seedance-2 Video Generation

Generate videos with seedance-2 and seedance-2-fast

POST /v1/video/generations

cURL (Text-to-Video)

curl --request POST \
  --url https://toapis.com/v1/video/generations \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "seedance-2",
    "prompt": "A calm sunrise over the ocean, with gentle waves and cinematic lighting.",
    "duration": 10,
    "ratio": "16:9"
  }'

200

{
  "id": "tsk_vid_xxx",
  "object": "generation.task",
  "model": "seedance-2",
  "status": "in_progress",
  "progress": 50,
  "created_at": 1778569600,
  "metadata": {}
}

• Supports seedance-2 and seedance-2-fast
• Supports text-to-video generation
• Supports first-frame image-to-video through image_urls
• Async task workflow with task ID based status queries

Authorizations

Authorization · string · required

All requests require Bearer Token authentication.

Authorization: Bearer YOUR_API_KEY

Body

model · string · default: "seedance-2" · required

Video generation model name.

Available models:

• seedance-2 - Standard version
• seedance-2-fast - Fast version

prompt · string · required

Video prompt. Describe subject, action, camera movement, and visual style clearly.

duration · integer · default: 5

Video duration in seconds.

Supported range: 4 to 15 seconds.

ratio · string

Video aspect ratio.

Options:

• 16:9
• 9:16
• 1:1
• 4:3

We recommend explicitly setting ratio.

image_urls · string[]

First-frame image URL array for image-to-video.

Only the first image is used.

Image requirements:

• Format: JPEG, JPG, PNG, WEBP
• File size: up to 10MB

Use Uploaded Assets

If you have already uploaded private assets through the avatar asset APIs and obtained an asset_id, you do not need to send the original file URL again in the generation request. Use:

• asset://<ASSET_ID>

This applies to:

• Private virtual avatar assets
• Private real avatar assets
• Image, video, and audio assets that have finished processing and are in active status

Before using an uploaded asset in video generation, make sure you complete this flow:

1. Create an asset group or complete real-person verification first
2. Upload the asset and obtain asset_id
3. Poll the asset status until it becomes active
4. Use asset://<ASSET_ID> in the generation request

You can use these guides to complete asset onboarding:

• Private Avatar Assets
• Real Avatar Assets

How To Reference Assets In Generation Requests

Image assets can be used as:

• first_frame
• last_frame
• reference_image

Video assets can be used as:

• reference_video

Audio assets can be used as:

• reference_audio

Minimal example:

{
  "model": "seedance-2",
  "prompt": "Keep the same person from image 1 and let them slowly turn around in front of a floor-to-ceiling window with a gentle camera push-in.",
  "image_with_roles": [
    {
      "url": "asset://asset_img_01KXXXXXXX",
      "role": "reference_image"
    }
  ]
}

Example with video and audio references:

{
  "model": "seedance-2",
  "prompt": "Keep the character from image 1, follow the camera movement of video 1, and match the rhythm and mood of audio 1 to generate an indoor talking-head video.",
  "image_with_roles": [
    {
      "url": "asset://asset_img_01KXXXXXXX",
      "role": "reference_image"
    }
  ],
  "video_with_roles": [
    {
      "url": "asset://asset_vid_01KXXXXXXX",
      "role": "reference_video"
    }
  ],
  "audio_with_roles": [
    {
      "url": "asset://asset_aud_01KXXXXXXX",
      "role": "reference_audio"
    }
  ]
}

Only assets with active status can be used for video generation.

If you pass asset://<ASSET_ID> while the asset is still processing or has already failed, the generation request may fail or produce poor results.

Response

id · string

Unique task ID for status queries.

object · string

Object type, always generation.task.

model · string

Model name used.

status · string

Task status.

• queued
• in_progress
• completed
• failed

progress · integer

Task progress percentage (0 to 100).

created_at · integer

Task creation timestamp (Unix timestamp).

metadata · object

Task metadata.

cURL (Text-to-Video)

curl --request POST \
  --url https://toapis.com/v1/video/generations \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "seedance-2",
    "prompt": "A calm sunrise over the ocean, with gentle waves and cinematic lighting.",
    "duration": 10,
    "ratio": "16:9"
  }'

200

{
  "id": "tsk_vid_xxx",
  "object": "generation.task",
  "model": "seedance-2",
  "status": "in_progress",
  "progress": 50,
  "created_at": 1778569600,
  "metadata": {}
}

Query Task

The submit endpoint returns a task ID. Query status and result with:

curl --request GET \
  --url https://toapis.com/v1/video/generations/{task_id} \
  --header 'Authorization: Bearer YOUR_API_KEY'