> ## Documentation Index > Fetch the complete documentation index at: https://evolink.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # Seedance 2.5 Complete Parameter Guide > Complete parameter reference for Seedance 2.5. Use `model` to select a model in this folder. Required fields, input requirements, and billing rules are listed for each mode. | Model ID | Mode | Required fields (excluding model) | |---|---|---| | `seedance-2.5-text-to-video` | Text-to-Video | `prompt` | | `seedance-2.5-image-to-video` | Image-to-Video | `prompt`, `image_urls` | | `seedance-2.5-reference-to-video` | Reference-to-Video | `prompt` | | `seedance-2.5-video-edit` | Video Edit | `prompt`, `video_urls` | | `seedance-2.5-video-extend` | Video Extend | `prompt`, `video_urls` | | `seedance-2.5-draft-to-video` | Draft-to-Video | `source_task_id` | ### Text-to-Video - Generate videos from text prompts, supports web search for enhanced timeliness - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Image-to-Video - Input 1 image for first-frame video generation, input 2 images for first-last-frame video generation, the model determines automatically - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Reference-to-Video - Input reference images (0–30) + reference videos (0–10) + reference audio (0–10) + a text prompt to generate video; provide at least one image, video, or audio reference - Best suited for generating new footage: images, videos, and audio act as style, subject, camera-movement, and soundtrack references - **To rewrite or continue an existing video, switch to [Video Edit](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-video-edit) / [Video Extend](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-video-extend)**: once this model's prompt carries an editing or extension intent, the request is validated as an edit / extend task — it is accepted without an error and the task only fails while it runs - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Video Edit - Video editing on an input video — replace subjects, add/remove objects, restyle or repaint regions - Prompt must state the editing intent (e.g. "Edit the video: remove the background music of @video1") - Input the video to edit (required) + optional reference images (0–30) + reference videos (up to 10 in total) + reference audio (0–10) - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Video Extend - Extend an input video forward or backward - Prompt must state the extension intent (e.g. "Extend @video1 backward, the character walks out of the frame") - Input the video to extend (required) + optional reference images (0–30) + reference videos (up to 10 in total) + reference audio (0–10) - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Draft-to-Video - Convert a completed Seedance 2.5 draft (a task created with `draft: true`) into a 1080p final video with the same content - Only four fields are accepted: `model`, `source_task_id`, `output_format`, and `callback_url`. The prompt, assets, duration, aspect ratio, seed, and audio settings are all reused from the draft. Do not send them again; any other field returns a 400 parameter error - Submit the conversion request before the draft's `draft_expires_at`. The default validity period is 24 hours after the draft completes; use the actual `draft_expires_at` returned by the task query as the deadline - Expiration is checked both when the request is accepted and before generation is submitted. If the draft expires while queued, the conversion task fails and its charge is refunded. Allow time for queuing - The same draft can be converted multiple times before it expires; each conversion is billed separately - Asynchronous processing mode: use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail). The query response includes `source_task_id` - Generated video links are valid for 24 hours; save them promptly **Billing rules:** - Billed separately from the draft, always at the `1080p` rate - The output duration is the actual output duration recorded for the source draft. If that value is unavailable, the effective output duration recorded for the source task is used - Billable duration: without video input, the output duration; with video input (including Reference-to-Video, Video Edit, and Video Extend), `output duration + max(input video duration, output duration)`. Both are rounded up to the next 0.1 second - If content filtering was disabled for the draft (`content_filter: false`), the final video is also billed at +10% (`1.1x`) - If conversion fails, the conversion charge is refunded; the draft charge is not refunded ## OpenAPI ````yaml en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-overview.json POST /v1/videos/generations openapi: 3.1.0 info: title: Seedance 2.5 Complete Parameter Guide description: >- Unified parameter reference for the 6 models in this Seedance 2.5 folder. Select a mode with model; required fields and limits for each mode are documented below. license: name: MIT version: 1.0.0 servers: - url: https://api.evolink.ai description: Production security: - bearerAuth: [] tags: - name: Video Generation description: AI video generation endpoints paths: /v1/videos/generations: post: tags: - Video Generation summary: Seedance 2.5 Complete Parameter Guide description: >+ Complete parameter reference for Seedance 2.5. Use `model` to select a model in this folder. Required fields, input requirements, and billing rules are listed for each mode. | Model ID | Mode | Required fields (excluding model) | |---|---|---| | `seedance-2.5-text-to-video` | Text-to-Video | `prompt` | | `seedance-2.5-image-to-video` | Image-to-Video | `prompt`, `image_urls` | | `seedance-2.5-reference-to-video` | Reference-to-Video | `prompt` | | `seedance-2.5-video-edit` | Video Edit | `prompt`, `video_urls` | | `seedance-2.5-video-extend` | Video Extend | `prompt`, `video_urls` | | `seedance-2.5-draft-to-video` | Draft-to-Video | `source_task_id` | ### Text-to-Video - Generate videos from text prompts, supports web search for enhanced timeliness - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Image-to-Video - Input 1 image for first-frame video generation, input 2 images for first-last-frame video generation, the model determines automatically - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Reference-to-Video - Input reference images (0–30) + reference videos (0–10) + reference audio (0–10) + a text prompt to generate video; provide at least one image, video, or audio reference - Best suited for generating new footage: images, videos, and audio act as style, subject, camera-movement, and soundtrack references - **To rewrite or continue an existing video, switch to [Video Edit](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-video-edit) / [Video Extend](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-video-extend)**: once this model's prompt carries an editing or extension intent, the request is validated as an edit / extend task — it is accepted without an error and the task only fails while it runs - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Video Edit - Video editing on an input video — replace subjects, add/remove objects, restyle or repaint regions - Prompt must state the editing intent (e.g. "Edit the video: remove the background music of @video1") - Input the video to edit (required) + optional reference images (0–30) + reference videos (up to 10 in total) + reference audio (0–10) - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Video Extend - Extend an input video forward or backward - Prompt must state the extension intent (e.g. "Extend @video1 backward, the character walks out of the frame") - Input the video to extend (required) + optional reference images (0–30) + reference videos (up to 10 in total) + reference audio (0–10) - **Now supports AIGC realistic human materials** - Supports **draft mode**: set `draft: true` to generate a 480p draft first. If you are satisfied, use [Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to convert it into a 1080p final video. The two requests are billed separately - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ### Draft-to-Video - Convert a completed Seedance 2.5 draft (a task created with `draft: true`) into a 1080p final video with the same content - Only four fields are accepted: `model`, `source_task_id`, `output_format`, and `callback_url`. The prompt, assets, duration, aspect ratio, seed, and audio settings are all reused from the draft. Do not send them again; any other field returns a 400 parameter error - Submit the conversion request before the draft's `draft_expires_at`. The default validity period is 24 hours after the draft completes; use the actual `draft_expires_at` returned by the task query as the deadline - Expiration is checked both when the request is accepted and before generation is submitted. If the draft expires while queued, the conversion task fails and its charge is refunded. Allow time for queuing - The same draft can be converted multiple times before it expires; each conversion is billed separately - Asynchronous processing mode: use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail). The query response includes `source_task_id` - Generated video links are valid for 24 hours; save them promptly **Billing rules:** - Billed separately from the draft, always at the `1080p` rate - The output duration is the actual output duration recorded for the source draft. If that value is unavailable, the effective output duration recorded for the source task is used - Billable duration: without video input, the output duration; with video input (including Reference-to-Video, Video Edit, and Video Extend), `output duration + max(input video duration, output duration)`. Both are rounded up to the next 0.1 second - If content filtering was disabled for the draft (`content_filter: false`), the final video is also billed at +10% (`1.1x`) - If conversion fails, the conversion charge is refunded; the draft charge is not refunded operationId: createSeedance25VideoGeneration requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/VideoGenerationRequest' examples: text_to_video_basic: summary: Basic text-to-video value: model: seedance-2.5-text-to-video prompt: >- A macro lens focuses on a green glass frog on a leaf. The focus gradually shifts from its smooth skin to its completely transparent abdomen, where a bright red heart is beating powerfully and rhythmically. duration: 8 quality: 720p aspect_ratio: '16:9' generate_audio: true content_filter: true text_to_video_web_search: summary: Text-to-video with web search value: model: seedance-2.5-text-to-video prompt: >- Today's New York weather forecast, with city skyline animation and temperature overlay display duration: 8 quality: 720p aspect_ratio: '16:9' model_params: web_search: true text_to_video_draft: summary: Draft mode (480p) value: model: seedance-2.5-text-to-video prompt: >- A macro lens focuses on a green glass frog on a leaf. The focus gradually shifts from its smooth skin to its completely transparent abdomen, where a bright red heart is beating powerfully and rhythmically. duration: 8 quality: 480p aspect_ratio: '16:9' generate_audio: true content_filter: true draft: true image_to_video_first_frame: summary: First-frame image-to-video value: model: seedance-2.5-image-to-video prompt: >- The camera slowly zooms in, the scene gradually comes to life image_urls: - https://example.com/first-frame.jpg duration: 5 quality: 720p aspect_ratio: adaptive image_to_video_first_last_frame: summary: First-last-frame image-to-video value: model: seedance-2.5-image-to-video prompt: A smooth transition between two scenes image_urls: - https://example.com/first.jpg - https://example.com/last.jpg duration: 6 quality: 720p aspect_ratio: adaptive image_to_video_draft: summary: Draft mode (480p) value: model: seedance-2.5-image-to-video prompt: >- The camera slowly zooms in, the scene gradually comes to life image_urls: - https://example.com/first-frame.jpg duration: 5 quality: 480p aspect_ratio: adaptive draft: true reference_to_video_multimodal: summary: Multimodal reference (image + video + audio) value: model: seedance-2.5-reference-to-video prompt: >- Use the first-person perspective framing of @video1 throughout, and use @audio1 as background music throughout. First-person perspective fruit tea promotional video... image_urls: - https://example.com/ref1.jpg - https://example.com/ref2.jpg video_urls: - https://example.com/reference.mp4 audio_urls: - https://example.com/bgm.mp3 duration: 10 quality: 720p aspect_ratio: '16:9' generate_audio: true content_filter: true reference_to_video_subject_reference: summary: Image + video reference (subject + camera movement) value: model: seedance-2.5-reference-to-video prompt: >- Make the face cream in @image1 the hero product, and follow the camera rhythm and lighting mood of @video1 to generate a clean product commercial image_urls: - https://example.com/cream.jpg video_urls: - https://example.com/reference.mp4 duration: 5 quality: 720p aspect_ratio: '16:9' reference_to_video_multi_video: summary: Multi-video reference (multi-shot narrative) value: model: seedance-2.5-reference-to-video prompt: >- The arched window in @video1 opens into the art gallery interior, continuing with @video2; then the camera enters the painting, continuing with @video3 video_urls: - https://example.com/part1.mp4 - https://example.com/part2.mp4 - https://example.com/part3.mp4 duration: 8 quality: 720p aspect_ratio: '16:9' generate_audio: true content_filter: true reference_to_video_draft: summary: Draft mode (480p) value: model: seedance-2.5-reference-to-video prompt: >- Use the first-person perspective framing of @video1 throughout, and use @audio1 as background music throughout. First-person perspective fruit tea promotional video... image_urls: - https://example.com/ref1.jpg - https://example.com/ref2.jpg video_urls: - https://example.com/reference.mp4 audio_urls: - https://example.com/bgm.mp3 duration: 10 quality: 480p aspect_ratio: '16:9' generate_audio: true content_filter: true draft: true video_edit_remove_objects: summary: Edit video (remove objects) value: model: seedance-2.5-video-edit prompt: >- Edit the video: remove all passers-by in @video1, keep only the main character video_urls: - https://example.com/original.mp4 quality: 720p generate_audio: true content_filter: true video_edit_replace_subject: summary: Edit video (replace element with image reference) value: model: seedance-2.5-video-edit prompt: >- Edit the video: replace the perfume in the gift box in @video1 with the face cream in @image1, keep the camera movement unchanged image_urls: - https://example.com/cream.jpg video_urls: - https://example.com/original.mp4 duration: -1 quality: 720p video_edit_draft: summary: Draft mode (480p) value: model: seedance-2.5-video-edit prompt: >- Edit the video: remove all passers-by in @video1, keep only the main character video_urls: - https://example.com/original.mp4 quality: 480p generate_audio: true content_filter: true draft: true video_extend_extend_backward: summary: Extend video backward (custom duration) value: model: seedance-2.5-video-extend prompt: >- Extend @video1 backward: the character walks out of the frame, the camera slowly pulls back to reveal the empty street video_urls: - https://example.com/original.mp4 duration: 12 quality: 720p generate_audio: true content_filter: true video_extend_extend_forward: summary: Extend video forward (automatic duration) value: model: seedance-2.5-video-extend prompt: >- Extend @video1 forward: the character keeps walking into the sunset as the light gradually fades video_urls: - https://example.com/original.mp4 duration: -1 quality: 720p video_extend_draft: summary: Draft mode (480p) value: model: seedance-2.5-video-extend prompt: >- Extend @video1 backward: the character walks out of the frame, the camera slowly pulls back to reveal the empty street video_urls: - https://example.com/original.mp4 duration: 12 quality: 480p generate_audio: true content_filter: true draft: true draft_to_video_basic: summary: Convert to a 1080p final video (MP4) value: model: seedance-2.5-draft-to-video source_task_id: task-unified-1774857405-abc123 draft_to_video_mov: summary: Convert to a 1080p final video (MOV) value: model: seedance-2.5-draft-to-video source_task_id: task-unified-1774857405-abc123 output_format: mov responses: '200': description: Video generation task created successfully content: application/json: schema: $ref: '#/components/schemas/VideoGenerationResponse' '400': description: Invalid request parameters content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: invalid_parameter message: Invalid request parameters type: invalid_request_error '401': description: Unauthenticated, token invalid or expired content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: unauthorized message: Invalid or expired token type: authentication_error '402': description: Insufficient quota, top-up required content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: insufficient_quota message: Insufficient quota. Please top up your account. type: insufficient_quota '403': description: Access denied content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: model_access_denied message: >- Token does not have access to model: seedance-2.5-text-to-video type: invalid_request_error '429': description: Rate limit exceeded content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: rate_limit_exceeded message: Too many requests, please try again later type: rate_limit_error '500': description: Internal server error content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: internal_error message: Internal server error type: api_error '503': description: >- Service temporarily unavailable; draft_unavailable may be returned when draft mode is unavailable content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' examples: draft_unavailable: summary: Draft mode temporarily unavailable value: error: code: draft_unavailable message: >- Draft mode is not available on the selected service right now. Retry later or send the request without draft. type: invalid_request_error components: schemas: VideoGenerationRequest: type: object description: >- This page combines the parameters of all modes. Required fields, accepted values, and defaults depend on the selected model; differences are listed by mode below. required: - model properties: model: type: string description: >- Video generation model name | Model ID | Mode | Required fields (excluding model) | |---|---|---| | `seedance-2.5-text-to-video` | Text-to-Video | `prompt` | | `seedance-2.5-image-to-video` | Image-to-Video | `prompt`, `image_urls` | | `seedance-2.5-reference-to-video` | Reference-to-Video | `prompt` | | `seedance-2.5-video-edit` | Video Edit | `prompt`, `video_urls` | | `seedance-2.5-video-extend` | Video Extend | `prompt`, `video_urls` | | `seedance-2.5-draft-to-video` | Draft-to-Video | `source_task_id` | enum: - seedance-2.5-text-to-video - seedance-2.5-image-to-video - seedance-2.5-reference-to-video - seedance-2.5-video-edit - seedance-2.5-video-extend - seedance-2.5-draft-to-video example: seedance-2.5-text-to-video prompt: type: string description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** **Text-to-Video** Text prompt describing the desired video. Supports both Chinese and English, recommended no more than 500 characters for Chinese or 1000 words for English. Maximum prompt length: 10000 tokens **Note:** - This model is text-to-video only and does not support `image_urls`, `video_urls`, or `audio_urls` input **Image-to-Video** Text prompt describing the desired video. Supports both Chinese and English, recommended no more than 500 characters for Chinese or 1000 words for English. Maximum prompt length: 10000 tokens **Note:** - This model does not support `video_urls` or `audio_urls` input **Reference-to-Video** Text prompt describing the desired video. Supports both Chinese and English, recommended no more than 500 characters for Chinese. Maximum prompt length: 10000 tokens **Details:** - When referring to input materials in the prompt, prefer explicit tags such as `@image1`, `@video1`, and `@audio1` instead of relying only on natural-language references such as "image 1", "video 1", or "audio 1" - Tag numbers follow the order of the corresponding URL array and start at 1: the first `image_urls` item is `@image1`, the first `video_urls` item is `@video1`, and the first `audio_urls` item is `@audio1` - Natural-language numbering may still be understood, but explicit tags provide a clearer mapping between each material and its intended use **Avoid being classified as an edit / extend task:** - The task type is inferred from the prompt intent: wording such as "edit the video", "add", "remove / delete", "change / replace" is treated as video editing; wording such as "extend forward / backward", "continue" is treated as video extension - Once classified as either task, `aspect_ratio` must be `adaptive`, and for video editing `duration` must be `-1`; otherwise the request is accepted without an error and the task later turns `failed` with error code `InvalidParameter.TaskTypeConstraint`. The pre-authorized quota is refunded, but the waiting time is wasted - If you really need editing / extension, use `seedance-2.5-video-edit` or `seedance-2.5-video-extend` instead: parameters are validated at submission, so invalid values return an immediate synchronous error - For pure reference generation, rephrase words like "replace" or "remove" descriptively, e.g. "the shot features ..." or "rendered in the style of @image1" **Video Edit** Text prompt describing the desired edit. Supports both Chinese and English, recommended no more than 500 characters for Chinese. Maximum prompt length: 10000 tokens **Details:** - The prompt must state the editing intent, e.g., "Edit the video: remove the background music of @video1" - You can use natural language to specify the purpose of each material, e.g., "replace the character in @video1 with the person in @image1" - The model will automatically understand the correspondence between material numbers and their intended uses **Video Extend** Text prompt describing the desired extension. Supports both Chinese and English, recommended no more than 500 characters for Chinese. Maximum prompt length: 10000 tokens **Details:** - The prompt must state the extension intent, e.g., "Extend @video1 backward, the character walks out of the frame" - You can use natural language to specify the purpose of each material, e.g., "extend @video1 forward following the style of @image1" - The model will automatically understand the correspondence between material numbers and their intended uses example: >- A macro lens focuses on a green glass frog on a leaf. The focus gradually shifts from its smooth skin to its completely transparent abdomen, where a bright red heart is beating powerfully and rhythmically. source_task_id: type: string description: >- Applies only to: **Draft-to-Video** Task ID of a completed Seedance 2.5 draft **Where to find it:** The `id` field returned when creating a draft (`draft: true`) **Requirements:** 1. The task must belong to the current account 2. The task must use one of the five Seedance 2.5 generation models 3. The task must have been created with `draft: true` 4. The task status must be `completed` 5. The current time must be earlier than `draft_expires_at` in the task query response example: task-unified-1774857405-abc123 image_urls: type: array description: >- Applies only to: **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** **Image-to-Video** Image URL array, **1–2 images** **Image count and behavior:** | Image Count | Behavior | Role | |:--------:|------|------| | 1 image | First-frame image-to-video | Automatically set as `first_frame` | | 2 images | First-last-frame image-to-video | 1st image → `first_frame`, 2nd image → `last_frame` | **Image requirements:** - Supported formats: `.jpeg`, `.png`, `.webp` - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Max size per image: `30MB` - Total request body size must not exceed `64MB` - When providing first and last frames, both images can be identical. If aspect ratios differ, the first frame takes priority and the last frame will be automatically cropped to fit - Image URLs must be directly accessible by the server **Reference-to-Video** Reference image URL array, **optional; 1–30 images when provided** **Requirement:** At least one of `image_urls`, `video_urls`, or `audio_urls` must be provided. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Image | `reference_image` | Style reference, product image, first/last frame (specified via prompt) | **Image requirements:** - Supported formats: `.jpeg`, `.png`, `.webp` - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Max size per image: `30MB` - Total request body size must not exceed `64MB`, do not use Base64 encoding - Image URLs must be directly accessible by the server **Video Edit** Reference image URL array, **optional; 1–30 images when provided** **Requirement:** `video_urls` is required for this model; images are optional supplementary references. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Image | `reference_image` | Style reference, product image, replacement subject (specified via prompt) | **Image requirements:** - Supported formats: `.jpeg`, `.png`, `.webp` - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Max size per image: `30MB` - Total request body size must not exceed `64MB`, do not use Base64 encoding - Image URLs must be directly accessible by the server **Video Extend** Reference image URL array, **optional; 1–30 images when provided** **Requirement:** `video_urls` is required for this model; images are optional supplementary references. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Image | `reference_image` | Style reference, product image, scene reference (specified via prompt) | **Image requirements:** - Supported formats: `.jpeg`, `.png`, `.webp` - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Max size per image: `30MB` - Total request body size must not exceed `64MB`, do not use Base64 encoding - Image URLs must be directly accessible by the server items: type: string format: uri minItems: 1 maxItems: 30 example: - https://example.com/first-frame.jpg video_urls: type: array description: >- Applies only to: **Reference-to-Video**, **Video Edit**, **Video Extend** **Reference-to-Video** Reference video URL array, **optional; 1–10 videos when provided** **Requirement:** At least one of `image_urls`, `video_urls`, or `audio_urls` must be provided. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Video | `reference_video` | Camera movement reference, motion reference, original video for editing/extension | **Video requirements:** - Supported formats: `.mp4`, `.mov` - Resolution: 480p, 720p, 1080p, 4K - Duration per video: `2` ~ `30` seconds, max 10 videos, total duration of all videos ≤ `30` seconds - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Frame pixels (width × height): `409,600` ~ `8,295,044` (e.g., 640×640 ~ 3326×2494) - Max size per video: `200MB` - Frame rate: `24` ~ `60` FPS - Total request body size must not exceed `64MB`, do not use Base64 encoding - Specific codec, resolution, and other media validity checks are performed by the model - Using video references will increase costs (input video duration is counted in billing) - Video URLs must be directly accessible by the server **Video Edit** Input video URL array, **required; 1–10 videos** **Requirement:** At least 1 video required — the video to edit. The first video is the video being edited; additional videos serve as references. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Video | `reference_video` | The video to edit (first video), camera movement or motion reference (additional videos) | **Video requirements:** - Supported formats: `.mp4`, `.mov` - Resolution: 480p, 720p, 1080p, 4K - Duration per video: `4` ~ `30` seconds (the first video is the edit target), max 10 videos, total duration of all videos ≤ `30` seconds - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Frame pixels (width × height): `409,600` ~ `8,295,044` (e.g., 640×640 ~ 3326×2494) - Max size per video: `200MB` - Frame rate: `24` ~ `60` FPS - Total request body size must not exceed `64MB`, do not use Base64 encoding - Specific codec, resolution, and other media validity checks are performed by the model - Using video references will increase costs (input video duration is counted in billing) - Video URLs must be directly accessible by the server **Video Extend** Input video URL array, **required; 1–10 videos** **Requirement:** At least 1 video required — the video to extend. The first video is the video being extended; additional videos serve as references. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Video | `reference_video` | The video to extend (first video), camera movement or motion reference (additional videos) | **Video requirements:** - Supported formats: `.mp4`, `.mov` - Resolution: 480p, 720p, 1080p, 4K - Duration per video: `2` ~ `30` seconds, max 10 videos, total duration of all videos ≤ `30` seconds - Aspect ratio (width/height): `0.4` ~ `2.5` - Width/height pixels: `300` ~ `6000` px - Frame pixels (width × height): `409,600` ~ `8,295,044` (e.g., 640×640 ~ 3326×2494) - Max size per video: `200MB` - Frame rate: `24` ~ `60` FPS - Total request body size must not exceed `64MB`, do not use Base64 encoding - Specific codec, resolution, and other media validity checks are performed by the model - Using video references will increase costs (input video duration is counted in billing) - Video URLs must be directly accessible by the server items: type: string format: uri minItems: 1 maxItems: 10 example: - https://example.com/reference.mp4 audio_urls: type: array description: >- Applies only to: **Reference-to-Video**, **Video Edit**, **Video Extend** **Reference-to-Video** Reference audio URL array, **optional; 1–10 clips when provided** **Requirement:** At least one of `image_urls`, `video_urls`, or `audio_urls` must be provided. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Audio | `reference_audio` | Background music, sound effects, voice/dialogue reference | **Audio requirements:** - Supported formats: `.wav`, `.mp3` - Duration per clip: `2` ~ `30` seconds, max 10 clips, total duration of all audio ≤ `30` seconds - Max size per clip: `15MB` - Total request body size must not exceed `64MB`, do not use Base64 encoding - Audio URLs must be directly accessible by the server **Video Edit / Video Extend** Reference audio URL array, **optional; 1–10 clips when provided** **Requirement:** `video_urls` is required for this model; audio clips are optional supplementary references. **Role description:** | Media Type | Role | Typical Usage | |---------|------|----------| | Audio | `reference_audio` | Background music, sound effects, voice/dialogue reference | **Audio requirements:** - Supported formats: `.wav`, `.mp3` - Duration per clip: `2` ~ `30` seconds, max 10 clips, total duration of all audio ≤ `30` seconds - Max size per clip: `15MB` - Total request body size must not exceed `64MB`, do not use Base64 encoding - Audio URLs must be directly accessible by the server items: type: string format: uri minItems: 1 maxItems: 10 example: - https://example.com/bgm.mp3 duration: type: integer description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** **Text-to-Video / Image-to-Video** Video duration (seconds), defaults to `5` seconds **Details:** - Supports any integer value between `4`–`30` seconds - `-1`: automatic duration, the model picks a length within `4`–`30` seconds and billing follows the actual output length - Duration directly affects billing **Reference-to-Video** Output video duration (seconds), defaults to `5` seconds **Details:** - Supports any integer value between `4`–`30` seconds - `-1`: automatic duration, the model picks a length within `4`–`30` seconds and billing follows the actual output length - Duration directly affects billing If the prompt is classified as a video edit task, only `-1` is accepted; an explicit number of seconds makes the task fail later, with no error at submission **Video Edit** Output video duration (seconds), defaults to `-1` **Details:** - Only `-1` is supported: the output length follows the input video (may be up to 0.4s shorter) - Custom durations are rejected - Duration directly affects billing **Video Extend** Output video duration (seconds), defaults to `5` seconds **Details:** - Supports any integer value between `4`–`30` seconds - `-1`: automatic duration, settled by actual output length - Duration directly affects billing example: 8 quality: type: string description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** Video resolution, defaults to `720p` **Options:** - `480p`: Lower clarity, lower cost - `720p`: Standard clarity, this is the default - `1080p`: Ultra HD clarity **Draft mode:** When `draft` is `true`, only `480p` is supported. If omitted, `quality` automatically defaults to `480p` enum: - 480p - 720p - 1080p default: 720p example: 720p draft: type: boolean description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** Draft mode switch, defaults to `false` **Details:** - `true`: Generate a 480p draft first, with faster turnaround and standard 480p billing - When enabled, `quality` must be `480p` or omitted. Any other resolution returns a 400 parameter error - Must be a top-level request body field. Placing it inside `model_params` returns a 400 parameter error **Convert to a final video:** - Once the draft is complete, [querying the task](/en/api-manual/task-management/get-task-detail) returns `draft_expires_at` - Before that time, pass the draft task ID to [Seedance 2.5 Draft-to-Video](/en/api-manual/video-series/seedance/seedance-2.5/seedance-2.5-draft-to-video) to generate a 1080p final video with the same content - The final video and draft are billed separately. If you are not satisfied with the draft, skip conversion and pay only for the draft - The default validity period is 24 hours after the draft completes; the actual deadline is `draft_expires_at`. Submit the conversion request early: expiration while queued also causes conversion to fail **Temporary unavailability:** A 503 error may be returned when the service is temporarily unavailable. If the error code is `draft_unavailable`, retry later or remove `draft` to use normal generation mode default: false example: false aspect_ratio: type: string description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** **Text-to-Video** Video aspect ratio, defaults to `adaptive` **Options:** - `16:9` (landscape), `9:16` (portrait), `1:1` (square), `4:3`, `3:4`, `21:9` (ultrawide) - `adaptive`: The model intelligently selects the best aspect ratio based on the prompt **Pixel values per resolution:** | Aspect Ratio | 480p | 720p | 1080p | |:------:|:----:|:----:|:----:| | 16:9 | 854×480 | 1280×720 | 1920×1080 | | 4:3 | 752×560 | 1112×834 | 1664×1248 | | 1:1 | 640×640 | 960×960 | 1440×1440 | | 3:4 | 560×752 | 834×1112 | 1248×1664 | | 9:16 | 480×854 | 720×1280 | 1080×1920 | | 21:9 | 992×432 | 1470×630 | 2206×946 | **Image-to-Video** Video aspect ratio, image-to-video only supports `adaptive` **Options:** - `adaptive`: Automatically selects the closest aspect ratio based on the first-frame image proportions, the only value this model accepts - Fixed ratios such as `16:9` or `9:16` cannot be specified for this model **Reference-to-Video** Video aspect ratio, defaults to `adaptive` **Options:** - `16:9` (landscape), `9:16` (portrait), `1:1` (square), `4:3`, `3:4`, `21:9` (ultrawide) - `adaptive`: Determined based on prompt intent, priority: video > image > prompt **Pixel values per resolution:** | Aspect Ratio | 480p | 720p | 1080p | |:------:|:----:|:----:|:----:| | 16:9 | 854×480 | 1280×720 | 1920×1080 | | 4:3 | 752×560 | 1112×834 | 1664×1248 | | 1:1 | 640×640 | 960×960 | 1440×1440 | | 3:4 | 560×752 | 834×1112 | 1248×1664 | | 9:16 | 480×854 | 720×1280 | 1080×1920 | | 21:9 | 992×432 | 1470×630 | 2206×946 | If the prompt is classified as a video edit / extend task, only `adaptive` is accepted; a fixed ratio makes the task fail later, with no error at submission **Video Edit / Video Extend** Video aspect ratio, defaults to `adaptive` **Options:** - `adaptive`: the output ratio follows the first input video, the only value this model accepts - Fixed ratios such as `16:9` or `9:16` are rejected enum: - '16:9' - '9:16' - '1:1' - '4:3' - '3:4' - '21:9' - adaptive default: adaptive example: '16:9' generate_audio: type: boolean description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** **Text-to-Video** Whether to generate synchronized audio, defaults to `true` **Options:** - `true`: Video includes synchronized audio (voice, sound effects, background music) at no additional charge. It is recommended to place dialogue within double quotes to optimize audio generation - `false`: Output silent video **Image-to-Video / Reference-to-Video / Video Edit / Video Extend** Whether to generate synchronized audio, defaults to `true` **Options:** - `true`: Video includes synchronized audio at no additional charge - `false`: Output silent video default: true example: true content_filter: type: boolean description: >- Applies only to: **Text-to-Video**, **Image-to-Video**, **Reference-to-Video**, **Video Edit**, **Video Extend** Content filter, enabled by default `true` **Options:** - `true`: Standard content safety check, this is the default - `false`: Relaxes content restrictions, billed at +10% (`1.1x`). Illegal and prohibited content is always enforced regardless of this setting **Mutually exclusive with web search:** - `content_filter: false` and `model_params.web_search: true` cannot be used together; sending both returns a 400 parameter error - Keep `content_filter: true` (the default) when web search is needed default: true example: true output_format: type: string description: >- **Text-to-Video / Image-to-Video / Reference-to-Video / Video Edit / Video Extend** Output container format, defaults to `mp4` **Options:** - `mp4`: H.264 encoding with the best compatibility and standard color precision, this is the default - `mov`: H.264 + yuv444p chroma sampling + PCM audio for higher color fidelity, recommended for color grading, keying and compositing workflows. Browser inline playback may not support it; download and play with VLC / mpv / ffplay. No extra charge **Draft-to-Video** Output container format, defaults to `mp4` **Options:** - `mp4`: Default container format. This endpoint outputs 1080p video with 10-bit color depth and H.265/HEVC encoding. Playback requires support for 10-bit HEVC; H.264-only players may not play it - `mov`: H.264 + yuv444p chroma sampling + PCM audio for higher color fidelity, recommended for color grading, keying and compositing workflows. Browser inline playback may not support it; download and play with VLC / mpv / ffplay. No extra charge The draft's output format is not inherited. Specify `mov` here if required enum: - mp4 - mov default: mp4 example: mp4 model_params: type: object description: |- Applies only to: **Text-to-Video** Model extension parameters properties: web_search: type: boolean description: >- Web search, defaults to `false` **Details:** - When enabled, the model autonomously decides whether to search internet content (e.g., products, weather) based on the prompt, improving timeliness - May increase latency - Fees are only charged when searches are actually triggered; multiple searches may occur once enabled **Mutually exclusive with the content filter:** - Keep `content_filter: true` when web search is enabled; using it together with `content_filter: false` returns a 400 parameter error default: false example: false callback_url: type: string description: >- HTTPS callback URL for task completion **Callback timing:** - Triggered when the task is completed or failed - Sent after billing confirmation is complete **Security restrictions:** - Only HTTPS protocol is supported - Callbacks to private IP addresses are prohibited (127.0.0.1, 10.x.x.x, 172.16-31.x.x, 192.168.x.x, etc.) - URL length must not exceed `2048` characters **Callback mechanism:** - Timeout: `10` seconds - Up to `3` retries after failure (at `1`/`2`/`4` seconds after failure respectively) - Callback response body format is consistent with the task query endpoint response format - A 2xx status code is considered successful; other status codes trigger retries format: uri example: https://your-domain.com/webhooks/video-task-completed VideoGenerationResponse: type: object properties: created: type: integer description: Task creation timestamp example: 1761313744 id: type: string description: Task ID example: task-unified-1774857405-abc123 model: type: string description: Actual model name used example: seedance-2.5-text-to-video object: type: string enum: - video.generation.task description: Specific type of the task progress: type: integer description: Task progress percentage (0-100) minimum: 0 maximum: 100 example: 0 status: type: string description: Task status enum: - pending - processing - completed - failed example: pending task_info: $ref: '#/components/schemas/VideoTaskInfo' description: Video task details type: type: string enum: - text - image - audio - video description: Output type of the task example: video usage: $ref: '#/components/schemas/VideoUsage' description: Usage and billing information ErrorResponse: type: object properties: error: type: object properties: code: type: string description: Error code identifier message: type: string description: Error description message type: type: string description: Error type VideoTaskInfo: type: object properties: can_cancel: type: boolean description: >- Whether the task can be cancelled. Seedance 2.5 does not support user cancellation; this field always returns `false`. example: false estimated_time: type: integer description: Estimated completion time (seconds) minimum: 0 example: 165 video_duration: type: integer description: Video duration (seconds) example: 8 VideoUsage: type: object description: Usage and billing information properties: billing_rule: type: string description: Billing rule enum: - per_call - per_token - per_second example: per_second credits_reserved: type: number description: Estimated credits consumed minimum: 0 example: 50 user_group: type: string description: User group category example: default securitySchemes: bearerAuth: type: http scheme: bearer description: >- ##All endpoints require Bearer Token authentication## **Get API Key:** Visit the [API Key Management Page](https://evolink.ai/dashboard/keys) to obtain your API Key **Add to request header:** ``` Authorization: Bearer YOUR_API_KEY ``` ```` This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.