Create Video

Use the unified video endpoint to submit asynchronous video generation tasks. New integrations should use the official-style top-level content array, with text, images, videos, and audio in the same array.

Endpoint

POST /v1/video/generations

Recommended Format: content

content is a multimodal content array. Text, images, videos, and audio inputs are all placed in the same array. Use type for the media type and role for the media purpose.

Text to Video

{
  "model": "doubao-seedance-2.0",
  "content": [
    {
      "type": "text",
      "text": "A cinematic aerial shot of a futuristic cubic city at sunrise"
    }
  ],
  "duration": 5,
  "ratio": "16:9",
  "generate_audio": false,
  "watermark": false
}

Image to Video

{
  "model": "doubao-seedance-2.0",
  "content": [
    {
      "type": "text",
      "text": "Animate image 1 naturally with a cinematic camera move"
    },
    {
      "type": "image_url",
      "role": "first_frame",
      "image_url": {
        "url": "https://example.com/input.jpg"
      }
    }
  ],
  "duration": 5,
  "ratio": "16:9"
}

First and Last Frame

{
  "model": "doubao-seedance-2.0",
  "content": [
    {
      "type": "text",
      "text": "Create a smooth transition from image 1 to image 2"
    },
    {
      "type": "image_url",
      "role": "first_frame",
      "image_url": {
        "url": "https://example.com/start.jpg"
      }
    },
    {
      "type": "image_url",
      "role": "last_frame",
      "image_url": {
        "url": "https://example.com/end.jpg"
      }
    }
  ],
  "duration": 5,
  "ratio": "16:9"
}

Multimodal Reference

{
  "model": "doubao-seedance-2.0",
  "content": [
    {
      "type": "text",
      "text": "Use image 1, video 1, and audio 1 to create a product showcase video"
    },
    {
      "type": "image_url",
      "role": "reference_image",
      "image_url": {
        "url": "asset://reviewed-image-asset-id"
      }
    },
    {
      "type": "video_url",
      "role": "reference_video",
      "video_url": {
        "url": "https://example.com/reference.mp4"
      }
    },
    {
      "type": "audio_url",
      "role": "reference_audio",
      "audio_url": {
        "url": "https://example.com/reference.wav"
      }
    }
  ],
  "duration": 5,
  "ratio": "1:1",
  "generate_audio": true
}

Top-level Fields

Field	Required	Description
model	Yes	Video model name, for example doubao-seedance-2.0 or doubao-seedance-2.0-fast
content	Recommended	Official-style multimodal content array. Recommended for new integrations
duration	No	Duration in seconds. Seedance 2.0 commonly supports 4 to 15; default is usually 5
ratio	No	Aspect ratio, for example adaptive, 16:9, 9:16, 1:1, 4:3, 3:4, 21:9
resolution	No	Output resolution, for example 480p, 720p, 1080p. Availability depends on the model and account
generate_audio	No	Whether to generate synchronized audio
watermark	No	Whether to add a watermark

content

Field	Required	Description
content[].type	Yes	text, image_url, video_url, or audio_url
content[].text	Conditional	Required when type is text
content[].image_url.url	Conditional	Public image URL, or an approved asset://<asset ID>
content[].video_url.url	Conditional	Public video URL
content[].audio_url.url	Conditional	Public audio URL
content[].role	Conditional	Asset role

Supported roles:

Asset	role
Image	first_frame, last_frame, reference_image
Video	reference_video
Audio	reference_audio

Asset Limits

• Base64 and inline binary media are not supported in the public request body.
• Image, video, and audio URLs must be directly reachable by the platform server.
• Up to 9 reference images.
• Up to 3 reference videos. Each video should be at most 15 seconds, and the total reference video duration should not exceed 15 seconds.
• Up to 3 reference audios. Each audio should be at most 15 seconds, and the total reference audio duration should not exceed 15 seconds.
• Audio cannot be the only asset. Provide at least one image or video together with audio.
• Do not mix first-frame/last-frame scenarios with reference_image, reference_video, or reference_audio.

Real-person Images and Moderation

If an input image contains a real person or private information, the upstream may return a PrivacyInformation safety error. For stable use of these assets, submit the image to the moderation endpoint first:

POST /v1/images/moderations

After approval, use the returned items[].asset_url in content[].image_url.url.

Response

Task submission returns a task ID:

{
  "id": "video_xxx",
  "task_id": "video_xxx",
  "object": "video",
  "model": "doubao-seedance-2.0",
  "status": "queued",
  "created_at": 1770000000
}

Query the task:

GET /v1/video/generations/{task_id}

Read the video URL from the task query response after the task succeeds. See the model-specific page for the exact response fields.