneUp`, `whipPan`, `fpvDrone`) to guide the shot. The preset is expanded into a descriptive instruction appended to the prompt. Any integer from `4` to `30` seconds. The default is `5` seconds. Seven aspect ratios: `film_horizontal_21_9` (21:9), `widescreen_16_9` (16:9, default), `classic_4_3` (4:3), `square_1_1` (1:1), `traditional_3_4` (3:4), `social_story_9_16` (9:16), and `film_vertical_9_21` (9:21). This parameter is ignored when a first-frame `image` is provided. Yes. Set `output_format` to `mp4` (default) or `mov`. `mov` trades a larger file size for higher colour fidelity. Yes. Provide a fixed `seed` value (from `0` to `4294967295`) together with the same prompt and parameters to get consistent results. Use `-1` (the default) for a random seed on each request. Rate limits and pricing depend on your subscription tier and the selected resolution and duration. See the [Rate Limits](/ratelimits) and [Pricing](/pricing) pages for current values. ## Best practices * **Resolution choice**: Use the `-1080p` endpoint for final, high-fidelity output and the `-480p` or `-720p` endpoints for faster, lower-cost iterations * **Prompt writing**: Be specific about the scene, subjects, motion, lighting, and visual style; when using references, cite them explicitly as `@Image1`, `@Audio1`, … * **References**: Use `reference_images` to keep a character or product consistent; remember it is mutually exclusive with the first-frame `image` * **Multishot**: Either omit all per-shot durations (let the model distribute time) or set every one so they sum to the total `duration` * **Audio**: Native audio can occasionally be moderated; if a request is rejected for audio content, retry with `sound_effects: false` * **Camera control**: Set `camera_fixed: true` for stable product or talking-head shots, or pick a `camera_motion` preset for deliberate movement * **Reproducibility**: Fix the `seed` while iterating on a prompt so you can compare changes consistently * **Production integration**: Use webhooks instead of polling for scalable applications * **Result retrieval**: Download the output promptly, as the delivery URL is valid for 24 hours ## Related APIs * **[Seedance 2.0 Pro](/api-reference/video/seedance-2-pro/overview)**: Previous Seedance Pro generation with 480p, 720p, 1080p, and 4K generate endpoints * **[Seedance 2.0 Fast](/api-reference/video/seedance-2-fast/overview)**: Speed-optimized Seedance 2.0 variant * **[Kling 3](/api-reference/video/kling-v3/overview)**: Higher-fidelity Kling 3.0 model for premium output # Seedance 2.5 Pro 1080p - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/task-by-id-1080p get /v1/ai/video/seedance-2-5-pro-1080p/{task-id} # Seedance 2.5 Pro 480p - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/task-by-id-480p get /v1/ai/video/seedance-2-5-pro-480p/{task-id} # Seedance 2.5 Pro 720p - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/task-by-id-720p get /v1/ai/video/seedance-2-5-pro-720p/{task-id} # Seedance 2.5 Pro 1080p - List tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/tasks-1080p get /v1/ai/video/seedance-2-5-pro-1080p # Seedance 2.5 Pro 480p - List tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/tasks-480p get /v1/ai/video/seedance-2-5-pro-480p # Seedance 2.5 Pro 720p - List tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-5-pro/tasks-720p get /v1/ai/video/seedance-2-5-pro-720p # Seedance 2.0 Fast 480p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-fast/generate-480p post /v1/ai/video/seedance-2-fast-480p Generate 480p video using Seedance 2.0 Fast. Faster generation with slightly reduced quality compared to Pro. Supports text-to-video and image-to-video with optional first and last frame images. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Fast 720p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-fast/generate-720p post /v1/ai/video/seedance-2-fast-720p Generate 720p video using Seedance 2.0 Fast. Faster generation with slightly reduced quality compared to Pro. Supports text-to-video and image-to-video with optional first and last frame images. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Fast API Source: https://docs.magnific.com/api-reference/video/seedance-2-fast/overview Generate AI videos quickly with Seedance 2.0 Fast. Text-to-video and image-to-video at 480p or 720p, durations from 4 to 15 seconds, first and last frame control, and seven aspect ratios. Seedance 2.0 Fast is the speed-optimized variant of the Seedance 2.0 family, generating videos from text or images at 480p and 720p for quick-turnaround creative workflows. Seedance 2.0 Fast is an AI video generation API that produces videos from a text prompt (text-to-video) or from one or two reference images (image-to-video). It is the speed-optimized variant of the Seedance 2.0 family, available at 480p and 720p with durations from 4 to 15 seconds. A single text `prompt` is always required; supplying a first-frame `image` (and optionally a last-frame `image_end`) switches the model into image-to-video or first-to-last-frame transition mode. Resolution is selected at generation time: there is a dedicated **480p** and **720p** generate endpoint. Task listing and status use a single generic endpoint regardless of the resolution used to create the task. ### Key capabilities * **Text-to-video and image-to-video**: Generate from a text prompt alone, or guide generation with a first-frame `image` and optional last-frame `image_end` * **First-to-last-frame transitions**: Provide both `image` and `image_end` to generate a video that transitions between the two frames * **Resolution by generate endpoint**: Dedicated `480p` and `720p` generate endpoints; resolution is not a body parameter * **Unified task tracking**: One generic list endpoint and one generic status endpoint cover tasks of any resolution * **Durations**: `5` or `10` seconds (default `5`) * **Seven aspect ratios**: From ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21`, default `widescreen_16_9` * **Camera control**: Set `camera_fixed` to `true` for a static tripod shot or `false` for natural camera movement * **Reproducible output**: Provide a `seed` (0 to 4294967295) to repeat results; `-1` selects a random seed * **Safety filtering**: Content safety checker is enabled by default via `enable_safety_checker` * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated MP4 video is returned via a URL valid for 24 hours ### Use cases * **Fast iteration**: Generate videos quickly when turnaround speed matters more than maximum fidelity * **Social media content**: Produce vertical `social_story_9_16` clips for TikTok, Instagram Reels, and YouTube Shorts * **High-volume generation**: Batch-generate many videos for testing, previews, and A/B variations * **Product animation**: Animate a product image, or transition between two product shots with first and last frame control * **Ad variations**: Generate multiple aspect ratios and durations from a single concept ### Generation modes | Mode | Required input | Optional input | Output behavior | | ----------------------- | -------------------------------- | -------------------------------------------------- | -------------------------------------------------------- | | **Text-to-Video** | `prompt` | `aspect_ratio`, `duration`, `seed`, `camera_fixed` | Video generated entirely from the text prompt | | **Image-to-Video** | `prompt` + `image` | `duration`, `seed`, `camera_fixed` | Video starts from the first-frame image | | **First-to-last frame** | `prompt` + `image` + `image_end` | `duration`, `seed`, `camera_fixed` | Video transitions from the first frame to the last frame | ## API Operations Generate a video using the `480p` or `720p` endpoint, then track it with the generic list and status endpoints. Generation returns a task ID for async polling or webhook notification.
Generate a 480p video from text or image Generate a 720p video from text or image List all Seedance 2.0 Fast tasks (any resolution) Get task status and result by ID
### Endpoint structure | Operation | Endpoint | | ----------------- | -------------------------------------------- | | **Generate 480p** | `POST /v1/ai/video/seedance-2-fast-480p` | | **Generate 720p** | `POST /v1/ai/video/seedance-2-fast-720p` | | **List tasks** | `GET /v1/ai/video/seedance-2-fast` | | **Get task** | `GET /v1/ai/video/seedance-2-fast/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-480p` or `-720p`) and is not a body parameter. The request body is identical for both endpoints. | Parameter | Type | Required | Default | Description | | ----------------------- | --------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video, including scene, motion, visual style, and camera movements. Up to 40000 characters (over 5,000 the results may become less predictable) | | `image` | `string` | No | - | First-frame image for image-to-video, as a public URL or Base64 string. When provided, the video starts from this image | | `image_end` | `string` | No | - | Last-frame image, as a public URL or Base64 string. When provided together with `image`, the video transitions from the first frame to this frame | | `duration` | `integer` | No | `5` | Video length in seconds: any integer from `4` to `15` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | Output ratio: `film_horizontal_21_9`, `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, `film_vertical_9_21` | | `camera_fixed` | `boolean` | No | `false` | When `true`, the camera stays fixed (tripod shot); when `false`, natural camera movement may be applied | | `seed` | `integer` | No | `-1` | Random seed for reproducible results (`0` to `4294967295`). Use `-1` for a random seed | | `sound_effects` | `boolean` | No | `true` | Generates native audio (sound effects and ambience). Set to `false` to produce a silent video | | `no_music` | `boolean` | No | `false` | Suppresses background music in the native audio, keeping dialogue and sound effects. Only applies when `sound_effects` is enabled | | `enable_safety_checker` | `boolean` | No | `true` | Enables content safety filtering on the generated video | | `webhook_url` | `string` | No | - | URL for async status notifications when the task completes | ## Frequently Asked Questions Seedance 2.0 Fast is the speed-optimized variant of the Seedance 2.0 family. You submit a text `prompt` (and optionally a first-frame `image` and last-frame `image_end`) to a resolution-specific generate endpoint and receive a task ID immediately. Poll the generic GET status endpoint or supply a `webhook_url` to be notified when the task completes, then download the MP4 video from the returned URL (valid for 24 hours). Resolution is selected by the generate endpoint, not by a body parameter. Call `POST /v1/ai/video/seedance-2-fast-480p` for 480p or `POST /v1/ai/video/seedance-2-fast-720p` for 720p. Task listing and status use the same generic endpoints regardless of resolution. Seedance 2.0 Fast is the speed-optimized variant and supports 480p and 720p. [Seedance 2.0 Pro](/api-reference/video/seedance-2-pro/overview) is the highest-quality variant and supports 480p, 720p, and 1080p. [Seedance 2.0 Mini](/api-reference/video/seedance-2-mini/overview) is the most affordable variant and supports 480p and 720p. All three share the same request parameters; choose Fast when turnaround speed is the priority. No. Status and listing are generic: `GET /v1/ai/video/seedance-2-fast` lists all tasks and `GET /v1/ai/video/seedance-2-fast/{task-id}` returns the status of a single task, whether it was generated at 480p or 720p. The `prompt` field is always required. For text-to-video, send only the `prompt` and the video is generated entirely from your description. For image-to-video, also send an `image` to use as the first frame; the video then starts from that image. To generate a transition video, send both `image` (first frame) and `image_end` (last frame). Seedance 2.0 Fast supports `5` or `10` second videos. The default is `5` seconds. Duration is an integer value. Seven aspect ratios are available: `film_horizontal_21_9` (21:9), `widescreen_16_9` (16:9, default), `classic_4_3` (4:3), `square_1_1` (1:1), `traditional_3_4` (3:4), `social_story_9_16` (9:16), and `film_vertical_9_21` (9:21). Provide `image` and `image_end` either as a publicly accessible URL or as a Base64-encoded string. `image` sets the first frame; adding `image_end` produces a first-to-last-frame transition. Yes. Provide a fixed `seed` value (from `0` to `4294967295`) together with the same prompt and parameters to get consistent results. Use `-1` (the default) for a random seed on each request. Seedance 2.0 Fast is optimized for quick turnaround. Poll the GET task endpoint or provide a `webhook_url` to be notified on completion. The result MP4 is delivered via a URL that remains valid for 24 hours. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Pricing varies by resolution and duration. See the [Pricing](/pricing) page for current rates and credit costs. ## Best practices * **Resolution choice**: Use the `-480p` endpoint for the fastest iterations, and the `-720p` endpoint when you need more detail * **Prompt writing**: Be specific about the scene, subjects, motion, camera movement, lighting, and visual style for better results * **Image-to-video**: Provide a clear, high-quality first-frame `image`; add `image_end` only when you want a transition between two frames * **Camera control**: Set `camera_fixed` to `true` for stable product or talking-head shots, and `false` to let the model add natural motion * **Reproducibility**: Fix the `seed` while iterating on a prompt so you can compare changes consistently * **Duration selection**: Start at `5` seconds for fast iterations, then switch to `10` seconds for final outputs * **Production integration**: Use webhooks instead of polling for scalable applications * **Result retrieval**: Download the output promptly, as the delivery URL is valid for 24 hours ## Related APIs * **[Seedance 2.0 Pro](/api-reference/video/seedance-2-pro/overview)**: Highest-quality Seedance 2.0 variant with 480p, 720p, and 1080p generate endpoints * **[Seedance 2.0 Mini](/api-reference/video/seedance-2-mini/overview)**: Most affordable Seedance 2.0 variant with 480p and 720p generate endpoints * **[Kling 3](/api-reference/video/kling-v3/overview)**: Higher-fidelity Kling 3.0 model for premium output # Seedance 2.0 Fast - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-fast/task-by-id get /v1/ai/video/seedance-2-fast/{task-id} # Seedance 2.0 Fast - List all tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-fast/tasks get /v1/ai/video/seedance-2-fast # Seedance 2.0 Mini 480p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-mini/generate-480p post /v1/ai/video/seedance-2-mini-480p Generate 480p video using Seedance 2.0 Mini, the most affordable and fastest variant of the Seedance 2.0 family. Available in 480p and 720p with durations from 4 to 15 seconds. Supports text-to-video and image-to-video with optional first and last frame images. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Mini 720p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-mini/generate-720p post /v1/ai/video/seedance-2-mini-720p Generate 720p video using Seedance 2.0 Mini, the most affordable and fastest variant of the Seedance 2.0 family. Available in 480p and 720p with durations from 4 to 15 seconds. Supports text-to-video and image-to-video with optional first and last frame images. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Mini API Source: https://docs.magnific.com/api-reference/video/seedance-2-mini/overview Generate affordable AI videos with Seedance 2.0 Mini. Text-to-video and image-to-video at 480p or 720p, durations from 4 to 15 seconds, first and last frame control, and seven aspect ratios. Seedance 2.0 Mini is the most affordable and fastest variant of the Seedance 2.0 family, generating short videos from text or images at 480p and 720p for high-volume, cost-sensitive workflows. Seedance 2.0 Mini is an AI video generation API that produces short videos from a text prompt (text-to-video) or from one or two reference images (image-to-video). It is the most affordable and fastest variant of the Seedance 2.0 family, available at 480p and 720p with durations from 4 to 15 seconds. A single text `prompt` is always required; supplying a first-frame `image` (and optionally a last-frame `image_end`) switches the model into image-to-video or first-to-last-frame transition mode. Resolution is selected at generation time: there is a dedicated **480p** generate endpoint and a dedicated **720p** generate endpoint. Task listing and status use a single generic endpoint regardless of the resolution used to create the task. ### Key capabilities * **Text-to-video and image-to-video**: Generate from a text prompt alone, or guide generation with a first-frame `image` and optional last-frame `image_end` * **First-to-last-frame transitions**: Provide both `image` and `image_end` to generate a video that transitions between the two frames * **Native audio**: Generate sound effects and ambience with `sound_effects`, and optionally suppress background music with `no_music` * **Resolution by generate endpoint**: Dedicated `480p` and `720p` generate endpoints; resolution is not a body parameter * **Unified task tracking**: One generic list endpoint and one generic status endpoint cover tasks of any resolution * **Flexible durations**: Any integer from `4` to `15` seconds (default `5`) * **Seven aspect ratios**: From ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21`, default `widescreen_16_9` * **Camera control**: Set `camera_fixed` to `true` for a static tripod shot or `false` for natural camera movement * **Reproducible output**: Provide a `seed` (0 to 4294967295) to repeat results; `-1` selects a random seed * **Safety filtering**: Content safety checker is enabled by default via `enable_safety_checker` * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated MP4 video is returned via a URL valid for 24 hours ### Use cases * **Social media content**: Produce vertical `social_story_9_16` clips for TikTok, Instagram Reels, and YouTube Shorts at low cost * **High-volume generation**: Batch-generate many short videos affordably for testing, previews, and A/B variations * **Product animation**: Animate a product image, or transition between two product shots with first and last frame control * **Rapid prototyping**: Iterate quickly on creative concepts before committing to higher-resolution models * **Ad variations**: Generate multiple aspect ratios and durations from a single concept ### Generation modes | Mode | Required input | Optional input | Output behavior | | ----------------------- | -------------------------------- | -------------------------------------------------- | -------------------------------------------------------- | | **Text-to-Video** | `prompt` | `aspect_ratio`, `duration`, `seed`, `camera_fixed` | Video generated entirely from the text prompt | | **Image-to-Video** | `prompt` + `image` | `duration`, `seed`, `camera_fixed` | Video starts from the first-frame image | | **First-to-last frame** | `prompt` + `image` + `image_end` | `duration`, `seed`, `camera_fixed` | Video transitions from the first frame to the last frame | ## API Operations Generate a video using the `480p` or `720p` endpoint, then track it with the generic list and status endpoints. Generation returns a task ID for async polling or webhook notification.
Generate a 480p video from text or image Generate a 720p video from text or image List all Seedance 2.0 Mini tasks (any resolution) Get task status and result by ID
### Endpoint structure | Operation | Endpoint | | ----------------- | -------------------------------------------- | | **Generate 480p** | `POST /v1/ai/video/seedance-2-mini-480p` | | **Generate 720p** | `POST /v1/ai/video/seedance-2-mini-720p` | | **List tasks** | `GET /v1/ai/video/seedance-2-mini` | | **Get task** | `GET /v1/ai/video/seedance-2-mini/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-480p` or `-720p`) and is not a body parameter. The request body is identical for both endpoints. | Parameter | Type | Required | Default | Description | | ----------------------- | --------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video, including scene, motion, visual style, and camera movements. Up to 40000 characters (over 5,000 the results may become less predictable) | | `image` | `string` | No | - | First-frame image for image-to-video, as a public URL or Base64 string. When provided, the video starts from this image | | `image_end` | `string` | No | - | Last-frame image, as a public URL or Base64 string. When provided together with `image`, the video transitions from the first frame to this frame | | `duration` | `integer` | No | `5` | Video length in seconds: any integer from `4` to `15` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | Output ratio: `film_horizontal_21_9`, `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, `film_vertical_9_21` | | `camera_fixed` | `boolean` | No | `false` | When `true`, the camera stays fixed (tripod shot); when `false`, natural camera movement may be applied | | `seed` | `integer` | No | `-1` | Random seed for reproducible results (`0` to `4294967295`). Use `-1` for a random seed | | `sound_effects` | `boolean` | No | `true` | Generates native audio (sound effects and ambience). Set to `false` to produce a silent video | | `no_music` | `boolean` | No | `false` | Suppresses background music in the native audio, keeping dialogue and sound effects. Only applies when `sound_effects` is enabled | | `enable_safety_checker` | `boolean` | No | `true` | Enables content safety filtering on the generated video | | `webhook_url` | `string` | No | - | URL for async status notifications when the task completes | ## Frequently Asked Questions Seedance 2.0 Mini is the most affordable and fastest variant of the Seedance 2.0 family. You submit a text `prompt` (and optionally a first-frame `image` and last-frame `image_end`) to a resolution-specific generate endpoint and receive a task ID immediately. Poll the generic GET status endpoint or supply a `webhook_url` to be notified when the task completes, then download the MP4 video from the returned URL (valid for 24 hours). Resolution is selected by the generate endpoint, not by a body parameter. Call `POST /v1/ai/video/seedance-2-mini-480p` for 480p or `POST /v1/ai/video/seedance-2-mini-720p` for 720p. Task listing and status use the same generic endpoints regardless of resolution. Seedance 2.0 Mini supports 480p and 720p only. No. Status and listing are generic: `GET /v1/ai/video/seedance-2-mini` lists all tasks and `GET /v1/ai/video/seedance-2-mini/{task-id}` returns the status of a single task, whether it was generated at 480p or 720p. The `prompt` field is always required. For text-to-video, send only the `prompt` and the video is generated entirely from your description. For image-to-video, also send an `image` to use as the first frame; the video then starts from that image. To generate a transition video, send both `image` (first frame) and `image_end` (last frame). Seedance 2.0 Mini supports any integer duration from `4` to `15` seconds. The default is `5` seconds. Seven aspect ratios are available: `film_horizontal_21_9` (21:9), `widescreen_16_9` (16:9, default), `classic_4_3` (4:3), `square_1_1` (1:1), `traditional_3_4` (3:4), `social_story_9_16` (9:16), and `film_vertical_9_21` (9:21). Provide `image` and `image_end` either as a publicly accessible URL or as a Base64-encoded string. `image` sets the first frame; adding `image_end` produces a first-to-last-frame transition. Yes. Provide a fixed `seed` value (from `0` to `4294967295`) together with the same prompt and parameters to get consistent results. Use `-1` (the default) for a random seed on each request. Seedance 2.0 Mini is optimized for fast, affordable generation. Poll the GET task endpoint or provide a `webhook_url` to be notified on completion. The result MP4 is delivered via a URL that remains valid for 24 hours. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Seedance 2.0 Mini is the most affordable variant of the Seedance 2.0 family. Pricing varies by resolution and duration. See the [Pricing](/pricing) page for current rates and credit costs. ## Best practices * **Resolution choice**: Use the `-480p` endpoint for the lowest cost and fastest generation, and the `-720p` endpoint when you need more detail * **Prompt writing**: Be specific about the scene, subjects, motion, camera movement, lighting, and visual style for better results * **Image-to-video**: Provide a clear, high-quality first-frame `image`; add `image_end` only when you want a transition between two frames * **Camera control**: Set `camera_fixed` to `true` for stable product or talking-head shots, and `false` to let the model add natural motion * **Reproducibility**: Fix the `seed` while iterating on a prompt so you can compare changes consistently * **Duration selection**: Start at `4` or `5` seconds for fast iterations, then extend up to `15` seconds for final outputs * **Production integration**: Use webhooks instead of polling for scalable applications * **Result retrieval**: Download the output promptly, as the delivery URL is valid for 24 hours ## Related APIs * **[Kling 3.0 Turbo](/api-reference/text-to-video/kling-v3-turbo/overview)**: Speed-optimized Kling model with 720p and 1080p generate endpoints * **[WAN 2.7](/api-reference/text-to-video/wan-2-7/overview)**: Alternative text-to-video and image-to-video model with multi-shot sequences * **[Kling 3](/api-reference/video/kling-v3/overview)**: Higher-fidelity Kling 3.0 model for premium output # Seedance 2.0 Mini - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-mini/task-by-id get /v1/ai/video/seedance-2-mini/{task-id} # Seedance 2.0 Mini - List all tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-mini/tasks get /v1/ai/video/seedance-2-mini # Seedance 2.0 Pro 1080p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/generate-1080p post /v1/ai/video/seedance-2-pro-1080p Generate 1080p video using Seedance 2.0 Pro. Supports text-to-video and image-to-video with optional first and last frame images for transition videos. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Pro 480p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/generate-480p post /v1/ai/video/seedance-2-pro-480p Generate 480p video using Seedance 2.0 Pro. Supports text-to-video and image-to-video with optional first and last frame images for transition videos. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Pro 4K - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/generate-4k post /v1/ai/video/seedance-2-pro-4k Generate 4K video using Seedance 2.0 Pro. Supports text-to-video and image-to-video with optional first and last frame images for transition videos. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Pro 720p - Create video from text or image Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/generate-720p post /v1/ai/video/seedance-2-pro-720p Generate 720p video using Seedance 2.0 Pro. Supports text-to-video and image-to-video with optional first and last frame images for transition videos. Also supports reference-to-video: attach up to 9 reference images via `reference_images` and mention them in the prompt with @Image1..@Image9. # Seedance 2.0 Pro API Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/overview Generate high-quality AI videos with Seedance 2.0 Pro. Text-to-video and image-to-video at 480p, 720p, or 1080p, durations from 4 to 15 seconds, first and last frame control, and seven aspect ratios. Seedance 2.0 Pro is the highest-quality variant of the Seedance 2.0 family, generating videos from text or images at 480p, 720p, and 1080p for premium creative and production workflows. Seedance 2.0 Pro is an AI video generation API that produces videos from a text prompt (text-to-video) or from one or two reference images (image-to-video). It is the highest-quality variant of the Seedance 2.0 family, available at 480p, 720p, and 1080p with durations from 4 to 15 seconds. A single text `prompt` is always required; supplying a first-frame `image` (and optionally a last-frame `image_end`) switches the model into image-to-video or first-to-last-frame transition mode. Resolution is selected at generation time: there is a dedicated **480p**, **720p**, and **1080p** generate endpoint. Task listing and status use a single generic endpoint regardless of the resolution used to create the task. ### Key capabilities * **Text-to-video and image-to-video**: Generate from a text prompt alone, or guide generation with a first-frame `image` and optional last-frame `image_end` * **First-to-last-frame transitions**: Provide both `image` and `image_end` to generate a video that transitions between the two frames * **Resolution by generate endpoint**: Dedicated `480p`, `720p`, and `1080p` generate endpoints; resolution is not a body parameter * **Unified task tracking**: One generic list endpoint and one generic status endpoint cover tasks of any resolution * **Durations**: `5` or `10` seconds (default `5`) * **Seven aspect ratios**: From ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21`, default `widescreen_16_9` * **Camera control**: Set `camera_fixed` to `true` for a static tripod shot or `false` for natural camera movement * **Reproducible output**: Provide a `seed` (0 to 4294967295) to repeat results; `-1` selects a random seed * **Safety filtering**: Content safety checker is enabled by default via `enable_safety_checker` * **Async processing**: Webhook notifications or polling for task completion * **24-hour delivery**: Generated MP4 video is returned via a URL valid for 24 hours ### Use cases * **Premium marketing content**: Produce high-resolution 1080p clips for ads, launches, and brand campaigns * **Cinematic storytelling**: Use ultra-wide `film_horizontal_21_9` and camera control for film-style sequences * **Product showcases**: Animate a product image, or transition between two product shots with first and last frame control * **Social media content**: Generate vertical `social_story_9_16` clips for TikTok, Instagram Reels, and YouTube Shorts * **High-fidelity prototyping**: Iterate on creative concepts at full quality before final delivery ### Generation modes | Mode | Required input | Optional input | Output behavior | | ----------------------- | -------------------------------- | -------------------------------------------------- | -------------------------------------------------------- | | **Text-to-Video** | `prompt` | `aspect_ratio`, `duration`, `seed`, `camera_fixed` | Video generated entirely from the text prompt | | **Image-to-Video** | `prompt` + `image` | `duration`, `seed`, `camera_fixed` | Video starts from the first-frame image | | **First-to-last frame** | `prompt` + `image` + `image_end` | `duration`, `seed`, `camera_fixed` | Video transitions from the first frame to the last frame | ## API Operations Generate a video using the `480p`, `720p`, or `1080p` endpoint, then track it with the generic list and status endpoints. Generation returns a task ID for async polling or webhook notification.
Generate a 480p video from text or image Generate a 720p video from text or image Generate a 1080p video from text or image List all Seedance 2.0 Pro tasks (any resolution) Get task status and result by ID
### Endpoint structure | Operation | Endpoint | | ------------------ | ------------------------------------------- | | **Generate 480p** | `POST /v1/ai/video/seedance-2-pro-480p` | | **Generate 720p** | `POST /v1/ai/video/seedance-2-pro-720p` | | **Generate 1080p** | `POST /v1/ai/video/seedance-2-pro-1080p` | | **List tasks** | `GET /v1/ai/video/seedance-2-pro` | | **Get task** | `GET /v1/ai/video/seedance-2-pro/{task-id}` | ### Parameters Resolution is determined by the generate endpoint you call (`-480p`, `-720p`, or `-1080p`) and is not a body parameter. The request body is identical for all three endpoints. | Parameter | Type | Required | Default | Description | | ----------------------- | --------- | -------- | ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Text description of the video, including scene, motion, visual style, and camera movements. Up to 40000 characters (over 5,000 the results may become less predictable) | | `image` | `string` | No | - | First-frame image for image-to-video, as a public URL or Base64 string. When provided, the video starts from this image | | `image_end` | `string` | No | - | Last-frame image, as a public URL or Base64 string. When provided together with `image`, the video transitions from the first frame to this frame | | `duration` | `integer` | No | `5` | Video length in seconds: any integer from `4` to `15` | | `aspect_ratio` | `string` | No | `widescreen_16_9` | Output ratio: `film_horizontal_21_9`, `widescreen_16_9`, `classic_4_3`, `square_1_1`, `traditional_3_4`, `social_story_9_16`, `film_vertical_9_21` | | `camera_fixed` | `boolean` | No | `false` | When `true`, the camera stays fixed (tripod shot); when `false`, natural camera movement may be applied | | `seed` | `integer` | No | `-1` | Random seed for reproducible results (`0` to `4294967295`). Use `-1` for a random seed | | `sound_effects` | `boolean` | No | `true` | Generates native audio (sound effects and ambience). Set to `false` to produce a silent video | | `no_music` | `boolean` | No | `false` | Suppresses background music in the native audio, keeping dialogue and sound effects. Only applies when `sound_effects` is enabled | | `enable_safety_checker` | `boolean` | No | `true` | Enables content safety filtering on the generated video | | `webhook_url` | `string` | No | - | URL for async status notifications when the task completes | ## Frequently Asked Questions Seedance 2.0 Pro is the highest-quality variant of the Seedance 2.0 family. You submit a text `prompt` (and optionally a first-frame `image` and last-frame `image_end`) to a resolution-specific generate endpoint and receive a task ID immediately. Poll the generic GET status endpoint or supply a `webhook_url` to be notified when the task completes, then download the MP4 video from the returned URL (valid for 24 hours). Resolution is selected by the generate endpoint, not by a body parameter. Call `POST /v1/ai/video/seedance-2-pro-480p` for 480p, `POST /v1/ai/video/seedance-2-pro-720p` for 720p, or `POST /v1/ai/video/seedance-2-pro-1080p` for 1080p. Task listing and status use the same generic endpoints regardless of resolution. Seedance 2.0 Pro is the highest-quality variant and supports 480p, 720p, and 1080p. [Seedance 2.0 Fast](/api-reference/video/seedance-2-fast/overview) is a faster variant that also supports 480p, 720p, and 1080p, optimized for quicker turnaround. [Seedance 2.0 Mini](/api-reference/video/seedance-2-mini/overview) is the most affordable variant and supports 480p and 720p only. All three share the same request parameters; choose Pro when output quality matters most. No. Status and listing are generic: `GET /v1/ai/video/seedance-2-pro` lists all tasks and `GET /v1/ai/video/seedance-2-pro/{task-id}` returns the status of a single task, whether it was generated at 480p, 720p, or 1080p. The `prompt` field is always required. For text-to-video, send only the `prompt` and the video is generated entirely from your description. For image-to-video, also send an `image` to use as the first frame; the video then starts from that image. To generate a transition video, send both `image` (first frame) and `image_end` (last frame). Seedance 2.0 Pro supports `5` or `10` second videos. The default is `5` seconds. Duration is an integer value. Seven aspect ratios are available: `film_horizontal_21_9` (21:9), `widescreen_16_9` (16:9, default), `classic_4_3` (4:3), `square_1_1` (1:1), `traditional_3_4` (3:4), `social_story_9_16` (9:16), and `film_vertical_9_21` (9:21). Provide `image` and `image_end` either as a publicly accessible URL or as a Base64-encoded string. `image` sets the first frame; adding `image_end` produces a first-to-last-frame transition. Yes. Provide a fixed `seed` value (from `0` to `4294967295`) together with the same prompt and parameters to get consistent results. Use `-1` (the default) for a random seed on each request. Poll the GET task endpoint or provide a `webhook_url` to be notified on completion. The result MP4 is delivered via a URL that remains valid for 24 hours. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Pricing varies by resolution and duration. See the [Pricing](/pricing) page for current rates and credit costs. ## Best practices * **Resolution choice**: Use the `-1080p` endpoint for final, high-fidelity output and the `-480p` or `-720p` endpoints for faster, lower-cost iterations * **Prompt writing**: Be specific about the scene, subjects, motion, camera movement, lighting, and visual style for better results * **Image-to-video**: Provide a clear, high-quality first-frame `image`; add `image_end` only when you want a transition between two frames * **Camera control**: Set `camera_fixed` to `true` for stable product or talking-head shots, and `false` to let the model add natural motion * **Reproducibility**: Fix the `seed` while iterating on a prompt so you can compare changes consistently * **Duration selection**: Start at `5` seconds for fast iterations, then switch to `10` seconds for final outputs * **Production integration**: Use webhooks instead of polling for scalable applications * **Result retrieval**: Download the output promptly, as the delivery URL is valid for 24 hours ## Related APIs * **[Seedance 2.0 Fast](/api-reference/video/seedance-2-fast/overview)**: Faster Seedance 2.0 variant with 480p, 720p, and 1080p generate endpoints * **[Seedance 2.0 Mini](/api-reference/video/seedance-2-mini/overview)**: Most affordable Seedance 2.0 variant with 480p and 720p generate endpoints * **[Kling 3](/api-reference/video/kling-v3/overview)**: Higher-fidelity Kling 3.0 model for premium output # Seedance 2.0 Pro - Get task status Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/task-by-id get /v1/ai/video/seedance-2-pro/{task-id} # Seedance 2.0 Pro - List all tasks Source: https://docs.magnific.com/api-reference/video/seedance-2-pro/tasks get /v1/ai/video/seedance-2-pro # VFX - Apply visual effects to video Source: https://docs.magnific.com/api-reference/video/vfx/apply-effects post /v1/ai/video/vfx Apply professional visual effects and filters to your videos using the VFX API. Transform any video with cinematic effects like film grain, motion blur, VHS retro style, and anamorphic lens distortion. **Available filters:** - `1` Film grain - Adds cinematic grain texture for a classic film look - `2` Motion blur - Creates motion blur effect (configurable kernel size and decay) - `3` Fish eye - Applies fish eye lens distortion - `4` VHS - Retro VHS tape effect with scan lines and color distortion - `5` Shake - Camera shake effect for dynamic footage - `6` VGA - Low resolution VGA effect for retro aesthetics - `7` Bloom - Glowing bloom effect (adjustable contrast) - `8` Anamorphic lens - Cinematic anamorphic lens effect with horizontal flares **Use cases:** Social media content, music videos, film production, advertising, retro-style videos, and creative projects. # VFX – Video Visual Effects API | Magnific API Source: https://docs.magnific.com/api-reference/video/vfx/overview Apply cinematic visual effects to videos with VFX API. 8 professional filters including film grain, VHS, bloom, and anamorphic lens. Per-second pricing at $0.017/second. Apply professional cinematic filters and visual effects to your videos with 8 distinct effect types for film, retro, and stylized looks. VFX is a video processing API that applies professional visual effects to any video. It offers 8 cinematic filter types including film grain, motion blur, VHS tape aesthetic, bloom lighting, and anamorphic lens flares. Processing costs \$0.017 per second of video, making it cost-effective for both short clips and longer content. ### Key capabilities * **8 cinematic filter types**: Film grain, motion blur, fish eye, VHS, camera shake, VGA, bloom, and anamorphic lens * **Adjustable frame rate**: Output FPS configurable from 1 to 60 frames per second (default: 24 fps) * **Filter-specific controls**: Fine-tune bloom contrast and motion blur parameters for precise effects * **Per-second pricing**: \$0.017 per second of video for predictable costs * **Async processing**: Webhook notifications or polling for task completion ### Available filters | Filter ID | Name | Description | | --------- | ------------------- | ---------------------------------------------------------------------------- | | `1` | **Film Grain** | Adds cinematic grain texture for a classic film look | | `2` | **Motion Blur** | Creates directional motion blur effect with adjustable kernel size and decay | | `3` | **Fish Eye** | Applies spherical fish eye lens distortion | | `4` | **VHS** | Retro VHS tape aesthetic with scan lines and color bleeding | | `5` | **Shake** | Simulates handheld camera shake for dynamic footage | | `6` | **VGA** | Low resolution VGA/CRT display look for retro computing aesthetic | | `7` | **Bloom** | Glowing bloom lighting effect with adjustable contrast | | `8` | **Anamorphic Lens** | Cinematic anamorphic lens flares for widescreen film look | ### Use cases * **Film and video production**: Add cinematic grain and anamorphic flares to professional footage * **Social media content**: Create trendy VHS and retro effects for TikTok, Instagram Reels, and YouTube Shorts * **Music videos**: Apply stylized effects like bloom and motion blur for artistic visuals * **Advertising campaigns**: Enhance product videos with professional-grade visual effects * **Gaming content**: Add camera shake and motion blur to gameplay recordings * **Nostalgic content**: Transform modern footage into retro VHS or VGA aesthetics ### Apply effects with VFX Submit a video URL and filter type to create a new processing task. The service returns a task ID for async polling or webhook notification.
Create a new VFX processing task List all VFX tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ---------------------------- | --------- | -------- | ------- | --------------------------------------------------------------------------------------------------------------------------------------- | | `video` | `string` | Yes | - | URL of the video to process (must be publicly accessible) | | `filter_type` | `integer` | No | `1` | Effect type: `1` (film grain), `2` (motion blur), `3` (fish eye), `4` (VHS), `5` (shake), `6` (VGA), `7` (bloom), `8` (anamorphic lens) | | `fps` | `integer` | No | `24` | Output frames per second (1-60) | | `bloom_filter_contrast` | `number` | No | - | Contrast for bloom filter (only for filter\_type `7`) | | `motion_filter_kernel_size` | `integer` | No | - | Kernel size for motion blur (only for filter\_type `2`) | | `motion_filter_decay_factor` | `number` | No | - | Decay factor for motion blur (only for filter\_type `2`) | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions The VFX API applies professional visual effects to videos. You submit a video URL and specify a filter type (1-8), receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. Typical processing time depends on video duration. The VFX API accepts videos via publicly accessible URLs. Common formats like MP4, MOV, and WebM are supported. Ensure your video URL is accessible without authentication. VFX processing costs $0.017 per second of video. For example, a 30-second video costs approximately $0.51, and a 60-second video costs approximately \$1.02. Each filter creates a distinct visual style: Film grain (1) adds cinematic texture, Motion blur (2) creates directional blur, Fish eye (3) applies lens distortion, VHS (4) creates retro tape aesthetics, Shake (5) simulates handheld camera, VGA (6) creates CRT display look, Bloom (7) adds glow lighting, and Anamorphic lens (8) adds cinematic lens flares. Yes. For bloom (filter\_type 7), use `bloom_filter_contrast` to adjust the glow intensity. For motion blur (filter\_type 2), use `motion_filter_kernel_size` and `motion_filter_decay_factor` to control blur strength and falloff. The VFX API supports output frame rates from 1 to 60 fps. The default is 24 fps, which is standard for cinematic content. Use higher values like 30 or 60 fps for smoother motion in fast-paced content. Processing time depends on video duration and complexity. Use webhooks for production workflows to receive notifications when processing completes, rather than polling. ## Best practices * **Input quality**: Use high-quality source videos for best results; effects amplify existing artifacts * **Filter selection**: Match the filter to your creative intent; film grain for cinematic, VHS for retro, bloom for dreamy aesthetics * **Frame rate tuning**: Use 24 fps for cinematic content, 30 fps for web video, 60 fps for gaming content * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Image Upscaler](/api-reference/image-upscaler-precision-v2/post-image-upscaler-precision-v2)**: Enhance image resolution before generating videos to apply VFX effects * **[Kling 2.6 Pro](/api-reference/image-to-video/kling-v2-6-pro)**: Generate AI videos from images to apply VFX effects to * **[WAN 2.6](/api-reference/image-to-video/wan-v2-6-720p)**: Create AI videos that can be enhanced with VFX filters # VFX - Get task status Source: https://docs.magnific.com/api-reference/video/vfx/task-by-id get /v1/ai/video/vfx/{task-id} Retrieve the status and results of a specific VFX video effect task by its task ID. # VFX - List tasks Source: https://docs.magnific.com/api-reference/video/vfx/vfx-tasks get /v1/ai/video/vfx Retrieve the status of all VFX video effect tasks for the authenticated user. # Video Upscaler Precision API Source: https://docs.magnific.com/api-reference/video/video-upscaler-precision/overview Upscale videos with AI diffusion-based enhancement using Video Upscaler Precision. Output at 720p, 1K, 2K, or 4K resolution with sharpening, smart grain, strength, and FPS boost controls. Video Upscaler Precision uses AI diffusion models for high-fidelity video upscaling with fine-grained control over sharpness, grain, and output strength. Video Upscaler Precision is an AI-powered video upscaling API that enhances video resolution using diffusion-based models for faithful, controlled quality improvement. Unlike the standard Video Upscaler, Precision focuses on detail recovery and controlled enhancement without adding AI-generated content -- it upscales what is already in the video. The API supports output resolutions of 720p, 1K, 2K (default), and 4K with adjustable sharpening, smart grain, and upscaling strength. For the standard upscaler with creativity controls and Turbo mode, see Video Upscaler. ### Key capabilities * **Resolution up to 4K**: Upscale videos to `720p`, `1k`, `2k` (default), or `4k` output resolution * **Strength control**: Adjustable upscaling intensity (0-100, default 60) to blend between original and fully upscaled output * **FPS boost**: Increase video frame rate for smoother motion. When enabled, the strength parameter is ignored and the fully upscaled result is returned * **Sharpen control**: Adjustable sharpening intensity (0-100, default 0) for edge definition and clarity * **Smart grain**: Film grain effect (0-100, default 0) to add natural texture and prevent artificial smoothness * **Async processing**: Webhook notifications or polling for task completion * **URL-based input**: Accepts a publicly accessible HTTPS URL pointing to the video ### Use cases * **Film restoration**: Upscale archival and classic footage to modern resolutions while preserving the original aesthetic * **Content repurposing**: Enhance low-resolution social media clips for large-screen display or broadcast delivery * **Video production**: Upscale B-roll and legacy footage to match project resolution requirements with precise control * **E-commerce**: Enhance product videos with higher resolution and sharpness for better detail visibility * **Surveillance review**: Improve clarity of security footage for analysis and evidence review * **Educational content**: Enhance older training videos and course recordings for modern high-resolution displays ### Upscale videos with Precision Create a precision upscaling task by submitting a video to the API. The service returns a task ID for async polling or webhook notification.
Create a new precision video upscaling task List all precision upscaler tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `video` | `string` | Yes | - | Source video as a publicly accessible HTTPS URL | | `resolution` | `string` | No | `"2k"` | Output resolution: `"720p"`, `"1k"`, `"2k"`, or `"4k"` | | `fps_boost` | `boolean` | No | `false` | Increase frame rate for smoother motion. When enabled, `strength` is ignored and the fully upscaled result is returned | | `sharpen` | `integer` | No | `0` | Sharpening intensity (0-100). Higher values increase edge definition | | `smart_grain` | `integer` | No | `0` | Film grain effect (0-100). Adds natural texture to prevent artificial smoothness | | `strength` | `integer` | No | `60` | Upscaling intensity (0-100). 0 returns the original video, 100 returns the fully upscaled result. Intermediate values blend original with upscaled output. Ignored when `fps_boost` is enabled | | `webhook_url` | `string` | No | - | URL for completion notification with task status and output URL | ## Frequently Asked Questions Video Upscaler Precision is an AI-powered video upscaling API that uses diffusion models to increase video resolution with faithful detail recovery. You submit a video via a publicly accessible HTTPS URL, receive a task ID, then poll for results or receive a webhook notification when processing completes. The API processes each frame with controlled enhancement based on your parameter settings. **Video Upscaler Precision** focuses on faithful, controlled upscaling with a `strength` parameter that lets you blend between original and upscaled output. It does not have creativity or flavor controls. The **standard [Video Upscaler](/api-reference/video/video-upscaler/overview)** offers creativity controls (0-100), processing flavors (`vivid`/`natural`), and a separate Turbo endpoint for faster processing. Choose Precision when you want the upscaled video to stay true to the original without AI-generated additions. The strength parameter (0-100, default 60) controls the blend between the original video and the fully upscaled result. A value of 0 returns the original video unchanged. A value of 100 returns the fully upscaled output. Values in between produce a weighted blend. This parameter is ignored when `fps_boost` is enabled, which always returns the fully upscaled result. Video Upscaler Precision supports four output resolutions: `720p`, `1k`, `2k` (default), and `4k`. Higher resolutions increase processing time. See the Pricing page for details on each resolution tier. Rate limits vary by subscription tier. See the Rate Limits page for current limits by plan. Pricing depends on the output resolution you select. See the Pricing page for current rates and subscription options. ## Best practices * **Input quality**: Start from the highest-quality source video available. Heavily compressed inputs may have artifacts amplified during upscaling * **Strength tuning**: Start with the default strength of 60. Use lower values (20-40) for subtle enhancement, higher values (80-100) for maximum upscaling effect * **FPS boost decision**: Enable FPS boost for footage with low frame rates (24fps or below) to create smoother playback. Note that it overrides the strength parameter * **Smart grain for realism**: Increase smart\_grain for footage that should retain a filmic quality rather than appearing digitally processed * **Resolution selection**: Choose 2K (default) for most use cases as a good balance of quality and processing time. Use 4K when the target display or delivery format requires it * **Production integration**: Use webhooks instead of polling for scalable, event-driven applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Video Upscaler](/api-reference/video/video-upscaler/overview)**: Standard video upscaler with creativity controls, processing flavors, and a Turbo endpoint for faster processing * **[Upscaler Precision V2](/api-reference/image-upscaler-precision-v2/overview)**: High-fidelity image upscaling with granular controls for sharpness, texture, and detail * **[VFX](/api-reference/video/vfx/overview)**: Apply cinematic visual effects to videos with professional filters # Video Upscaler Precision - Get task status Source: https://docs.magnific.com/api-reference/video/video-upscaler-precision/task-by-id get /v1/ai/video-upscaler-precision/{task-id} Returns the current status and output URL of a specific video upscaler precision task. The output URL is included only if the task has completed successfully. # Video Upscaler Precision - Upscale video Source: https://docs.magnific.com/api-reference/video/video-upscaler-precision/upscale-video post /v1/ai/video-upscaler-precision Upscales a video with precision-grade quality enhancement. Designed for frame-accurate upscaling with fine-grained control over sharpening, grain, and output strength. Supports output resolutions of 720p, 1k, 2k, or 4k with optional FPS boost. # Video Upscaler Precision - List tasks Source: https://docs.magnific.com/api-reference/video/video-upscaler-precision/video-upscaler-precision-tasks get /v1/ai/video-upscaler-precision Returns a list of all video upscaler precision tasks. Each task includes its ID, current status, and output URL if completed. # Video Upscaler Topaz API Source: https://docs.magnific.com/api-reference/video/video-upscaler-topaz/overview Upscale videos with Topaz Starlight generative AI enhancement. Output at 720p, 1K, 2K, or 4K resolution with model selection, AI frame interpolation, and noise control. Video Upscaler Topaz uses Topaz Starlight generative models for high-quality video upscaling with model selection, AI frame interpolation, and noise control. Video Upscaler Topaz is an AI-powered video upscaling API built on Topaz Starlight generative models. It enhances video resolution with generative detail synthesis and supports optional frame-rate changes through AI frame interpolation. The API offers two enhancement models -- `starlight_precise_2_5` for the highest quality and `starlight_fast_2` for faster processing -- with output resolutions of 720p, 1K, 2K (default), and 4K. For faithful, non-generative upscaling with a strength blend, see Video Upscaler Precision. ### Key capabilities * **Topaz Starlight models**: Choose `starlight_precise_2_5` (highest quality, default) or `starlight_fast_2` (faster, slight quality trade-off) * **Resolution up to 4K**: Upscale videos to `720p`, `1k`, `2k` (default), or `4k` output resolution while preserving the source aspect ratio * **AI frame interpolation**: Change the output frame rate with `target_fps`. When it differs from the source, an interpolation model is applied -- `apollo` (recommended) or `chronos` (optimized for slow-motion and large frame-rate changes) * **Noise control**: Adjustable noise reduction / addition intensity (0-1, default 0) * **Async processing**: Webhook notifications or polling for task completion * **Flexible input**: Accepts a publicly accessible HTTPS URL or an uploaded file reference (`upl_...`) ### Use cases * **Film restoration**: Upscale archival and classic footage to modern resolutions with generative detail recovery * **Content repurposing**: Enhance low-resolution clips for large-screen display or broadcast delivery * **Video production**: Upscale B-roll and legacy footage to match project resolution requirements * **Smooth motion**: Boost frame rate with AI interpolation for smoother playback of low-fps footage * **E-commerce**: Enhance product videos with higher resolution and detail * **Educational content**: Modernize older training videos and recordings for high-resolution displays ### Upscale videos with Topaz Create a Topaz upscaling task by submitting a video to the API. The service returns a task ID for async polling or webhook notification.
Create a new Topaz video upscaling task List all Topaz upscaler tasks Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | --------------------- | --------- | -------- | ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `video` | `string` | Yes | - | Source video as a publicly accessible HTTPS URL or an uploaded file reference (`upl_...`) | | `enhancement_model` | `string` | Yes | `"starlight_precise_2_5"` | Topaz Starlight model: `"starlight_precise_2_5"` (highest quality) or `"starlight_fast_2"` (faster) | | `resolution` | `string` | No | `"2k"` | Output resolution: `"720p"`, `"1k"`, `"2k"`, or `"4k"`. Aspect ratio is preserved | | `target_fps` | `integer` | No | - | Target output frame rate (1-60). When it differs from the source, AI frame interpolation is applied. Omit to keep the source frame rate. Interpolation requires a target of at least 15 fps | | `frame_interpolation` | `string` | No | `"apollo"` | Interpolation model used when `target_fps` changes the frame rate: `"apollo"` (recommended) or `"chronos"` (slow-motion / large changes) | | `noise` | `number` | No | `0` | Noise reduction / addition intensity (0-1) applied by the enhancement model | | `webhook_url` | `string` | No | - | URL for completion notification with task status and output URL | ## Frequently Asked Questions Video Upscaler Topaz is an AI-powered video upscaling API built on Topaz Starlight generative models. You submit a video via a publicly accessible HTTPS URL (or an `upl_` upload reference), receive a task ID, then poll for results or receive a webhook notification when processing completes. The service synthesizes generative detail while upscaling to your selected resolution, and can optionally change the frame rate with AI interpolation. `starlight_precise_2_5` is the highest-quality generative upscaler and is the default. `starlight_fast_2` is a faster generative upscaler with a slight trade-off in quality. Choose Fast when turnaround time matters more than maximum fidelity. Set `target_fps` to change the output frame rate. When `target_fps` differs from the source frame rate, an interpolation model generates the intermediate frames -- `apollo` (recommended for most videos) or `chronos` (optimized for slow-motion and large frame-rate changes). If you omit `target_fps`, the source frame rate is kept and no interpolation is applied. Frame interpolation requires a target of at least 15 fps. **Video Upscaler Topaz** uses Topaz Starlight generative models that synthesize new detail while upscaling, and adds AI frame interpolation. **[Video Upscaler Precision](/api-reference/video/video-upscaler-precision/overview)** focuses on faithful, non-generative upscaling with a `strength` blend between original and upscaled output. Choose Topaz for generative enhancement and frame-rate changes; choose Precision to stay true to the original without AI-generated additions. Video Upscaler Topaz supports four output resolutions: `720p`, `1k`, `2k` (default), and `4k`. The source aspect ratio is always preserved. Higher resolutions increase processing time. See the Pricing page for details on each resolution tier. Rate limits vary by subscription tier. See the Rate Limits page for current limits by plan. Pricing depends on the video and the output resolution you select. See the Pricing page for current rates and subscription options. ## Best practices * **Input quality**: Start from the highest-quality source video available. Heavily compressed inputs may have artifacts amplified during upscaling * **Model selection**: Use `starlight_precise_2_5` for maximum quality; switch to `starlight_fast_2` when turnaround time is the priority * **Frame interpolation**: Use `apollo` for most footage; choose `chronos` for slow-motion or large frame-rate increases. Keep `target_fps` at 15 or above for interpolation * **Noise tuning**: Start with the default of 0 and increase only if the source needs additional noise handling * **Resolution selection**: Choose 2K (default) for most use cases as a good balance of quality and processing time. Use 4K when the target display or delivery format requires it * **Production integration**: Use webhooks instead of polling for scalable, event-driven applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Video Upscaler Precision](/api-reference/video/video-upscaler-precision/overview)**: Faithful, non-generative video upscaling with a strength blend and FPS boost * **[Video Upscaler](/api-reference/video/video-upscaler/overview)**: Standard video upscaler with creativity controls, processing flavors, and a Turbo endpoint * **[VFX](/api-reference/video/vfx/overview)**: Apply cinematic visual effects to videos with professional filters # Video Upscaler Topaz - Get task status Source: https://docs.magnific.com/api-reference/video/video-upscaler-topaz/task-by-id get /v1/ai/video-upscaler-topaz/{task-id} Returns the current status and output URL of a specific video upscaler topaz task. The output URL is included only if the task has completed successfully. # Video Upscaler Topaz - Upscale video Source: https://docs.magnific.com/api-reference/video/video-upscaler-topaz/upscale-video post /v1/ai/video-upscaler-topaz Upscales a video with Topaz Starlight generative enhancement. Supports output resolutions of 720p, 1k, 2k, or 4k, an optional frame-rate change with AI frame interpolation, and noise control. # Video Upscaler Topaz - List tasks Source: https://docs.magnific.com/api-reference/video/video-upscaler-topaz/video-upscaler-topaz-tasks get /v1/ai/video-upscaler-topaz Returns a list of all video upscaler topaz tasks. Each task includes its ID, current status, and output URL if completed. # AI Video Upscaling - Magnific API Source: https://docs.magnific.com/api-reference/video/video-upscaler/overview Upscale videos to 720p, 1K, 2K, or 4K with the Video Upscaler API. Standard and Turbo endpoints with creativity, FPS boost, sharpening, and smart grain controls. Video Upscaler enhances video resolution up to 4K with AI-powered detail enhancement, FPS boost, and fine-grained controls for sharpness and texture. Video Upscaler is an AI-powered video upscaling API that increases video resolution while intelligently enhancing details and improving visual quality. Built on the same Magnific technology used for image upscaling, this API extends those capabilities to video content with additional controls for frame rate enhancement and processing speed. The API offers two processing modes via separate endpoints: **Standard** for maximum quality and **Turbo** for faster processing with premium quality automatically applied. Both support output resolutions up to 4K with granular control over creativity, sharpness, grain, and processing style. ### Key capabilities * **Resolution up to 4K**: Upscale videos to 720p, 1K, 2K, or 4K output resolution * **Creativity control**: Adjustable creativity intensity (0-100) to balance fidelity vs enhancement * **FPS boost**: Increase video frame rate for smoother motion * **Sharpen control**: Adjustable sharpening intensity (0-100) for edge definition and clarity * **Smart grain**: Intelligent grain/texture enhancement (0-100) for natural film grain preservation * **Processing flavors**: Two modes: `vivid` for enhanced colors and detail, `natural` for faithful reproduction * **Standard and Turbo endpoints**: Standard for maximum quality control, Turbo for faster processing with premium quality built in * **Frame-based pricing**: Cost calculated per frame, varying by output resolution * **Async processing**: Webhook notifications or polling for task completion * **URL-based input**: Accepts a publicly accessible HTTPS URL pointing to the video ### Use cases * **Film restoration**: Upscale classic footage to modern resolutions while preserving the original aesthetic * **Content repurposing**: Enhance low-resolution social media clips for large-screen display or broadcast * **Video production**: Upscale B-roll and archival footage to match project resolution requirements * **E-commerce**: Enhance product videos with higher resolution and sharpness for better customer experience * **Surveillance**: Improve clarity of security footage for review and analysis * **Gaming content**: Upscale gameplay recordings and streaming highlights to 4K * **Educational content**: Enhance older educational videos and training materials for modern displays ### Standard endpoint Submit a publicly accessible HTTPS video URL to create a new upscaling task. The service returns a task ID for async polling or webhook notification.
Create a new video upscaling task List all video upscaler tasks Get task status and results by ID
### Turbo endpoint Turbo processing with premium quality automatically applied. Faster processing while maintaining high visual quality. Uses the same parameters as the standard endpoint. Turbo tasks appear in the same task list and can be retrieved with the same get-task endpoint as standard tasks.
Create a turbo video upscaling task
### Parameters Both standard and turbo endpoints accept the same request body: | Parameter | Type | Required | Default | Description | | ------------- | --------- | -------- | ------- | --------------------------------------------------------------------------------------------- | | `video` | `string` | Yes | - | Source video as a publicly accessible HTTPS URL | | `creativity` | `integer` | No | `0` | Creativity intensity (0-100). Higher values introduce more AI-generated detail | | `resolution` | `string` | No | `2k` | Output resolution: `720p`, `1k`, `2k`, or `4k` | | `fps_boost` | `boolean` | No | `false` | Boost video frame rate for smoother motion | | `sharpen` | `integer` | No | `0` | Sharpening intensity (0-100). Higher values increase edge definition | | `smart_grain` | `integer` | No | `0` | Grain/texture enhancement (0-100). Adds natural film grain effect | | `flavor` | `string` | No | `vivid` | Processing style: `vivid` for enhanced colors and detail, `natural` for faithful reproduction | | `webhook_url` | `string` | No | - | URL for completion notification with task status and output URL | ### Pricing Video Upscaler uses frame-based pricing. The cost per frame varies depending on the output resolution selected. Key pricing factors: * **Number of frames**: Total frame count in the input video determines base cost * **Output resolution**: 4K costs more per frame than 2K, 1K, or 720p * **FPS boost**: Enabling frame rate boost increases the total frame count and cost * **Endpoint**: Standard and Turbo have different per-frame rates **Choosing between Standard and Turbo:** Use the Turbo endpoint (`/v1/ai/video-upscaler/turbo`) for faster processing with premium quality automatically applied. Use the Standard endpoint (`/v1/ai/video-upscaler`) when you need maximum control over processing options. See the [Pricing page](/pricing) for current rates. ### Rate limits | Tier | Requests per day | | ------ | ---------------- | | Free | 10 | | Tier 1 | 125 | For higher rate limits, [contact our sales team](https://www.magnific.com/api#contact) with details about your use case and expected traffic. ## Frequently Asked Questions Video Upscaler is an AI-powered video upscaling API that increases video resolution while enhancing details. You submit a video via a publicly accessible HTTPS URL, receive a task ID, then poll for results or receive a webhook notification when processing completes. The API analyzes each frame and applies intelligent upscaling with your chosen parameters. Video Upscaler supports four output resolutions: `720p`, `1k`, `2k` (default), and `4k`. Choose based on your target display or delivery requirements. Higher resolutions increase processing time and cost per frame. The **Standard** endpoint (`/v1/ai/video-upscaler`) provides full control over upscaling parameters. The **Turbo** endpoint (`/v1/ai/video-upscaler/turbo`) delivers faster processing with premium quality automatically applied. Both accept the same request parameters. Choose Turbo for high-volume workflows or when faster turnaround is needed, and Standard when you want maximum quality control. Video Upscaler pricing is calculated per frame. The cost per frame varies by output resolution (720p, 1K, 2K, or 4K). The total cost equals the number of frames in your video multiplied by the per-frame rate for your chosen resolution. Enabling FPS boost increases the total frame count and therefore the cost. See the [Pricing page](/pricing) for current per-frame rates. **Vivid** enhances colors and adds more detail during upscaling, producing visually striking results. **Natural** preserves the original color palette and characteristics for faithful reproduction. Choose vivid for marketing and creative content, natural for documentary or archival work. The creativity parameter (0-100, default 0) controls how much AI-generated detail is introduced during upscaling. At 0, the upscaler focuses on faithful resolution enhancement. Higher values allow the AI to infer and add new details, similar to the creative image upscaler. Start with low values and increase gradually. FPS boost increases the video frame rate for smoother motion. This is especially useful for older footage with low frame rates, creating a more fluid viewing experience on modern displays. Note that FPS boost increases the total frame count, which affects pricing. Processing time depends on video duration, input resolution, output resolution, and selected options. The Turbo endpoint significantly reduces processing time compared to Standard. Use webhooks for production workflows to receive notifications when processing completes, rather than polling. Free tier allows 10 requests per day. Tier 1 allows 125 requests per day. Rate limits apply across both Standard and Turbo endpoints. For higher limits, contact our sales team. ## Best practices * **Input quality**: Start from the highest-quality source video available. Avoid heavily compressed inputs as artifacts may be amplified * **Resolution selection**: Choose the output resolution based on your delivery needs. 2K is a good balance of quality and cost for most use cases * **Creativity tuning**: Start with creativity at 0 for faithful upscaling. Increase gradually if you want the AI to enhance and infer additional detail * **Flavor selection**: Use `vivid` for marketing and social media content, `natural` for documentary, archival, or clinical footage * **Standard vs Turbo**: Use the Turbo endpoint for faster turnaround with premium quality. Use Standard for full quality control * **Smart grain for realism**: Increase smart\_grain for footage that should retain a filmic quality rather than appearing digitally smooth * **Production integration**: Use webhooks instead of polling for scalable, event-driven applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors during high traffic ## Related APIs * **[Upscaler Creative](/api-reference/image-upscaler-creative/image-upscaler)**: Prompt-guided image upscaling with creative detail generation and stylization * **[Upscaler Precision V2](/api-reference/image-upscaler-precision-v2/overview)**: High-fidelity image upscaling with granular controls for sharpness, texture, and detail * **[VFX](/api-reference/video/vfx/overview)**: Apply cinematic visual effects to videos with professional filters # Video Upscaler - Get task status Source: https://docs.magnific.com/api-reference/video/video-upscaler/task-by-id get /v1/ai/video-upscaler/{task-id} Returns the current status and output URL of a specific video upscaler task. The output URL is included only if the task has completed successfully. # Video Upscaler - Upscale video Source: https://docs.magnific.com/api-reference/video/video-upscaler/upscale-video post /v1/ai/video-upscaler Upscales a video while enhancing visual quality and resolution. Supports various output resolutions (720p, 1k, 2k, 4k) with optional creativity, sharpening, and grain controls. # Video Upscaler Turbo - Upscale video Source: https://docs.magnific.com/api-reference/video/video-upscaler/upscale-video-turbo post /v1/ai/video-upscaler/turbo Upscales a video using turbo processing with premium quality enhancement applied automatically. Turbo mode reduces processing time while maintaining high visual quality. Supports various output resolutions (720p, 1k, 2k, 4k) with optional creativity, sharpening, and grain controls. # Video Upscaler - List tasks Source: https://docs.magnific.com/api-reference/video/video-upscaler/video-upscaler-tasks get /v1/ai/video-upscaler Returns a list of all video upscaler tasks. Each task includes its ID, current status, and output URL if completed. # Download a video by option id. Source: https://docs.magnific.com/api-reference/videos/download-an-option-video get /v1/videos/{id}/options/{option-id}/download # Download a video by ID. Source: https://docs.magnific.com/api-reference/videos/download-an-video get /v1/videos/{id}/download # Search and filter videos by specified order Source: https://docs.magnific.com/api-reference/videos/get-all-videos-by-order get /v1/videos # Get detailed video information by ID Source: https://docs.magnific.com/api-reference/videos/get-one-video-by-id get /v1/videos/{id} # Videos API Source: https://docs.magnific.com/api-reference/videos/videos-api Search, retrieve, and download videos programmatically with the Magnific Videos API. The Magnific Videos API lets you discover video resources with AI-powered search and download the assets for use in your applications. ### What you can do * List videos with sorting and filters * Get a single video by ID * Download a video asset ### Authentication All requests require an API key via the `x-magnific-api-key` header. See the [Quickstart](/quickstart) and [Authentication](/authentication). ### Endpoints
List videos with sorting and filters Retrieve a single video by ID Download a video asset
### Use cases * Build a video picker for your CMS * AI-powered search by keyword and sort by popularity * Programmatically download videos during build/deploy ### FAQ #### Is there a rate limit? Yes. See [Rate limits](/ratelimits). #### What license applies to videos? Refer to the [License agreement](https://www.magnific.com/legal/terms-of-use#api-services). # Voiceover - Generate speech from text Source: https://docs.magnific.com/api-reference/voiceover/generate post /v1/ai/voiceover/elevenlabs-turbo-v2-5 Generate natural-sounding speech from text using ElevenLabs AI voices. Create professional voiceovers for videos, podcasts, presentations, and more. Supports multiple languages, voice customization, and high-quality audio output. # ElevenLabs Voiceover - Text-to-Speech API Source: https://docs.magnific.com/api-reference/voiceover/overview Generate natural-sounding voiceovers from text with ElevenLabs AI. Professional speech synthesis with voice customization, multi-language support, and high-quality audio output. Powered by ElevenLabs technology, this API generates natural-sounding speech from text with customizable voice settings and multi-language support. ElevenLabs Voiceover is a text-to-speech API that converts text into natural-sounding audio using AI voice synthesis. Generate professional voiceovers for videos, podcasts, presentations, e-learning content, and accessibility applications. The API supports up to 40,000 characters per request with the Turbo model and delivers high-quality audio output suitable for production use. ### Key capabilities * **AI model**: `eleven_turbo_v2_5` optimized for fast, high-quality generation (up to 10,000 characters) * **Voice customization**: Control stability (0-1), similarity boost (0-1), and speech speed (0.7-1.2x) * **Speaker boost**: Enhanced voice clarity and presence in generated audio * **Multi-language support**: UTF-8 text including accented letters and non-Latin scripts * **Maximum text length**: Up to 40,000 characters per request * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Video production**: Create voiceovers for marketing videos, tutorials, and social media content * **Podcast production**: Generate intro/outro narration or read content aloud * **E-learning**: Convert educational materials to audio for accessibility and engagement * **Accessibility**: Provide audio versions of written content for visually impaired users * **IVR systems**: Generate professional voice prompts for phone systems * **Audiobook creation**: Convert written content into natural-sounding audio narration ### Generate voiceover with ElevenLabs Create voiceovers by submitting text to the API with a voice ID. The service returns a task ID for async polling or webhook notification.
Create a new voiceover generation task List all voiceover tasks with status Get task status and results by ID
### Parameters | Parameter | Type | Required | Default | Description | | ------------------- | --------- | -------- | --------------------- | --------------------------------------------------------------------- | | `text` | `string` | Yes | - | Text to convert to speech. UTF-8 supported, 1-40,000 characters | | `voice_id` | `string` | Yes | - | ElevenLabs voice ID from the voice library | | `model` | `string` | No | `"eleven_turbo_v2_5"` | AI model for speech synthesis | | `stability` | `number` | No | `0.5` | Voice consistency: 0.0 (expressive) to 1.0 (stable) | | `similarity_boost` | `number` | No | `0.2` | Voice matching: 0.0 (varied) to 1.0 (close match, may have artifacts) | | `speed` | `number` | No | `1.0` | Speech rate: 0.7 (30% slower) to 1.2 (20% faster) | | `use_speaker_boost` | `boolean` | No | `true` | Enable enhanced voice clarity and presence | | `webhook_url` | `string` | No | - | URL for task completion notification | ## Frequently Asked Questions ElevenLabs Voiceover is a text-to-speech API that converts text into natural-sounding audio using AI voice synthesis. You submit text with a voice ID, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The output is a high-quality audio file containing the synthesized speech. The `eleven_turbo_v2_5` model is used, optimized for fast, high-quality speech synthesis. It supports up to 10,000 characters per request and is ideal for both real-time applications and production use. Voice IDs are unique identifiers for each voice in the ElevenLabs voice library. You can find voice IDs in the ElevenLabs Voice Library. A common example voice ID is `21m00Tcm4TlvDq8ikWAM` (Rachel - a calm, professional female voice). `stability` (0-1) controls voice consistency: lower values produce more expressive, varied speech; higher values produce more consistent, stable output. `similarity_boost` (0-1) controls how closely the output matches the original voice sample: higher values match more closely but may introduce artifacts. ElevenLabs Voiceover supports multiple languages through UTF-8 text encoding, including accented letters (e.g., e, n, u) and non-Latin scripts. The specific languages available depend on the voice selected - many voices support multiple languages natively. Processing time depends on text length and model selection. The `eleven_turbo_v2_5` model is optimized for speed and generates audio faster. For production workflows with longer texts, use webhooks instead of polling to receive completion notifications. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits by tier. For high-volume production use, consider using webhooks for efficient task management. ## Best practices * **Voice selection**: Choose a voice appropriate for your content and audience * **Model**: The `eleven_turbo_v2_5` model provides fast, high-quality results for all use cases * **Stability tuning**: Start with 0.5, decrease for more expressive reads, increase for more consistent output * **Speed adjustment**: Use 0.7-0.9 for slower, clearer speech; 1.0-1.2 for faster narration * **Text preparation**: Use proper punctuation for natural pauses; avoid very long sentences * **Production integration**: Use webhooks instead of polling for scalable applications * **Error handling**: Implement retry logic with exponential backoff for 503 errors ## Related APIs * **[Sound Effects](/api-reference/sound-effects/post-sound-effects)**: Generate sound effects from text descriptions * **[Audio Isolation](/api-reference/audio-isolation/overview)**: Extract specific sounds from audio or video files * **[Lip Sync](/api-reference/lip-sync/latent-sync/overview)**: Synchronize lip movements with audio * **[OmniHuman 1.5](/api-reference/video/omni-human-1-5)**: Generate human animations driven by audio # Voiceover - Get task status Source: https://docs.magnific.com/api-reference/voiceover/task-by-id get /v1/ai/voiceover/elevenlabs-turbo-v2-5/{task-id} Get the status of one voiceover task # Voiceover - List tasks Source: https://docs.magnific.com/api-reference/voiceover/voiceover-tasks get /v1/ai/voiceover/elevenlabs-turbo-v2-5 Get the status of all voiceover tasks # WAN 2.5 Text-to-Video API Source: https://docs.magnific.com/api-reference/wan-2-5-t2v/overview Generate videos from text prompts with WAN 2.5. Choose from 480p, 720p, or 1080p resolutions with 5 or 10 second durations for creative and marketing content. WAN 2.5 is developed by Alibaba and delivers high-quality text-to-video generation with multiple resolution options and prompt expansion capabilities. WAN 2.5 Text-to-Video generates videos directly from text descriptions. Describe your scene, characters, motion, and camera movements in a prompt, and the API produces video output at your chosen resolution. Available in 480p, 720p, and 1080p variants with 5 or 10 second durations, suitable for rapid prototyping, social media content, and marketing assets. ### Key capabilities * **Multiple resolutions**: Choose between 480p, 720p, or 1080p based on quality and speed requirements * **Flexible duration**: Generate 5-second clips for fast iteration or 10-second videos for more developed action * **Prompt expansion**: AI-powered prompt optimizer expands simple ideas into detailed video scripts * **Negative prompts**: Exclude unwanted elements like blur, watermarks, or distortion from output * **Reproducible results**: Use seed values to regenerate similar videos with identical parameters * **Async processing**: Webhook notifications or polling for task completion * **Maximum prompt length**: 800 characters for main prompt, 500 characters for negative prompt ### Resolution comparison | Resolution | Best for | Processing speed | Output quality | | ---------- | ---------------------------------------------------------- | ---------------- | -------------- | | **480p** | Rapid prototyping, previews, mobile-first content | Fastest | Good | | **720p** | Social media, web content, balanced quality/speed | Medium | Better | | **1080p** | Marketing assets, professional content, high-detail scenes | Slower | Best | ### Use cases * **Social media content**: Quick video clips for TikTok, Instagram Reels, and YouTube Shorts * **Marketing previews**: Rapid concept visualization before full production * **Product demonstrations**: Animated product showcases from text descriptions * **Educational content**: Explainer videos and visual learning materials * **Creative exploration**: Experimental motion and abstract visualizations * **Storyboarding**: Visual previews for film and video pre-production ### API operations Generate videos by submitting a text prompt to the appropriate resolution endpoint. The service returns a task ID for async polling or webhook notification.
Generate 480p video from text List 480p tasks Get 480p task status
Generate 720p video from text List 720p tasks Get 720p task status
Generate 1080p video from text List 1080p tasks Get 1080p task status
### Parameters | Parameter | Type | Required | Default | Description | | ------------------------- | --------- | -------- | ------- | ------------------------------------------------------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Main description of the video including scene, characters, motion, camera moves, and style. Maximum 800 characters. | | `duration` | `string` | No | `"5"` | Video length: `"5"` seconds (faster) or `"10"` seconds (more action) | | `negative_prompt` | `string` | No | - | Elements to avoid in the output (e.g., "blurry, low quality, watermark"). Maximum 500 characters. | | `enable_prompt_expansion` | `boolean` | No | `true` | AI optimizer expands shorter prompts into detailed scripts | | `seed` | `integer` | No | Random | Seed for reproducibility (0 to 2147483647). Use same seed with identical parameters for similar results. | | `webhook_url` | `string` | No | - | URL for async completion notifications | ## Frequently Asked Questions WAN 2.5 Text-to-Video is an AI video generation API developed by Alibaba. You submit a text prompt describing your desired video, receive a task ID immediately, then poll for results or receive a webhook notification when processing completes. The model interprets your description and generates a video matching the scene, motion, and style you specified. Choose based on your use case: 480p is fastest and ideal for rapid prototyping or mobile-first content. 720p balances quality and speed, suitable for most social media and web content. 1080p delivers the highest quality for marketing assets and professional content but takes longer to process. Processing time depends on resolution and duration. 480p processes fastest, followed by 720p, then 1080p. A 5-second clip generates faster than a 10-second clip. For production workflows, use webhooks instead of polling. Be specific about scenes and visual details. Describe camera movements (zoom, pan, tilt), lighting, atmosphere, and subject actions. Example: "fluffy orange cat on wooden windowsill, looking at snow falling outside, soft warm lighting, slow camera zoom in" produces better results than "cat looking outside." When enabled (default), the AI optimizer expands shorter prompts into detailed video scripts. This is useful when you have a simple idea but want richer video output. Disable it if you want precise control over exactly what the model generates. Rate limits vary by subscription tier. See [Rate Limits](/ratelimits) for current limits. See the [Pricing page](/pricing) for current rates and subscription options. WAN 2.5 Text-to-Video offers 480p, 720p, and 1080p resolution options with prompt expansion. WAN 2.6 provides enhanced quality and is available for both text-to-video and image-to-video workflows. Choose WAN 2.5 for more resolution flexibility; choose WAN 2.6 for the latest quality improvements. ## Best practices * **Prompt writing**: Be specific about scenes, camera movements, lighting, and subject actions. Detailed prompts produce better results than vague descriptions. * **Resolution selection**: Start with 480p for rapid iteration, then switch to higher resolutions for final output. * **Duration choice**: Use 5-second clips for quick previews; 10-second clips allow more complex motion and narrative development. * **Negative prompts**: Include common issues to avoid: "blurry, low quality, watermark, text, distortion, extra limbs." * **Reproducibility**: Save the seed value if you like a result and want to generate variations with similar characteristics. * **Production integration**: Use webhooks for scalable applications instead of polling. * **Error handling**: Implement retry with exponential backoff for 503 errors during high-demand periods. ## Related APIs * **[WAN 2.6 Text-to-Video](/api-reference/text-to-video/wan-v2-6-720p)**: Latest WAN generation with enhanced quality at 720p and 1080p * **[WAN 2.5 Image-to-Video](/api-reference/image-to-video/wan-2-5-i2v-480p)**: Generate video from an existing image with WAN 2.5 * **[WAN 2.6 Image-to-Video](/api-reference/image-to-video/wan-v2-6-720p)**: Latest WAN generation for image-to-video at 720p and 1080p * **[LTX 2 Pro Text-to-Video](/api-reference/text-to-video/ltx-2-pro)**: Alternative text-to-video model with different style characteristics # WAN 2.5 T2V 1080p - Get task status Source: https://docs.magnific.com/api-reference/wan-2-5-t2v/task-by-id-1080p get /v1/ai/text-to-video/wan-2-5-t2v-1080p/{task-id} Get the status of a WAN 2.5 Text-to-Video 1080p task # WAN 2.5 T2V 480p - Get task status Source: https://docs.magnific.com/api-reference/wan-2-5-t2v/task-by-id-480p get /v1/ai/text-to-video/wan-2-5-t2v-480p/{task-id} Get the status of a WAN 2.5 Text-to-Video 480p task # WAN 2.5 T2V 720p - Get task status Source: https://docs.magnific.com/api-reference/wan-2-5-t2v/task-by-id-720p get /v1/ai/text-to-video/wan-2-5-t2v-720p/{task-id} Get the status of a WAN 2.5 Text-to-Video 720p task # WAN 2.6 Video – Image-to-Video & Text-to-Video API Source: https://docs.magnific.com/api-reference/wan-v2-6-i2v/overview Generate high-quality AI videos from images or text with WAN 2.6. 720p and 1080p resolutions, 5-15 second durations, multi-shot sequences, and prompt expansion for cinematic results. WAN 2.6 is Alibaba's latest video generation model, delivering smooth motion, high visual fidelity, and advanced features like multi-shot sequences and AI prompt expansion. WAN 2.6 is a versatile AI video generation API that supports both image-to-video (i2v) and text-to-video (t2v) workflows. It generates cinematic videos with smooth motion, strong temporal consistency, and detailed visuals at 720p or 1080p resolution. The model offers durations of 5, 10, or 15 seconds and includes advanced features like prompt expansion and multi-shot composition for richer narratives. ### Key capabilities * **Dual input modes**: Generate video from an image (i2v) or purely from text (t2v) * **Resolution options**: 720p (1280x720) and 1080p (1920x1080) in landscape or portrait * **Flexible durations**: 5, 10, or 15 second video outputs * **Prompt expansion**: AI optimizer expands short prompts into detailed scripts for richer output * **Multi-shot sequences**: Break prompts into multiple shots for narrative depth (requires prompt expansion) * **Negative prompts**: Exclude unwanted elements like watermarks, blur, or distortion * **Reproducible results**: Fixed seed support for consistent generation * **Async processing**: Webhook notifications or polling for task completion ### Use cases * **Marketing videos**: Create product showcases and brand content from images or descriptions * **Social media content**: Generate short-form videos for TikTok, Instagram, and YouTube * **Concept visualization**: Transform static designs or ideas into motion * **Storyboarding**: Use multi-shot mode to prototype narrative sequences * **Educational content**: Illustrate concepts with AI-generated video explanations * **Creative exploration**: Experiment with text prompts to generate unique visual content ### Generation modes | Mode | Input | Best for | | ------------------------ | ------------------ | -------------------------------------------------------------------- | | **Image-to-Video (i2v)** | Image URL + prompt | Animating existing images, product videos, bringing artwork to life | | **Text-to-Video (t2v)** | Text prompt only | Creating videos from scratch, concept exploration, narrative content | ### Resolution variants | Variant | Resolution | Orientation | Use case | | ------------------- | ---------- | ----------- | ------------------------------- | | **720p Landscape** | 1280x720 | Horizontal | Web videos, presentations | | **720p Portrait** | 720x1280 | Vertical | Social media stories, mobile | | **1080p Landscape** | 1920x1080 | Horizontal | High-quality marketing, YouTube | | **1080p Portrait** | 1080x1920 | Vertical | Premium social content, ads | ## API Operations ### Image-to-Video (i2v) Generate videos from an input image with motion guidance via prompts.
Generate 720p video from image Check 720p task status Generate 1080p video from image Check 1080p task status
### Text-to-Video (t2v) Generate videos purely from text descriptions without an input image.
Generate 720p video from text Check 720p task status Generate 1080p video from text Check 1080p task status
### Parameters | Parameter | Type | Required | Default | Description | | ------------------------- | --------- | -------- | ---------- | ------------------------------------------------------------------- | | `prompt` | `string` | Yes | - | Scene description, motion, camera moves, style. Max 2000 characters | | `image` | `string` | i2v only | - | URL of the keyframe image to animate (JPEG, PNG, WebP) | | `size` | `string` | No | `1280*720` | Output size: `1280*720`, `720*1280`, `1920*1080`, `1080*1920` | | `duration` | `string` | No | `5` | Video length: `5`, `10`, or `15` seconds | | `negative_prompt` | `string` | No | - | Elements to avoid (e.g., "blurry, watermark"). Max 1000 characters | | `enable_prompt_expansion` | `boolean` | No | `false` | Enable AI to expand prompts into detailed scripts | | `shot_type` | `string` | No | `single` | `single` for continuous shot, `multi` for scene transitions | | `seed` | `integer` | No | `-1` | Random seed for reproducibility (-1 for random) | | `webhook_url` | `string` | No | - | URL for async status notifications | ## Frequently Asked Questions WAN 2.6 is Alibaba's video generation model that creates AI videos from images (i2v) or text prompts (t2v). It produces smooth, cinematic videos up to 15 seconds at 720p or 1080p resolution with features like multi-shot sequences and prompt expansion for richer narratives. Image-to-video (i2v) animates an existing image you provide, giving you control over the visual starting point. Text-to-video (t2v) generates videos purely from text descriptions, creating visuals entirely from your prompt without an input image. WAN 2.6 supports 5, 10, and 15 second durations. Shorter durations (5s) generate faster, while longer durations (15s) allow for more developed scenes and narratives. WAN 2.6 offers 720p (1280x720) and 1080p (1920x1080) in both landscape and portrait orientations. Use 720p for faster generation and 1080p for higher quality output. Prompt expansion uses AI to transform short, simple prompts into detailed scripts before generation. Enable it when you have a basic idea but want richer, more cinematic output. It is required for multi-shot mode. Multi-shot mode (`shot_type: multi`) breaks your prompt into multiple scene transitions, creating a narrative sequence instead of a single continuous shot. It requires `enable_prompt_expansion: true`. Generation time varies by resolution, duration, and server load. Typical processing ranges from 30 seconds to several minutes. Use webhooks for production workflows instead of polling. Rate limits depend on your subscription tier. See the [Rate Limits](/ratelimits) page for current limits by plan. Pricing varies by resolution and duration. See the [Pricing](/pricing) page for current rates and credit costs. Image-to-video accepts JPEG, PNG, and WebP images via publicly accessible URLs. Use high-quality images with clear subjects for best results. ## Best practices * **Prompt writing**: Be specific about scenes, camera movements (zoom, pan, tilt), lighting, and atmosphere for better results * **Image selection (i2v)**: Use high-resolution images with clear subjects and balanced lighting; avoid compressed or noisy inputs * **Negative prompts**: Always include common artifacts to avoid: "blurry, low quality, watermark, text, distortion" * **Duration selection**: Start with 5 seconds for quick iterations, then increase to 10-15s for final outputs * **Prompt expansion**: Enable for short prompts or when you want the AI to add cinematic details * **Multi-shot planning**: For multi-shot mode, structure your prompt with clear scene descriptions * **Production integration**: Use webhooks for scalable applications instead of polling * **Reproducibility**: Save the `seed` value from successful generations to recreate similar results ## Related APIs * **[WAN 2.5](/api-reference/image-to-video/wan-2-5/task-by-id-720p)**: Previous WAN version with 480p, 720p, and 1080p options * **[Kling 2.5 Turbo Pro](/api-reference/image-to-video/kling-v2.5-pro/overview)**: Alternative i2v model with cinematic quality * **[Kling Pro v2.1](/api-reference/image-to-video/kling-v2.1-pro/overview)**: High-fidelity i2v with expressive motion * **[PixVerse V5](/api-reference/image-to-video/pixverse/overview)**: Fast i2v with style consistency # WAN 2.6 T2V 1080p - Get task status Source: https://docs.magnific.com/api-reference/wan-v2-6-t2v/task-by-id-1080p get /v1/ai/text-to-video/wan-v2-6-1080p/{task-id} Get the status of a WAN 2.6 Text-to-Video 1080p task # WAN 2.6 T2V 720p - Get task status Source: https://docs.magnific.com/api-reference/wan-v2-6-t2v/task-by-id-720p get /v1/ai/text-to-video/wan-v2-6-720p/{task-id} Get the status of a WAN 2.6 Text-to-Video 720p task # Authentication Source: https://docs.magnific.com/authentication Learn how to authenticate your requests to the Magnific API ## API Key Authentication Magnific API uses API keys to authenticate requests. You need to include your API key in the header of every API request to access the Magnific resources. Currently, private API keys are the only way to authenticate with the Magnific API. This means that only server-to-server calls can be made to the API All API endpoints are authenticated using API keys and picked up from the specification file.
Download the OpenAPI specification file
## Using the API Key Include your API key in the `x-magnific-api-key` header of your HTTP requests: ```bash theme={null} curl -H "x-magnific-api-key: YOUR_API_KEY" https://api.magnific.com/v1/resources ``` ```javascript JavaScript theme={null} const axios = require('axios'); const response = await axios.get('https://api.magnific.com/v1/resources', { headers: { 'x-magnific-api-key': 'YOUR_API_KEY' } }); ``` ```python Python theme={null} import requests headers = { 'x-magnific-api-key': 'YOUR_API_KEY' } response = requests.get('https://api.magnific.com/v1/resources', headers=headers) ``` ```ruby Ruby theme={null} require 'net/http' require 'uri' uri = URI.parse('https://api.magnific.com/v1/resources') request = Net::HTTP::Get.new(uri) request['x-magnific-api-key'] = 'YOUR_API_KEY' response = Net::HTTP.start(uri.hostname, uri.port, use_ssl: uri.scheme == 'https') do |http| http.request(request) end ``` ## Rate limiting Be aware that API requests are subject to [rate limiting](/ratelimits). The specific limits may vary based on your account type and agreement with Magnific. Always check the API response headers for rate limit information. # Changelog Source: https://docs.magnific.com/changelog/2024 This section documents changes to the Magnific API, including all patch updates, bug fixes, new version releases, version deprecations, versions unsets, etc. ### Mystic AI: Enhanced High-Resolution Image Generation We've made significant improvements to our Mystic Image Generation mode, offering advanced text-to-image generation with higher resolution outputs and fine-tuned controls. * **POST /v1/ai/mystic**: Enhanced high-resolution image generation with advanced controls [Go to the docs](/api-reference/mystic/post-mystic) ### Text-to-Image Classic Fast Model: Image Size Parameter Updates We've made important changes to the image size parameters for our Text-to-Image Classic Fast model for generating images with AI. These updates provide more precise aspect ratio information and improve clarity for our users. * **POST /v1/ai/text-to-image**: Updated with new image size parameters and aspect ratio controls [Go to the docs](/api-reference/text-to-image) ### Image Relight on Magnific API We're excited to announce the launch of our new Image Relight feature on the API! This powerful feature allows you to transform the lighting of your images using advanced AI algorithms. * **POST /v1/ai/image-relight**: Transform scene lighting using text prompts, reference images, or custom light maps [Go to the docs](/api-reference/image-relight/post-image-relight) ### Image Upscaler on Magnific API We're excited to announce the launch of our new Image Upscaler API! This powerful tool allows you to enhance your images using advanced AI algorithms, providing higher resolution and improved quality. * **POST /v1/ai/image-upscaler**: Upscale images to 2x, 4x, 8x, or 16x their original size [Go to the docs](/api-reference/image-upscaler-creative/post-image-upscaler) # Changelog Source: https://docs.magnific.com/changelog/2025 This section documents changes to the Magnific API, including all patch updates, bug fixes, new version releases, version deprecations, versions unsets, etc. ### AI Skin Enhancer Modes (Creative, Faithful, Flexible) Add AI skin enhancement endpoints for portraits with Creative, Faithful, and Flexible modes, returning task-based results for simple async integration. * **POST /v1/ai/skin-enhancer/flexible**: Skin enhancer using AI (Flexible) * **POST /v1/ai/skin-enhancer/faithful**: Skin enhancer using AI (Faithful) * **POST /v1/ai/skin-enhancer/creative**: Skin enhancer using AI (Creative) [Go to the docs](/api-reference/skin-enhancer) ### Upscale With Precision An Image With Magnific Get ready for something amazing! We're thrilled to announce our new upscale with precision an image with magnific that will revolutionize how you enhance your images with AI-powered upscaling in your applications. * **POST /v1/ai/image-upscaler-precision-v2**: Upscale with precision an image with Magnific V2 [Go to the docs](/api-reference/image-upscaler-precision-v2) ### Generate Sound Effects From Text We're delighted to unveil our new generate sound effects from text! This game-changing addition to our API makes generate content with AI-powered technology more accessible and powerful than ever before. * **POST /v1/ai/sound-effects**: Generate sound effects from text [Go to the docs](/api-reference/sound-effects) ### Improve Prompt We've added 3 new endpoints to enhance our API capabilities across multiple services for improving AI generation prompts. * **POST /v1/ai/improve-prompt**: Improve AI generation prompts [Go to the docs](/api-reference/improve-prompt) ### Image To Prompt We've added 3 new endpoints to enhance our API capabilities across multiple services for generating AI prompts from images. * **POST /v1/ai/image-to-prompt**: Generate AI prompt from image [Go to the docs](/api-reference/image-to-prompt) ### Text to Image - Flux Pro 1.1 We've added 6 new endpoints to enhance our API capabilities across multiple services for text-to-image generation and lip-sync video generation. * **POST /v1/ai/text-to-image/flux-pro-v1-1**: Create image from text - Flux pro 1.1 [Go to the docs](/api-reference/text-to-image) ### Image to Video - Kling 2.5 Pro We've added 3 new endpoints to enhance our API capabilities across multiple services for creating videos from images. * **POST /v1/ai/image-to-video/kling-v2-5-pro**: Create a video from an image - Kling 2.5 pro model [Go to the docs](/api-reference/image-to-video) ### Image to Video - PixVerse V5 Transition We've added 3 new endpoints to enhance our API capabilities across multiple services for video transitions. * **POST /v1/ai/image-to-video/pixverse-v5-transition**: Video transition using PixVerse-V5 [Go to the docs](/api-reference/image-to-video) ### Image to Video - PixVerse V5 We've added 6 new endpoints to enhance our API capabilities across multiple services for video extensions. * **POST /v1/ai/image-to-video/pixverse-v5**: Video extension using PixVerse-V5 [Go to the docs](/api-reference/image-to-video) ### Text to Image - Seedream V4 We've added 6 new endpoints to enhance our API capabilities across multiple services for text-to-image generation and editing. * **POST /v1/ai/text-to-image/seedream-v4**: Create image from text - Seedream v4 [Go to the docs](/api-reference/text-to-image) ### Text to Image - Gemini 2.5 Flash We've added 3 new endpoints to enhance our API capabilities across multiple services for image generation and editing. * **POST /v1/ai/gemini-2-5-flash-image-preview**: Image generation and editing using Gemini 2.5 Flash [Go to the docs](/api-reference/text-to-image) ### Text to Image - Seedream We've added 12 new endpoints to enhance our API capabilities across multiple services for text-to-image generation and image-to-video conversion. * **POST /v1/ai/text-to-image/seedream**: Create image from text - Seedream These endpoints were removed on August 27, 2026 and are no longer available. Use [Seedream 5.0 Pro](/api-reference/text-to-image/post-seedream-v5-pro), [Seedream 5.0 Lite](/api-reference/text-to-image/post-seedream-v5-lite) or [Seedream 4.5](/api-reference/text-to-image/post-seedream-v4-5) instead. [Go to the docs](/api-reference/text-to-image) ### Text to Image - Seedream We've added 3 new endpoints to enhance our API capabilities across multiple services for text-to-image generation. * **POST /v1/ai/text-to-image/seedream**: Create image from text - Seedream These endpoints were removed on August 27, 2026 and are no longer available. Use [Seedream 5.0 Pro](/api-reference/text-to-image/post-seedream-v5-pro), [Seedream 5.0 Lite](/api-reference/text-to-image/post-seedream-v5-lite) or [Seedream 4.5](/api-reference/text-to-image/post-seedream-v4-5) instead. [Go to the docs](/api-reference/text-to-image) ### Text to Image - HyperFlux We've added 3 new endpoints to enhance our API capabilities across multiple services for image generation. * **POST /v1/ai/text-to-image/hyperflux**: Image generation using HyperFlux [Go to the docs](/api-reference/text-to-image) ### Image Upscaler Precision We've added 3 new endpoints to enhance our API capabilities across multiple services for precision image upscaling. * **POST /v1/ai/image-upscaler-precision**: Upscales an image while adding new visual elements or details [Go to the docs](/api-reference/image-upscaler-precision/post-image-upscaler-precision) ### Image to Video - MiniMax Hailuo-02 We've added 6 new endpoints to enhance our API capabilities across multiple services for creating videos from text/images. * **POST /v1/ai/image-to-video/minimax-hailuo-02-1080p**: Create a video from text/image - MiniMax Hailuo-02 1080p [Go to the docs](/api-reference/image-to-video) ### Text to Icon We've added 2 new endpoints to enhance our API capabilities across multiple services for AI icon generation. * **POST /v1/ai/text-to-icon/preview**: AI Icon generation [Go to the docs](/api-reference/icon-generation/post-preview) ### Image to Video - Kling 2.1 Master We've added 3 new endpoints to enhance our API capabilities across multiple services for creating videos from images. * **POST /v1/ai/image-to-video/kling-v2-1-master**: Create a video from an image - Kling 2.1 master model [Go to the docs](/api-reference/image-to-video) # Changelog Source: https://docs.magnific.com/changelog/2026 This section documents changes to the Magnific API, including all patch updates, bug fixes, new version releases, version deprecations, versions unsets, etc. ### Seedance 2.x: prompt limit raised to 40,000 characters Every Seedance 2.x model now accepts prompts of up to 40,000 characters, twenty times the previous limit of 2,000. Longer prompts cost no more to run. * **Applies to**: Seedance 2.0 Pro (480p, 720p, 1080p and 4K), Seedance 2.0 Fast, Seedance 2.0 Mini and Seedance 2.5 Pro * **Multishot**: the per-shot `prompt` of a Seedance 2.5 Pro sequence shares the new limit * **Quality note**: beyond roughly 5,000 characters the results may become less predictable, so the extra room is best spent on detail that matters * **Unchanged**: Seedance 1.5 Pro and Seedance 1.0 keep their 2,000-character limit [Go to the docs](/api-reference/video/seedance-2-pro/overview) ### Seedance 2.0: reference videos Seedance 2.0 Pro, Fast and Mini now accept **video** references, alongside the reference images they already supported: * **`reference_videos`** (new): up to 3 items, each a publicly accessible HTTPS URL or a video upload `file_id` (`upl_vid_...`) * Reference them in the prompt as **`@Video1`**, `@Video2`, `@Video3`, matching the 1-indexed array order — the same convention as `@ImageN` * Each video must be **MP4 or MOV**, 2 to 15 seconds, 24 to 60 fps, at most 50 MB, between 409600 and 2086876 pixels, and with an aspect ratio from 0.4 to 2.5 * The **combined** duration of all reference videos must not exceed 15 seconds Billing is unchanged: credits scale with the requested `duration`, and reference videos do not affect the price. [Seedance 2.0 Pro docs](/api-reference/video/seedance-2-pro/overview) ### Seedance 2.0: two input combinations now return 400 instead of failing later Both combinations below have always been rejected **by the model provider**, which means the request was accepted, credits were consumed, and the failure arrived in the webhook minutes later. They are now validated up front and answered with a `400` naming the field, so nothing is charged: * **Keyframes and visual references are mutually exclusive.** `image` or `image_end` cannot be combined with `reference_images` or `reference_videos`. Send keyframes for a first/last-frame video, or references for a reference-to-video — not both. * **`image_end` requires `image`.** A last frame with no first frame is not a valid request. Send both to generate a transition, or drop `image_end`. If you were sending either shape, those requests were already failing. The change is that you now find out immediately, and at no cost. [Seedance 2.0 Pro docs](/api-reference/video/seedance-2-pro/overview) ### Seedance 2.0 Mini: extended durations and audio control Seedance 2.0 Mini now matches Pro and Fast: * **`duration`**: any integer from `4` to `15` seconds (previously limited to `5` or `10`) * **`sound_effects`** (new, default `true`): set to `false` to generate a silent video * **`no_music`** (new, default `false`): set to `true` to suppress background music while keeping dialogue and sound effects Billing is unchanged: credits scale with the requested duration, and audio does not affect the price. [Seedance 2.0 Mini docs](/api-reference/video/seedance-2-mini/overview) ### GPT Image 2 - Text-to-Image and Image Editing API Generate and edit images with OpenAI GPT Image 2, the image model known for close prompt following and legible text rendered inside the image. Pick the output with `resolution` (`1k`/`2k`/`4k`) and `aspect_ratio`, choose the cost/quality trade-off with one of three quality tiers, and get up to 10 images from a single request. * **POST /v1/ai/text-to-image/gpt-image-2**: Generate images from a text prompt * **POST /v1/ai/text-to-image/gpt-image-2-edit**: Edit existing images from a text instruction Both endpoints also expose a task list and a task-by-id endpoint for polling. **Features:** * **Three quality tiers**: `quality` accepts `low`, `medium` and `high`, changing both render quality and cost per image * **Batch up to 10 images**: `num_images` returns up to 10 images or edited variants per request, each billed individually * **In-image text**: renders legible words and short phrases inside the image, for mockups, posters and packaging * **Prompt-driven editing**: the edit endpoint applies changes described in natural language, with no mask required * **Async by default**: every request returns a task id, with webhook or polling delivery [Go to the docs](/api-reference/text-to-image/gpt-image-2/overview) ### Seedream (legacy) API removal The legacy `/v1/ai/text-to-image/seedream` endpoints have been removed from the Magnific API documentation and are no longer available. Use [Seedream 5.0 Pro](/api-reference/text-to-image/post-seedream-v5-pro), [Seedream 5.0 Lite](/api-reference/text-to-image/post-seedream-v5-lite) or [Seedream 4.5](/api-reference/text-to-image/post-seedream-v4-5) for text-to-image generation. ### Video Upscaler Topaz API Upscale videos with Topaz Starlight generative AI. Choose the Starlight Precise 2.5 or Fast 2 model, output at 720p, 1K, 2K, or 4K, optionally change the frame rate with AI interpolation (Apollo or Chronos), and control noise. Accepts an HTTPS URL or an uploaded file reference. * **POST /v1/ai/video-upscaler-topaz**: Create a Topaz upscaling task **Features:** * **Starlight models**: `starlight_precise_2_5` (highest quality) or `starlight_fast_2` (faster) * **Resolution up to 4K**: `720p`, `1k`, `2k` (default), or `4k`, aspect ratio preserved * **AI frame interpolation**: set `target_fps` and pick `apollo` or `chronos` (interpolation needs at least 15 fps) * **Noise control**: `noise` intensity from `0` to `1` (default `0`) * Async polling or webhook notifications [Go to the docs](/api-reference/video/video-upscaler-topaz/overview) ### Veo 3.1 Image-to-Video: start and end frame Google Veo 3.1 image-to-video now accepts an optional last frame, so you can define both ends of the shot and let the model interpolate the transition between them. * **`image_end`** (new, optional): last frame as a Base64 string or HTTPS URL. Veo 3.1 animates from `image` towards it * Available on both **POST /v1/ai/image-to-video/veo-3-1** and **POST /v1/ai/image-to-video/veo-3-1-fast**, at every resolution including 4K * `image` is still required — the model needs a first frame to interpolate from, so sending only `image_end` returns a `400` * Both frames may use different formats and independent sources (URL or Base64) Billing is unchanged: credits still scale with the requested duration, and the end frame does not affect the price. [Go to the docs](/api-reference/image-to-video/veo-3-1/overview) ### Seedance 2.5 Pro - Text, Image, Reference & Multishot to Video API Generate AI videos with Seedance 2.5 Pro, the latest Seedance video model. Create from a text prompt, animate a first-frame image (with optional last-frame transition), drive generation with subject/style/audio references, or storyboard up to six shots with multishot. Seedance 2.5 Pro adds native audio, camera-motion presets, and durations from 4 to 30 seconds. Resolution is selected at generation time via dedicated 480p, 720p, and 1080p endpoints; task listing and status are per resolution. * **POST /v1/ai/video/seedance-2-5-pro-480p**: Generate a 480p video * **POST /v1/ai/video/seedance-2-5-pro-720p**: Generate a 720p video * **POST /v1/ai/video/seedance-2-5-pro-1080p**: Generate a 1080p video * **GET /v1/ai/video/seedance-2-5-pro-\{resolution}**: List tasks for a resolution * **GET /v1/ai/video/seedance-2-5-pro-\{resolution}/\{task-id}**: Get task status and result **Features:** * **Four generation modes**: text-to-video, image-to-video (first/last frame), reference-to-video (`reference_images`, `reference_audios` cited as `@Image1`, `@Audio1`, …), and `multishot` storyboards of up to 6 shots * **Native audio**: sound effects and ambience via `sound_effects`, with optional `no_music` * **Camera motion**: 50+ `camera_motion` presets, plus `camera_fixed` for static shots * **Flexible durations**: any integer from `4` to `30` seconds (default `5`) * **Seven aspect ratios** and selectable `output_format` (`mp4` or `mov`) * Async polling or webhook notifications; result delivered via a URL valid for 24 hours [Go to the docs](/api-reference/video/seedance-2-5-pro/overview) ### Seedance 2.0 Pro & Fast: extended durations and audio control Seedance 2.0 Pro and Seedance 2.0 Fast now support more flexible video generation: * **`duration`**: any integer from `4` to `15` seconds (previously limited to `5` or `10`) * **`sound_effects`** (new, default `true`): set to `false` to generate a silent video with no native audio track. Disabling audio also avoids audio-related content moderation blocks * **`no_music`** (new, default `false`): set to `true` to suppress background music while keeping dialogue and sound effects Billing is unchanged: credits scale with the requested duration, and audio does not affect the price. [Seedance 2.0 Pro docs](/api-reference/video/seedance-2-pro/overview) · [Seedance 2.0 Fast docs](/api-reference/video/seedance-2-fast/overview) ### Google Imagen 3 & Imagen 4 API removal The Google Imagen 3, Imagen 4 Fast, and Imagen 4 Ultra endpoints have been removed from the Magnific API documentation and are no longer available. Use [Nano Banana Pro](/api-reference/text-to-image/nano-banana-pro-flash/overview) or [Gemini 2.5 Flash Image Preview](/api-reference/text-to-image/post-gemini-2-5-flash-image-preview) for text-to-image generation. ### Audit Logs API Pull your enterprise team's audit trail programmatically with the new Audit Logs API. Track who did what, to which resource, and how access changed — ready to stream into your SIEM or compliance pipeline. * **GET /v1/analytics/audit-logs/events**: Search audit events with filters and cursor pagination **Features:** * Rich filters: actor, group, folder, event type, request ID, and date window * Cursor pagination with no depth limit to walk your entire history * Server-side project and folder name resolution * Customer-safe payloads: free-form event data narrowed to a vetted allow-list * Enterprise-only, admin/owner access; no credit cost [Go to the docs](/api-reference/audit-logs/overview) ### Google Veo 3.1 Lite - Text to Video & Image to Video API Veo 3.1 Lite is the cost-efficient variant of Google Veo 3.1. It generates cinematic videos from a text prompt or animates a source image, at 720p or 1080p resolution with optional native audio. Ideal for prototyping, iteration, and high-volume workflows. **Text to Video:** * **POST /v1/ai/text-to-video/veo-3-1-lite**: Create a video from text * **GET /v1/ai/text-to-video/veo-3-1-lite**: List all Veo 3.1 Lite text-to-video tasks * **GET /v1/ai/text-to-video/veo-3-1-lite/\{task-id}**: Get task status and results **Image to Video:** * **POST /v1/ai/image-to-video/veo-3-1-lite**: Create a video from an image * **GET /v1/ai/image-to-video/veo-3-1-lite**: List all Veo 3.1 Lite image-to-video tasks * **GET /v1/ai/image-to-video/veo-3-1-lite/\{task-id}**: Get task status and results **Features:** * Required `prompt` (max 20,000 chars); image-to-video also requires an `image` (Base64 or HTTPS URL) * Resolutions `720p` and `1080p` (default `720p`); no 4K on the Lite variant * Durations of `4`, `6`, or `8` seconds (default `8`); aspect ratios `16:9` or `9:16` (default `16:9`) * Optional synchronized audio via `generate_audio` (default `true`), plus `negative_prompt` and `seed` * Audio and video delivered via URL; async polling or `webhook_url` notifications **Use cases:** Cost-efficient marketing and social clips, animating product photos and concept art, and rapid video prototyping. [Text to Video](/api-reference/text-to-video/veo-3-1-lite/overview) · [Image to Video](/api-reference/image-to-video/veo-3-1-lite/overview) ### Google Lyria & Lyria 3 - Text to Music API Two new Google music generation models. Google Lyria generates fixed 30-second instrumental tracks from a text prompt. Lyria 3 adds selectable `clip` and `pro` model variants and optional reference images to guide the mood and style of the composition. Both deliver the generated track as an audio URL. **Google Lyria:** * **POST /v1/ai/music-generation/google-lyria**: Generate a 30-second track from text * **GET /v1/ai/music-generation/google-lyria**: List all Google Lyria tasks * **GET /v1/ai/music-generation/google-lyria/\{task-id}**: Get task status and results **Lyria 3:** * **POST /v1/ai/music-generation/lyria-3**: Generate a track from text * **GET /v1/ai/music-generation/lyria-3**: List all Lyria 3 tasks * **GET /v1/ai/music-generation/lyria-3/\{task-id}**: Get task status and results **Features:** * **Google Lyria**: required `prompt` (1-2500 chars), optional `negative_prompt` and `seed`, fixed 30-second duration * **Lyria 3**: required `prompt` (10-2000 chars), `model` variant `clip` (default) or `pro`, and optional `reference_images` (PNG, JPEG, or WebP via public URL or GCS) * Audio delivered via URL; async polling or `webhook_url` notifications **Use cases:** Background music for video and games, music beds for podcasts and social content, and original tracks for ads. [Google Lyria docs](/api-reference/music-generation/google-lyria/overview) · [Lyria 3 docs](/api-reference/music-generation/lyria-3/overview) ### PixVerse Text to Video & Transitions - V5, V5.5 & V6 API Five new PixVerse video endpoints. Text to Video generates a video directly from a text prompt with no input image across V5, V5.5, and V6. The Transition endpoints for V5.5 and V6 interpolate a video between a required first and last frame, guided by a prompt. **Text to Video:** * **POST /v1/ai/text-to-video/pixverse-v5**: Create a video from text (PixVerse V5) * **POST /v1/ai/text-to-video/pixverse-v5-5**: Create a video from text (PixVerse V5.5) * **POST /v1/ai/text-to-video/pixverse-v6**: Create a video from text (PixVerse V6) **Transitions:** * **POST /v1/ai/image-to-video/pixverse-v5-5-transition**: First-to-last frame transition (PixVerse V5.5) * **POST /v1/ai/image-to-video/pixverse-v6-transition**: First-to-last frame transition (PixVerse V6) Each model also exposes `GET` endpoints to list tasks and to retrieve a task by `\{task-id\}`. **Features:** * **V5**: 5 or 8 second durations, resolutions up to 1080p, five aspect ratios, and visual styles * **V5.5**: adds native synchronized audio (`generate_audio_switch`), multi-clip output (`generate_multi_clip_switch`), prompt reasoning (`thinking_type`), and a 10-second duration * **V6**: flexible 1-15 second durations plus extra aspect ratios (`portrait_2_3`, `landscape_3_2`, `cinematic_21_9`) on text to video * **Transitions**: required `first_image_url` and `end_image_url`, 20 camera movement presets, native audio, and prompt reasoning; V6 Transition adds multi-clip output and flexible 1-15s durations * Async tasks with polling or `webhook_url` notifications **Use cases:** Text-driven marketing and social clips, cinematic sequences, scene-to-scene transitions, and product reveals. [V6 Text to Video](/api-reference/text-to-video/pixverse-v6/overview) · [V6 Transition](/api-reference/image-to-video/pixverse-v6-transition/overview) ### Flux Kontext Max & FLUX.2 Flex - Image Editing & Text to Image API Two new Black Forest Labs models. Flux Kontext Max is the highest-quality tier of the Kontext family for context-aware image editing, delivering maximum prompt adherence and detail from a reference image and a text prompt. FLUX.2 Flex is a professional-grade text-to-image model with fine-grained control over guidance, steps, and a wide custom resolution range, plus optional image-to-image input. **Flux Kontext Max:** * **POST /v1/ai/text-to-image/flux-kontext-max**: Edit an image from a reference image and text prompt * **GET /v1/ai/text-to-image/flux-kontext-max**: List all Flux Kontext Max tasks * **GET /v1/ai/text-to-image/flux-kontext-max/\{task-id}**: Get task status and results **FLUX.2 Flex:** * **POST /v1/ai/text-to-image/flux-2-flex**: Create an image from text * **GET /v1/ai/text-to-image/flux-2-flex**: List all FLUX.2 Flex tasks * **GET /v1/ai/text-to-image/flux-2-flex/\{task-id}**: Get task status and results **Features:** * **Flux Kontext Max**: required `input_image` reference plus `prompt`, 10 aspect ratios (`square_1_1` default), guidance (1-10), steps (1-100), and safety tolerance (0-6) * **FLUX.2 Flex**: custom `width` and `height` from 256 to 1920px (default 1024x768), up to 4 optional Base64 input images for image-to-image, guidance (1-10), and steps (1-100) * `prompt_upsampling` for creative prompt enhancement and reproducible output with `seed` * PNG or JPEG output; async polling or `webhook_url` notifications **Use cases:** Context-aware product and campaign edits, large-format and print imagery, image-to-image refinement, and professional creative workflows. [Flux Kontext Max docs](/api-reference/text-to-image/flux-kontext-max/overview) · [FLUX.2 Flex docs](/api-reference/text-to-image/flux-2-flex/overview) ### Seedream 5.0 Pro - Text to Image & Image Editing API Generate and edit high-quality images with Seedream 5.0 Pro, ByteDance's flagship image model. The text-to-image endpoint delivers high-fidelity detail, strong prompt adherence, and superior composition, while the edit endpoint applies text-guided edits across up to 10 reference images, preserving subject details, lighting, and color tone. * **POST /v1/ai/text-to-image/seedream-v5-pro**: Create image from text * **GET /v1/ai/text-to-image/seedream-v5-pro**: List all Seedream 5.0 Pro tasks * **GET /v1/ai/text-to-image/seedream-v5-pro/\{task-id}**: Get task status and results * **POST /v1/ai/text-to-image/seedream-v5-pro-edit**: Edit images with text guidance and reference images * **GET /v1/ai/text-to-image/seedream-v5-pro-edit**: List all Seedream 5.0 Pro edit tasks * **GET /v1/ai/text-to-image/seedream-v5-pro-edit/\{task-id}**: Get edit task status and results **Features:** * Resolution tiers: `2k` (base output, up to \~4.19M pixels, 2048x2048 at 1:1) and `1.5k` (economical, capped at \~2.36M pixels, 1536x1536 at 1:1); default `2k` * 8 aspect ratios: `square_1_1`, `widescreen_16_9`, `social_story_9_16`, `portrait_2_3`, `traditional_3_4`, `standard_3_2`, `classic_4_3`, `cinematic_21_9` * Image editing with 1 to 10 reference images as public URLs or Base64 (JPG, JPEG, PNG; min 256x256, max 10MB each) * Deterministic generation with `seed` (0 - 4,294,967,295) * Async tasks with polling or `webhook_url` notifications **Use cases:** Professional and large-format imagery, detailed scenes and complex compositions, cinematic and artistic visuals, and consistent multi-image reference editing. [Create from text](/api-reference/text-to-image/post-seedream-v5-pro) · [Edit images](/api-reference/text-to-image/post-seedream-v5-pro-edit) ### Seedance 2.0 Pro & Seedance 2.0 Fast - Text to Video & Image to Video API Generate AI videos with Seedance 2.0 Pro, the highest-quality variant of the Seedance 2.0 family, and Seedance 2.0 Fast, its speed-optimized counterpart. Both support text-to-video and image-to-video with optional first and last frame transitions. Resolution is selected at generation time via dedicated endpoints — Pro at 480p, 720p, and 1080p; Fast at 480p and 720p — while task listing and status use a single generic endpoint per model. **Seedance 2.0 Pro:** * **POST /v1/ai/video/seedance-2-pro-480p**: Generate a 480p video from text or image * **POST /v1/ai/video/seedance-2-pro-720p**: Generate a 720p video from text or image * **POST /v1/ai/video/seedance-2-pro-1080p**: Generate a 1080p video from text or image * **GET /v1/ai/video/seedance-2-pro**: List all Seedance 2.0 Pro tasks * **GET /v1/ai/video/seedance-2-pro/\{task-id}**: Get task status and result **Seedance 2.0 Fast:** * **POST /v1/ai/video/seedance-2-fast-480p**: Generate a 480p video from text or image * **POST /v1/ai/video/seedance-2-fast-720p**: Generate a 720p video from text or image * **GET /v1/ai/video/seedance-2-fast**: List all Seedance 2.0 Fast tasks * **GET /v1/ai/video/seedance-2-fast/\{task-id}**: Get task status and result **Features:** * **Text-to-video and image-to-video**: text `prompt` always required; add a first-frame `image` (and optional last-frame `image_end`) for image-driven generation and transitions * **Resolution by generate endpoint**: dedicated endpoints per model — Pro at `480p`, `720p`, and `1080p`; Fast at `480p` and `720p` * **Generic task tracking**: one list endpoint and one status endpoint per model, for tasks of any resolution * **Durations**: `5` or `10` seconds (default `5`) * **Seven aspect ratios** from ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21` (default `widescreen_16_9`) * **Camera control** with `camera_fixed` and reproducible output with `seed` * Async polling or webhook notifications; result MP4 delivered via a URL valid for 24 hours [Go to the Pro docs](/api-reference/video/seedance-2-pro/overview) · [Go to the Fast docs](/api-reference/video/seedance-2-fast/overview) ### Seedance 2.0 Mini - Text to Video & Image to Video API Generate affordable AI videos with Seedance 2.0 Mini, the most affordable and fastest variant of the Seedance 2.0 family. Create videos from a text prompt, animate a first-frame image, or transition between a first and last frame. Resolution is selected at generation time via dedicated 480p and 720p endpoints, while task listing and status use a single generic endpoint. * **POST /v1/ai/video/seedance-2-mini-480p**: Generate a 480p video from text or image * **POST /v1/ai/video/seedance-2-mini-720p**: Generate a 720p video from text or image * **GET /v1/ai/video/seedance-2-mini**: List all Seedance 2.0 Mini tasks * **GET /v1/ai/video/seedance-2-mini/\{task-id}**: Get task status and result **Features:** * **Affordable, fast generation** for high-volume and cost-sensitive workflows * **Text-to-video and image-to-video**: text `prompt` always required; add a first-frame `image` (and optional last-frame `image_end`) for image-driven generation and transitions * **Resolution by generate endpoint**: dedicated `480p` and `720p` endpoints * **Generic task tracking**: one list endpoint and one status endpoint for tasks of any resolution * **Durations**: `5` or `10` seconds (default `5`) * **Seven aspect ratios** from ultra-wide `film_horizontal_21_9` to ultra-tall `film_vertical_9_21` (default `widescreen_16_9`) * **Camera control** with `camera_fixed` and reproducible output with `seed` * Async polling or webhook notifications; result MP4 delivered via a URL valid for 24 hours [Go to the docs](/api-reference/video/seedance-2-mini/overview) ### Creations API Retrieve a user's most recent creations programmatically, enriched with each creation's metadata. Identity is resolved automatically from the API key, results are paginated newest-first, and it does not consume credits. * **GET /v1/creations/recent**: List the user's most recent creations **Features:** * Pagination via `page` and `per_page` * Filter by tool with `tool_name` * Server-side identity resolution (no cross-user access) [Go to the docs](/api-reference/creations/overview) ### Kling 3.0 Turbo - Text to Video & Image to Video API Generate AI videos fast with Kling 3.0 Turbo, the speed-optimized model in the Kling 3.0 family. Create videos from a text prompt or animate a single first-frame image, with flexible 3-15 second durations. Resolution is selected at generation time via dedicated 720p and 1080p endpoints, while task listing and status use a single generic endpoint per mode. **Text to Video** * **POST /v1/ai/text-to-video/kling-v3-turbo-720p**: Generate a 720p video from a text prompt * **POST /v1/ai/text-to-video/kling-v3-turbo-1080p**: Generate a 1080p video from a text prompt * **GET /v1/ai/text-to-video/kling-v3-turbo**: List all Kling 3.0 Turbo T2V tasks * **GET /v1/ai/text-to-video/kling-v3-turbo/\{task-id}**: Get T2V task status and result **Image to Video** * **POST /v1/ai/image-to-video/kling-v3-turbo-720p**: Generate a 720p video from a first-frame image * **POST /v1/ai/image-to-video/kling-v3-turbo-1080p**: Generate a 1080p video from a first-frame image * **GET /v1/ai/image-to-video/kling-v3-turbo**: List all Kling 3.0 Turbo I2V tasks * **GET /v1/ai/image-to-video/kling-v3-turbo/\{task-id}**: Get I2V task status and result **Features:** * **Fast Turbo generation** for high-volume workflows and rapid iteration * **Resolution by generate endpoint**: dedicated `720p` and `1080p` endpoints per mode * **Generic task tracking**: one list endpoint and one status endpoint per mode, for tasks of any resolution * **Durations**: any whole number from `3` to `15` seconds (default `5`) * **Aspect ratios** (text-to-video): `16:9`, `9:16`, and `1:1` * **Image-to-video** accepts a public URL or Base64 first-frame image with optional prompt-guided motion * Async polling or webhook notifications; result MP4 delivered via a URL valid for 24 hours [Go to the docs](/api-reference/text-to-video/kling-v3-turbo/overview) ### Analytics API Track your team's AI credit consumption over time with the new Analytics API. Query usage with per-user, per-project, and per-group breakdowns using day, week, month, or year granularity. * **POST /v1/analytics/team-credit-usage**: Query team credit consumption with time-series aggregation * **GET /v1/analytics/team-members**: List team members with email, role, and status * **GET /v1/analytics/team-api-keys**: List API keys with name and status * **GET /v1/analytics/team-groups**: List team groups with member count * **GET /v1/analytics/team-projects**: List team projects with name **Features:** * Time-series aggregation: day, week, month, or year granularity * Per-user breakdown with email addresses, per-project and per-group context * Filter by group IDs, project references, or API keys * Discovery endpoints to retrieve filter values for your team [Go to the docs](/api-reference/analytics/overview) ### PixVerse V6 - Image to Video API Generate cinematic videos from a single image or interpolate a transition between two keyframes with PixVerse V6. Builds on PixVerse V5.5 with flexible 1-15 second durations (any integer length) and built-in transition mode via the new optional `last_frame_image` field. * **POST /v1/ai/image-to-video/pixverse-v6**: Generate a video from an image (or first/last frame pair) with PixVerse V6 * **GET /v1/ai/image-to-video/pixverse-v6**: List all PixVerse V6 tasks * **GET /v1/ai/image-to-video/pixverse-v6/\{task-id}**: Get PixVerse V6 task status and result **What's new vs PixVerse V5.5:** * **Flexible duration**: any integer value from `1` to `15` seconds (previously fixed to `5`, `8`, or `10`) * **Transition mode** via the new optional `last_frame_image` field: generate a smooth interpolation between `image_url` (first frame) and `last_frame_image` (last frame) in a single request, replacing the separate transition endpoint * Preserves V5.5 features: native synchronized audio (`generate_audio_switch`), multi-clip output (`generate_multi_clip_switch`), prompt reasoning (`thinking_type`) * Same resolutions (`360p`, `540p`, `720p`, `1080p`), aspect ratios, styles, and camera presets as V5.5 * Async polling or webhook notifications on completion [Go to the docs](/api-reference/image-to-video/pixverse-v6/overview) ### PixVerse V5.5 - Image to Video API Generate cinematic videos from a single image with PixVerse V5.5. Builds on PixVerse V5 with a new 10-second duration option (up to 720p), native synchronized audio, multi-clip output with dynamic camera changes, and a prompt reasoning mode. * **POST /v1/ai/image-to-video/pixverse-v5-5**: Generate a video from an image with PixVerse V5.5 * **GET /v1/ai/image-to-video/pixverse-v5-5**: List all PixVerse V5.5 tasks * **GET /v1/ai/image-to-video/pixverse-v5-5/\{task-id}**: Get PixVerse V5.5 task status and result **What's new vs PixVerse V5:** * **Durations of 5, 8, or 10 seconds** (10s only up to 720p; 1080p capped at 5 or 8 seconds) * **Native synchronized audio** via `generate_audio_switch` (music, SFX, or dialogue with the video) * **Multi-clip output with dynamic cameras** via `generate_multi_clip_switch` * **Prompt reasoning** via `thinking_type` (`enabled`, `disabled`, `auto`) * Same resolutions (`360p`, `540p`, `720p`, `1080p`), aspect ratios, styles, and camera presets as V5 * Async polling or webhook notifications on completion [Go to the docs](/api-reference/image-to-video/pixverse-v5-5/overview) ### Happy Horse 1.0 - Video Generation, Reference, and Editing API Generate AI videos with Happy Horse 1.0, Alibaba ATH's flagship video model, ranked **#1 on the Artificial Analysis Video Arena (April 2026)** for both text-to-video and image-to-video. Four endpoint categories: text-to-video, image-to-video, reference-to-video (up to 9 character references), and video editing with optional audio control. * **POST /v1/ai/text-to-video/happy-horse-1**: Generate video from a text prompt (3-15 seconds) * **POST /v1/ai/image-to-video/happy-horse-1**: Animate a single image into video (3-15 seconds) * **POST /v1/ai/reference-to-video/happy-horse-1**: Generate video from up to 9 reference images, referenced as `character1`..`character9` in the prompt (3-15 seconds) * **POST /v1/ai/video-edit/happy-horse-1**: Edit videos with natural-language instructions and up to 5 reference images via `@Image1`..`@Image5` tokens (output capped at 15 seconds) **Features:** * State-of-the-art quality: #1 on Artificial Analysis Video Arena (April 2026) for T2V and I2V * Resolutions: 720P (1280x720) and 1080P (1920x1080) * 5 aspect ratios: `16:9`, `9:16`, `1:1`, `4:3`, `3:4` * Durations: 3 to 15 seconds (default 5) * R2V supports up to 9 character references with optional bottom-right "Happy Horse" watermark * Video Edit supports `audio_setting` (`auto` or `origin`) to control audio handling * Reproducible generation via `seed` (0 to 2147483647) * Webhook notifications and async polling **Use cases:** Marketing videos, social media content, multi-character narratives, branded campaigns with consistent characters, video editing and style transfer, storyboarding. [Go to the docs](/api-reference/text-to-video/happy-horse-1/overview) ### Kling 4K I2V – Video Generation API Generate 4K videos from an input image with Kling 4K I2V. Async jobs return a task ID for polling or webhooks, with duration and prompt control for 4K output. * **POST /v1/ai/video/kling-4k-i2v**: Generate video from an image with Kling 4K I2V [Go to the docs](/api-reference/video/post-kling-4k-i2v) ### Kling 4K T2V – Video Generation API Generate 4K videos from a text prompt with Kling 4K T2V. Async jobs return a task ID for polling or webhooks, with duration and prompt control for 4K output. * **POST /v1/ai/video/kling-4k-t2v**: Generate video from text with Kling 4K T2V [Go to the docs](/api-reference/video/post-kling-4k-t2v) ### Google Imagen 4 API (Fast & Ultra) Generate high-quality images from text with Google Imagen 4. Two variants: **Imagen 4 Fast** for speed-optimized and cost-effective generation, and **Imagen 4 Ultra** for highest-quality professional output. Both variants share the same parameter schema, multilingual prompt support (9 languages), 5 aspect ratios, configurable person-generation and safety controls, PNG/JPEG output, and deterministic generation via seed. * **POST /v1/ai/text-to-image/imagen4-fast**: Create image with Imagen 4 Fast * **GET /v1/ai/text-to-image/imagen4-fast**: List all Imagen 4 Fast tasks * **GET /v1/ai/text-to-image/imagen4-fast/\{task-id}**: Get Imagen 4 Fast task status and results * **POST /v1/ai/text-to-image/imagen4-ultra**: Create image with Imagen 4 Ultra * **GET /v1/ai/text-to-image/imagen4-ultra**: List all Imagen 4 Ultra tasks * **GET /v1/ai/text-to-image/imagen4-ultra/\{task-id}**: Get Imagen 4 Ultra task status and results **Features:** * Two quality tiers: Fast (speed/cost-optimized) and Ultra (maximum quality) * Multilingual prompts: `en`, `es`, `pt`, `ja`, `ko`, `hi`, `zh`, `zh-CN`, `auto` * Aspect ratios: `square_1_1`, `social_story_9_16`, `widescreen_16_9`, `traditional_3_4`, `classic_4_3` * Person generation controls: `dont_allow`, `allow_adult`, `allow_all` * Safety settings: from `block_low_and_above` to `block_none` * Output formats: PNG or JPEG with configurable compression * Deterministic generation with `seed` (1 - 4,294,967,295) * LLM-based prompt rewriting via `enhance_prompt` **Use cases:** Marketing creatives, e-commerce imagery, editorial illustrations, localized content, rapid prototyping with Fast and final production assets with Ultra. These endpoints were removed on August 20, 2026 and are no longer available. Use [Nano Banana Pro](/api-reference/text-to-image/nano-banana-pro-flash/overview) or [Gemini 2.5 Flash Image Preview](/api-reference/text-to-image/post-gemini-2-5-flash-image-preview) instead. ### WAN 2.7 - Video Generation, Editing, and Reference API Generate AI videos with WAN 2.7, Alibaba's latest video model. Four endpoint categories: text-to-video, image-to-video (with video continuation), reference-to-video for character-consistent generation, and video editing with style transfer. All categories support 720P and 1080P resolution, audio-guided generation, and automatic prompt expansion. * **POST /v1/ai/text-to-video/wan-2-7**: Generate video from a text prompt (2-15 seconds) * **POST /v1/ai/image-to-video/wan-2-7**: Generate video from image, first+last frame, or extend a video (2-15 seconds) * **POST /v1/ai/reference-to-video/wan-2-7**: Generate video featuring characters from reference images/videos (2-10 seconds) * **POST /v1/ai/video-edit/wan-2-7**: Edit an existing video with text instructions, reference images, or style transfer (720P/1080P) [T2V docs](/api-reference/text-to-video/wan-2-7/overview) | [I2V docs](/api-reference/image-to-video/wan-2-7/overview) | [R2V docs](/api-reference/reference-to-video/wan-2-7/overview) | [Video Edit docs](/api-reference/video-edit/wan-2-7/overview) ### Veed Fabric 1.0 and Veed Fabric 1.0 Fast - Lip Sync API Generate realistic talking videos from a portrait image and audio file with Veed Fabric 1.0. Two variants available: standard for highest quality and Fast for reduced generation time. Output at 720p or 480p resolution in MP4 format. * **POST /v1/ai/lip-sync/veed-fabric-1-0**: Generate a lip-synced talking video * **POST /v1/ai/lip-sync/veed-fabric-1-0-fast**: Generate a lip-synced talking video (faster processing) * **GET /v1/ai/lip-sync/veed-fabric-1-0**: List all Veed Fabric 1.0 tasks * **GET /v1/ai/lip-sync/veed-fabric-1-0-fast**: List all Veed Fabric 1.0 Fast tasks [Veed Fabric 1.0 docs](/api-reference/lip-sync/veed-fabric-1-0/overview) | [Veed Fabric 1.0 Fast docs](/api-reference/lip-sync/veed-fabric-1-0-fast/overview) ### Video Upscaler Precision - AI Video Upscaling API AI diffusion-based precision video upscaling with faithful detail recovery. Supports output resolutions of 1K, 2K, and 4K with adjustable sharpening, smart grain, and upscaling strength controls. FPS boost available for smoother motion. * **POST /v1/ai/video-upscaler-precision**: Create a precision video upscaling task * **GET /v1/ai/video-upscaler-precision**: List all precision upscaler tasks * **GET /v1/ai/video-upscaler-precision/\{task-id}**: Get task status and results [Go to the docs](/api-reference/video/video-upscaler-precision/overview) ### Kling 3 Motion Control - Video Generation API Transfer motion from reference videos to character images with Kling 3 Motion Control. Preserves character appearance while applying motion patterns from 3-30 second reference videos. Available in Pro and Standard tiers. * **POST /v1/ai/video/kling-v3-motion-control-pro**: Generate motion-controlled video with Kling 3 Pro * **POST /v1/ai/video/kling-v3-motion-control-std**: Generate motion-controlled video with Kling 3 Standard [Go to the docs](/api-reference/video/kling-v3-motion-control/overview) ### Sound Effects API Search, filter, and download royalty-free sound effects from the Magnific catalog. Browse 42 categories including ambience, foley, and transitions with full-text search, category filtering, and flexible sorting. * **GET /v1/sfx**: Search and filter sound effects * **GET /v1/sfx/download**: Download a sound effect audio file [Go to the docs](/api-reference/sfx/overview) ### Music API Search, filter, and download royalty-free music from the Magnific Music catalog. Filter by genre, mood, and premium status with sorting by relevance, popularity, duration, or tempo. * **GET /v1/music**: Search and filter music * **GET /v1/music/download**: Download a music audio file [Go to the docs](/api-reference/music/overview) ### Nano Banana Pro Flash - Text To Image API Generate images from text with Nano Banana Pro Flash, powered by Google's Gemini 3.1 Flash model. Faster generation with Google Search grounding for real-world accuracy, reference image support (up to 3 images), 10 aspect ratios, and resolutions up to 4K. * **POST /v1/ai/text-to-image/nano-banana-pro-flash**: Create a new image generation task * **GET /v1/ai/text-to-image/nano-banana-pro-flash**: List all Nano Banana Pro Flash tasks * **GET /v1/ai/text-to-image/nano-banana-pro-flash/task-id**: Get task status and results by ID [Go to the docs](/api-reference/text-to-image/nano-banana-pro-flash/overview) ### Video Upscaler Turbo Endpoint and Frame-Based Pricing New dedicated Turbo endpoint for Video Upscaler with faster processing and premium quality automatically applied. Pricing model updated to frame-based billing that varies by output resolution. The `turbo` and `premium_quality` parameters have been removed from the standard endpoint in favor of the separate Turbo path. Turbo tasks use the same list and get-task endpoints as standard tasks. * **POST /v1/ai/video-upscaler/turbo**: Create a turbo video upscaling task [Go to the docs](/api-reference/video/video-upscaler/overview) ### Runway Gen 4.5 - Video Generation API Generate high-quality AI videos from text prompts or images with Runway Gen 4.5. Supports both text-to-video and image-to-video workflows with async task processing, polling, and webhook notifications. * **POST /v1/ai/text-to-video/runway-4-5**: Generate video from a text prompt * **POST /v1/ai/image-to-video/runway-4-5**: Generate video from an image [Go to the docs](/api-reference/video/runway-gen-4-5/overview) ### Change Camera - Image Perspective API Transform the camera angle and perspective of any image with AI. Control horizontal rotation (0-360 degrees), vertical tilt (-30 to 90 degrees), and zoom level (0-10) to generate multi-angle views from a single photo. * **POST /v1/ai/image-change-camera**: Create a new camera angle transformation task [Go to the docs](/api-reference/image-change-camera/overview) ### Seedream V4.5 – Image Expand API Outpaint and expand images with Seedream V4.5 by setting per-edge pixel growth. Optional prompt guidance with async tasks, polling, and webhooks support. * **POST /v1/ai/image-expand/seedream-v4-5**: Expand an image beyond its boundaries using Seedream V4.5 outpainting [Go to the docs](/api-reference/image-expand/post-seedream-v4-5) ### Ideogram Image Edit – Inpainting API Edit images with Ideogram inpainting using masks and prompts. Choose TURBO/DEFAULT/QUALITY modes, MagicPrompt, and async tasks with webhooks and polling. * **POST /v1/ai/ideogram-image-edit**: Inpaint and edit an image using a mask plus a prompt with Ideogram Image Edit [Go to the docs](/api-reference/ideogram-image-edit/post-ideogram-image-edit) ### Ideogram Image Expand API Expand images beyond their original boundaries with AI-powered outpainting using the Ideogram model. Control expansion independently on each edge (left, right, top, bottom) up to 2048 pixels, with optional prompt guidance and auto-prompt generation. * **POST /v1/ai/image-expand/ideogram**: Create a new image expansion task [Go to the docs](/api-reference/image-expand/ideogram/overview) ### Kling 3 - Video Generation API Generate AI videos with Kling 3, the latest video generation model from Kuaishou. * **POST /v1/ai/video/kling-v3-pro**: Generate video with Kling 3 Pro * **POST /v1/ai/video/kling-v3-std**: Generate video with Kling 3 Standard [Go to the docs](/api-reference/video/kling-v3/overview) ### Google Veo 3.1 Reference-to-Video API Generate videos with consistent characters and objects using reference images. Maintain visual identity across scenes for storytelling and multi-scene projects. * **POST /v1/ai/reference-to-video/veo-3-1**: Create video with reference images * **GET /v1/ai/reference-to-video/veo-3-1**: List all reference-to-video tasks * **GET /v1/ai/reference-to-video/veo-3-1/task-id**: Get task status and results **Features:** * Character/object consistency using 1-3 reference images * Multi-resolution output: 720p, 1080p, or 4K * Native audio generation with dialogue and sound effects * Fixed 8-second duration at 24 FPS * Aspect ratios: 16:9 (landscape) or 9:16 (portrait) * Up to 20,000 character prompts **Use cases:** Multi-scene storytelling, brand mascot videos, product showcases, character animation, episodic content [Go to the docs](/api-reference/reference-to-video/veo-3-1/overview) ### Nano Banana Pro (Text-to-Image) Generate high-quality images with Google's Nano Banana Pro (Gemini 3) model. Supports reference images for guided generation, multiple aspect ratios, and resolution options. * **POST /v1/ai/text-to-image/nano-banana-pro**: Create image from text with optional reference images **Features:** * Up to 3 reference images for guided generation * Multiple aspect ratios (1:1, 16:9, 9:16, 4:3, 3:4, etc.) * Resolution options: low, medium (default), high (4K) * Webhook notifications [Go to the docs](/api-reference/text-to-image/post-nano-banana-pro) # Welcome to Magnific API Source: https://docs.magnific.com/introduction

Start Building Today

Get your API key and start building today.

Quick start!
## Powerful API tools at your fingertips
Create custom visuals from text prompts using our state-of-the-art AI models. Enhance, upscale, relight, and transform images with powerful AI-driven tools. Convert static images into dynamic videos with customizable motion effects. Create stunning icons in different styles from text prompts using our advanced AI models. Automatically detect AI-generated content with our advanced classification system. Access millions of vectors, photos, icons, and templates to enhance your applications.
## Why developers choose Magnific API
Comprehensive documentation and dedicated support for seamless integration. Our API grows with your needs, scaling seamlessly as your usage expands. Dependable infrastructure designed for high performance and scalability.
## LLM-Friendly Documentation Our documentation is also available in an LLM-friendly format, making it easy to integrate with large language models and AI tools. You can access it in two ways: * [llms.txt](https://docs.magnific.com/llms.txt) – A lightweight sitemap that lists all documentation pages. * [llms-full.txt](https://docs.magnific.com/llms-full.txt) – The full documentation in markdown format. (Note: This may exceed your model’s context window.) Additionally, any page can be accessed as Markdown by appending .md to the URL. For example: [https://docs.magnific.com/introduction](https://docs.magnific.com/introduction) → [https://docs.magnific.com/introduction.md](https://docs.magnific.com/introduction.md) ## AI Integration with MCP Want to connect Magnific to AI assistants like Claude, ChatGPT, Cursor, and more? Add the Magnific MCP server to your client and your agent can generate images, create videos, upscale assets, and browse your creations straight from the chat. No API keys, no local setup. Learn about MCP Integration # Magnific MCP Source: https://docs.magnific.com/modelcontextprotocol Magnific MCP is a remote server built on the [Model Context Protocol](https://modelcontextprotocol.io). Once you connect it to an AI assistant, your agent can generate images and video, train consistent characters, upscale assets, and browse your generation history straight from the chat. Everything runs on your Magnific account and uses your existing credits. You don't need to manage an API key. The first time a client connects, you sign in to your Magnific account and that's it. ## Getting started
Any Magnific account works. MCP tools share the same credit balance as the in-app product. Add `https://mcp.magnific.com` in your AI client's MCP settings. Sign in to Magnific in your browser to grant the client access. Ask your assistant to generate, upscale, or browse. It will pick the right Magnific tool on its own.
## Server endpoint The Magnific MCP server lives at: ```text theme={null} https://mcp.magnific.com ``` It uses the **streamable HTTP** MCP transport, so any modern MCP client can connect directly. You don't need to install a bridge or run anything locally. ## Authentication Magnific MCP uses **OAuth 2.0** with your Magnific account. The first time a client connects, it opens a browser window so you can sign in and approve access. After that, the client keeps the session and won't ask you again. ### Discovery endpoints MCP clients discover the OAuth configuration through standard well-known URLs: | Resource | URL | | ----------------------------- | ----------------------------------------------------------------- | | Protected resource metadata | `https://mcp.magnific.com/.well-known/oauth-protected-resource` | | Authorization server metadata | `https://mcp.magnific.com/.well-known/oauth-authorization-server` | Most clients pick these up automatically when they hit the MCP endpoint without a valid token, so you usually don't have to touch them. ## Connect your client Every client asks for the same handful of details. Keep these on hand before you start: | Field | Value | | ------------------ | -------------------------------------------------------- | | **Name** | `Magnific` | | **URL** | `https://mcp.magnific.com` | | **Authentication** | OAuth (sign in with your Magnific account when prompted) | 1. Open Claude and click your profile menu, then choose **Customize**. 2. Go to **Connectors** and click **Add connector**. 3. Pick **Add custom connector**. 4. Enter the **Name** (`Magnific`) and **URL** (`https://mcp.magnific.com`). 5. Finish the **OAuth** sign-in to Magnific when Claude opens it. 6. **Start chatting.** Ask Claude to generate, upscale, or look through your creations and it will reach for the right Magnific tool on its own. 1. Open ChatGPT and click the **settings icon** in the top-right corner. 2. Go to **Advanced settings** and turn **Developer mode** on. 3. Open **Apps** and click **Create app**. 4. Enter the **Name** (`Magnific`) and **URL** (`https://mcp.magnific.com`), then choose **OAuth** as the authentication method. 5. Sign in to Magnific in the OAuth window that opens. 6. **Start chatting.** ChatGPT will use Magnific whenever you ask for an image, a video, or an upscale. Developer-mode custom apps are available on Pro, Business, and Enterprise plans. Add the Magnific MCP server with a single command: ```bash theme={null} claude mcp add --transport http magnific https://mcp.magnific.com ``` The first call from inside Claude Code opens the OAuth flow in your browser. Once you approve it, the Magnific tools are available in the session. 1. Open Cursor settings and go to the **MCP** tab. 2. Click **Add new global MCP server**. 3. Paste the configuration below: ```json theme={null} { "mcpServers": { "magnific": { "url": "https://mcp.magnific.com" } } } ``` 4. Reload the server entry. Cursor will open your browser for the Magnific sign-in. For more details, see [Cursor's MCP documentation](https://docs.cursor.com/context/model-context-protocol). Other MCP-compatible clients (Windsurf, VS Code, OpenClaw, Hermes, Codex, and so on) follow the same idea. Add `https://mcp.magnific.com` as a streamable HTTP MCP server, name it `Magnific`, and finish the OAuth sign-in when it pops up. ## Available tools These are the tools the MCP server exposes. The names are stable across clients, so you can reference them directly in your prompts. ### Account * `account_balance`: current credit balance * `project_report`: overview of project usage ### Creations * `creations_search`: search your creations by query and filters * `creations_get`: fetch a single creation by identifier * `creations_show`: render creations inline in supported clients * `creations_wait`: wait for an in-progress creation to finish * `creation_status`: poll the status of an in-progress creation * `creations_request_upload` / `creations_upload` / `creations_finalize_upload`: upload an asset * `creations_move`: move a creation between folders ### Image generation and editing * `images_generate`: generate images from text and optional references * `images_generate_svg`: generate SVG output * `images_to_svg`: convert raster to SVG * `images_upscale`: Magnific upscaler * `images_crop`: smart crop * `images_resize`: resize * `images_remove_background`: alpha cutout * `images_models_list` / `images_models_show`: image model catalog ### Video generation * `video_generate`: generate video * `video_models_list` / `video_models_show`: video model catalog ### Audio * `audio_tts`: text-to-speech * `audio_voices_list` / `audio_voices_show`: voice catalog ### 3D * `models3d_generate`: generate a 3D model ### Custom references * `custom_references_create`: train a Soul character or style * `custom_references_list`: list trained references ### Folders and Spaces * `folders_list` / `folders_get` / `folders_create` / `folders_rename` / `folders_delete` * `spaces_list`: list your Spaces * `spaces_view`: inspect a single Space ### Discovery * `tools_show`: surface the picker UI for available tools The MCP server is the source of truth. The live `tools/list` response always reflects the latest set of tools available. ## Frequently asked questions Magnific runs a remote server that speaks the Model Context Protocol over streamable HTTP. Any MCP-compatible client can connect to `https://mcp.magnific.com` and call Magnific tools right from the chat. Magnific MCP works with Claude (Web and Claude Code), ChatGPT, Cursor, and any other client that supports the MCP streamable HTTP transport. That includes Windsurf, VS Code, OpenClaw, Hermes, and Codex. No. You sign in once with your Magnific account through OAuth and the client keeps the session. There's nothing to manage or rotate. MCP tools draw from the same credit balance as the rest of Magnific. Each generation costs credits based on the model and resolution you pick. See [Pricing](/pricing) for the current rates. Yes. The `creations_search`, `creations_get`, and `creations_show` tools let your agent browse your generation history and use any past image or video as the starting point for a new one. The REST API documented on this site is still fully supported for direct, programmatic integrations using API keys. The MCP server is the path for AI assistants and chat-based agents. # Pricing Source: https://docs.magnific.com/pricing ## Credit-based system Enterprise uses a flexible, credit-based system. Admins can set credit caps per API Key, monitor activity, or limit access to specific tools, giving you complete control over usage and spending. ## Credit-based API consumption Every AI generation request you make through the API — image, video, audio, upscaling, editing, and any other model — deducts credits from your organization’s balance on each call. This applies to every paid plan. If your subscription includes **"Unlimited"** allowances, note that they only cover usage inside the Magnific web app. API calls always run on the credit model, so make sure your balance is provisioned for your expected API volume. ## Stock content via API Stock downloads through the API follow a different model than AI generations: * **Premium, Premium+ and Pro plans**: stock downloads via API don’t consume credits, but they’re capped at **100 downloads per day**. Once you hit the cap, further download requests will be rejected until the next day. * **Business and Enterprise plans**: no daily cap. Stock downloads are unlimited, and each one is deducted from your credit balance like any other API call. For the full breakdown of credits charged per stock download (by content type and file quality), see [Stock content download credits →](https://www.magnific.com/ai/docs/stock-content-download-credits). If your integration needs to download more than 100 stock assets per day, you’ll need to upgrade to Business or Enterprise. [Contact sales →](https://www.magnific.com/enterprise#contact) [View detailed pricing →](https://www.magnific.com/ai/docs/ai-image-generator-credits) # Quickstart Source: https://docs.magnific.com/quickstart Ready to harness the power of Magnific API? ## Get Started Follow these steps to begin: Sign in with the Administrator's email address and go to [API Keys](https://www.magnific.com/user/organization/api-keys) in your dashboard to generate a key. Browse our [API Reference](https://docs.magnific.com/api-reference/mystic/post-mystic) to learn how to integrate Magnific into your projects. Begin experimenting with the API and unlock the full range of models and resources available to your organization. Need help or have questions? Contact our dedicated support team through our [support page](https://www.magnific.com/api#contact) for prompt assistance. Contact your designated Magnific representative to request API access. Once it's enabled, you'll be able to generate API Keys directly from your Dashboard.
# Rate limiting Source: https://docs.magnific.com/ratelimits To ensure the stability and performance of our API, we have implemented rate limiting based on IP address and API key. This is a crucial security measure to prevent abuse and overuse of the API. Here’s how our rate limiting works: Under your `API KEY`, no **daily rate limit** (Requests Per Day, RPD) applies to API endpoints, with the following exceptions: | Service | Premium RPD | | :----------------- | :---------: | | ai-image-to-prompt | 100,000 | | ai-improve-prompt | 100,000 | | ai-powered-search | 100,000 | #### Analytics API All [Analytics API](/api-reference/analytics/overview) endpoints (`/v1/analytics/*`) are limited to **10,000 RPD** and **do not consume credits**. Available exclusively for **Business** and **Enterprise** plans. The [Audit Logs API](/api-reference/audit-logs/overview) (`/v1/analytics/audit-logs/events`) has its own **50,000 RPD** budget on **Enterprise**. ### General rate limit Our general rate limit policy applies based on `IP address` and allows: * **50 hits per second** over a **5-second period**. This means that requests originating from the same IP address can make up to 50 API requests per second. If you exceed this limit, your requests may be throttled or temporarily blocked. The limit resets every 5 seconds. ### Average rate limit Additionally, we enforce an average rate limit policy based on IP address: * **10 hits per second** averaged over a **2-minute period**. This means that over a 2-minute window, the average number of requests from the same IP address should not exceed 10 per second. If you surpass this average rate, your API access might be restricted until your request rate falls back within acceptable limits. In addition to the per-second limits, we enforce a per-minute rate limit policy based on `API key`: * **30,000 requests per minute (RPM)**. This means that requests made with the same API key cannot exceed 30,000 per minute. If you surpass this limit, additional requests will be throttled or temporarily rejected until the limit resets.
If your application requires a higher rate limit than what's currently provided, we're here to help! Please [reach out to our sales](https://www.magnific.com/api#contact) team with details about your use case and expected traffic. We will set you up with a custom plan in no time.
# Upload files Source: https://docs.magnific.com/upload-files Upload your own images, videos or audio and get a ready-to-use URL to feed any AI endpoint. ## The problem Most AI endpoints take your media as a **URL** (`image_url`, `video_url`, `audio_url`, `reference_images`, …). That's easy when your file already lives on the public internet — but if it's a local file, or something you generated on the fly, you have nowhere to point those endpoints to. The **Upload Files API** solves this: you upload the file once and get back a **readable URL** (`asset_url`) that you can drop into any endpoint that accepts a URL — images, video or audio alike. `asset_url` is a plain, publicly-fetchable URL. Anywhere an endpoint asks for an `image_url` / `video_url` / `audio_url` (or a reference image), pass the `asset_url` of your upload. ## How it works Call `POST /v1/ai/uploads/request-url` with the `content_type` of each file. You get back, per file, a short-lived `upload_url` (to send the bytes to) and a readable `asset_url` (to use later). `PUT` the raw file bytes to `upload_url`, including the returned `headers`. The upload happens directly against storage — the `upload_url` is short-lived (see `expires_in`). Pass the `asset_url` as the input URL of any AI endpoint that accepts one. Re-fetch it any time from `GET /v1/ai/uploads`. ## Supported file types Pass one of these as `content_type`: | Type | MIME types | | ----- | ----------------------------------------------------------------- | | Image | `image/png`, `image/jpeg`, `image/webp` | | Video | `video/mp4`, `video/webm`, `video/quicktime` | | Audio | `audio/mpeg`, `audio/wav`, `audio/mp4`, `audio/webm`, `audio/ogg` | Maximum size: **1 GiB** per file. ## 1. Request an upload URL ```bash theme={null} curl -X POST "https://api.magnific.com/v1/ai/uploads/request-url" \ -H "x-magnific-api-key: YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "files": [ { "content_type": "image/png" } ] }' ``` You can request up to **14** files in a single call (`files` array). ```json theme={null} { "files": [ { "file_id": "upl_img_a1b2c3d4e5f6", "upload_url": "https://.../v1/ai/uploads/...", "headers": { "Content-Type": "image/png", "x-goog-content-length-range": "0,1073741824" }, "expires_in": 120, "asset_url": "https://cdn-magnific.freepik.com/uploads/.../...?token=...", "asset_url_expires_in": 86400 } ] } ``` Identifier of the uploaded file (`upl_…`). Pre-signed URL to `PUT` the raw file bytes to. Short-lived — see `expires_in`. Headers you **must** include in the `PUT` request (they are part of the URL signature). Seconds until `upload_url` expires (do the `PUT` right after requesting it). Readable URL of the uploaded file. Valid once the `PUT` completes. Use it as the input URL in any endpoint that accepts one. Expires in \~24h — don't persist it. Seconds until `asset_url` expires (counted from this response). ## 2. Upload the bytes `PUT` the file to the `upload_url`, echoing the `headers` from step 1. Do **not** send your API key here — the request goes straight to storage and is authorized by the signed URL. ```bash theme={null} curl -X PUT "PASTE_upload_url_HERE" \ -H "Content-Type: image/png" \ -H "x-goog-content-length-range: 0,1073741824" \ --data-binary "@./my-image.png" ``` The `Content-Type` you send here **must match** the `content_type` you requested in step 1, and the `x-goog-content-length-range` header is required — both are covered by the URL signature. ## 3. Use the `asset_url` Once the upload completes, pass the `asset_url` wherever an endpoint expects a URL: ```bash theme={null} curl -X POST "https://api.magnific.com/v1/ai/classifier/image" \ -H "x-magnific-api-key: YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "image": "PASTE_asset_url_HERE" }' ``` The same `asset_url` works for video and audio inputs on endpoints that accept them (`video_url`, `audio_url`, …). ## List your uploads Lost the response from step 1, or your `asset_url` expired? List your files — each item comes with a **freshly-signed** `asset_url`: ```bash theme={null} curl "https://api.magnific.com/v1/ai/uploads" \ -H "x-magnific-api-key: YOUR_API_KEY" ``` ```json theme={null} { "files": [ { "file_id": "upl_img_a1b2c3d4e5f6", "content_type": "image/png", "size_bytes": 20480, "created_at": "2026-07-03T10:00:00Z", "expires_at": "2026-07-10T10:00:00Z", "ttl_seconds": 604800, "asset_url": "https://cdn-magnific.freepik.com/uploads/.../...?token=...", "asset_url_expires_in": 86400 } ] } ``` ## Expiration Two independent clocks: * **`asset_url_expires_in`** — how long the URL itself stays valid (**\~1 day**). An expired URL returns `403` — just call `GET /v1/ai/uploads` again to get a fresh one (a new URL is signed on every listing). Don't persist the `asset_url`. * **`ttl_seconds`** — how long until the **file** becomes eligible for auto-deletion (**\~7 days**). Deletion is eventual, so a file may linger a little past this — but once it is deleted, its `asset_url` stops working. Treat uploads as a staging area, not storage. # Webhooks Source: https://docs.magnific.com/webhooks Webhooks are a powerful way to connect different systems and services. Learn about webhooks and how to integrate them securely. ### What are webhooks? Webhooks are a way for one system to send real-time data to another system. They are a powerful tool for integrating different services and automating workflows. With webhooks, you can receive notifications, updates, and data from external systems without having to poll for changes. ### How do webhooks work? Webhooks work by allowing you to register a URL with a service that supports them. When an event occurs in the service, it sends an HTTP POST request to the registered URL with relevant data. The receiving system can then process the data and take appropriate actions based on the event. ### Why use webhooks? Webhooks offer several advantages over traditional polling-based methods: * **Real-time updates**: Webhooks provide instant notifications, allowing you to react to events as they happen. * **Efficiency**: They reduce the overhead of constant requests by delivering data only when an event occurs. * **Automation**: Webhooks trigger automatic workflows, reducing manual processes and streamlining tasks. * **Seamless integration**: They facilitate easy data exchange between systems, enabling efficient communication between different platforms. ### Common use cases for webhooks Webhooks are widely used in the following scenarios: * **Notifications**: Sending real-time alerts and updates to users or systems. * **Data synchronization**: Ensuring that data remains consistent across multiple platforms. * **Workflow automation**: Initiating tasks like sending emails, updating databases, or processing transactions based on specific events. By using webhooks, you can streamline your workflows, improve efficiency, and create seamless integrations between different systems. ## Webhook security ### Why webhook security is important? Webhooks are a powerful way to connect different systems and services. They allow you to send real-time data from one system to another. However, with great power comes great responsibility. Webhooks can be a security risk if not implemented correctly, as they can be exploited by attackers to send malicious data to your system. ### Webhook security headers To ensure the integrity and authenticity of incoming webhook requests, we deliver three headers with each request: * `webhook-id`: A unique identifier for the webhook request. This helps to detect and prevent replay attacks. * `webhook-timestamp`: A timestamp indicating when the webhook request was sent. This is used to ensure that the request is recent and prevents replay attacks within a specific time window. * `webhook-signature`: A signature generated using a secret key. This is used to verify the authenticity of the request, ensuring that it was sent by a trusted source. ### Generating the string to sign for verification You must generate a content string that will be signed and verified. This content is created by concatenating the `webhook-id`, `webhook-timestamp`, and the request body with a period (`.`) separator. You can do this by following these steps: 1. **Retrieve the headers**: Extract the `webhook-id` and `webhook-timestamp` from the request headers. 2. **Access the request body**: Obtain the raw body of the webhook request. 3. **Concatenate the values**: Combine the `webhook-id`, `webhook-timestamp`, and body into a single string using the format mentioned earlier. Here is an example of how you can generate the content to sign in Python: ```python python.py theme={null} content_to_sign = f"{webhook_id}.{webhook_timestamp}.{body}" ``` ```javascript javascript.js theme={null} const contentToSign = `${webhookId}.${webhookTimestamp}.${body}`; ``` ```php php.php theme={null} $content_to_sign = "$webhook_id.$webhook_timestamp.$body"; ``` ```java java.java theme={null} String contentToSign = webhookId + "." + webhookTimestamp + "." + body; ``` ### Obtaining the secret key The secret key is a shared secret between your system and the webhook provider. It is used to generate the signature and verify the authenticity of the request. Make sure to keep the secret key secure and never expose it in your code or configuration files. You obtain your secret key when you generate your API Key at your [User Dashboard](https://www.magnific.com/user/organization/api-keys). Copy the secret key and store it securely in your system. ### Generating the signature For the webhook signature, we use HMAC-SHA256 as the hashing algorithm. You can generate the signature by following these steps: 1. Encode the secret key as bytes. 2. Obtain the HMAC-SHA256 hash as bytes of the content to sign using the secret key. 3. Encode the hash in base64 to get the signature. Here is an example of how you can generate the signature in Python: ```python python.py theme={null} import hmac import hashlib import base64 def generate_signature(secret_key, content_to_sign): secret_key_bytes = secret_key.encode() hmac_bytes = hmac.new(secret_key_bytes, content_to_sign.encode(), hashlib.sha256).digest() signature = base64.b64encode(hmac_bytes).decode() return signature ``` ```javascript javascript.js theme={null} const crypto = require('crypto'); function generateSignature(secretKey, contentToSign) { const secretKeyBytes = Buffer.from(secretKey, 'utf-8'); const hmac = crypto.createHmac('sha256', secretKeyBytes); hmac.update(contentToSign); const signature = hmac.digest('base64'); return signature; } ``` ```php php.php theme={null} function generateSignature($secretKey, $contentToSign) { $secretKey = base64_decode($secretKey); $hmac = hash_hmac( 'sha256', $contentToSign, $secretKey, true ); $signature = base64_encode($hmac); return $signature; } ``` ```java java.java theme={null} import javax.crypto.Mac; import javax.crypto.spec.SecretKeySpec; import java.util.Base64; public static String generateSignature(String secretKey, String contentToSign) throws Exception { byte[] secretKeyBytes = secretKey.getBytes("UTF-8"); Mac mac = Mac.getInstance("HmacSHA256"); SecretKeySpec secretKeySpec = new SecretKeySpec(secretKeyBytes, "HmacSHA256"); mac.init(secretKeySpec); byte[] hmacBytes = mac.doFinal(contentToSign.getBytes("UTF-8")); String signature = Base64.getEncoder().encodeToString(hmacBytes); return signature; } ``` The obtained signature must be compared with the `webhook-signature` header in the incoming request to verify the authenticity of the request. If the signatures match, the request is considered valid, and you can process it further. The `webhook-signature` header is composed of a list of space-delimited signatures and their corresponding version identifiers. This allows you to rotate the secret key without breaking existing webhook integrations. For example, the header might look like this: ``` v1,signature1 v2,signature2 ``` You should iterate over the list of signatures and verify each one using the corresponding secret key version. If any of the signatures match, the request is considered valid. For example, you can implement this logic in Python as follows: ```python python.py theme={null} def verify_signature(generated_signature, header_signatures): for signature in header_signatures.split(): version, expected_signature = signature.split(',') if expected_signature == generated_signature: return True return False ``` ```javascript javascript.js theme={null} function verifySignature(generatedSignature, headerSignatures) { const signatures = headerSignatures.split(' '); for (const signature of signatures) { const [version, expectedSignature] = signature.split(','); if (expectedSignature === generatedSignature) { return true; } } return false; } ``` ```php php.php theme={null} function verifySignature($generatedSignature, $headerSignatures) { $signatures = explode(' ', $headerSignatures); foreach ($signatures as $signature) { list($version, $expectedSignature) = explode(',', $signature); if ($expectedSignature === $generatedSignature) { return true; } } return false; } ``` ```java java.java theme={null} public static boolean verifySignature(String generatedSignature, String headerSignatures) { String[] signatures = headerSignatures.split(" "); for (String signature : signatures) { String[] parts = signature.split(","); String version = parts[0]; String expectedSignature = parts[1]; if (expectedSignature.equals(generatedSignature)) { return true; } } return false; } ``` By following these steps, you can ensure the security of your webhook implementation and protect your system from unauthorized access and data tampering.