Create Image-to-Video Task
- Official docs: https://platform.vidu.cn/docs/image-to-video
- Create an image-to-video task; returns task_id or task details
- Control generation behavior with request parameters
Authorizations
Request Body
Text prompt
Text description for video generation.
Note: Must not exceed 2000 characters
Model name
Allowed values: viduq2, viduq1, viduq3-turbo, viduq3-pro
- viduq3-turbo: Faster generation than viduq3-pro
- viduq3-pro: Efficient high-quality audio-video; more vivid, expressive, dimensional output
- viduq2: Latest model
- viduq1: Clear visuals, smooth transitions, stable camera motion
Video duration; default depends on model:
- viduq3-pro, viduq3-turbo: default 5s, range 1–16
- viduq2: default 5s, range 1–10
- viduq1: default 5s, allowed 5 only
Random seed
If omitted or 0, a random value is used
If set manually, that seed is used
Resolution; default depends on model and duration:
- viduq3-pro, viduq3-turbo (1–16s): default 720p, allowed 540p, 720p, 1080p
- viduq2 (1–10s): default 720p, allowed 540p, 720p, 1080p
- viduq1 (5s): default 1080p, allowed 1080p
Motion amplitude
Default auto; allowed: auto, small, medium, large
Note: Not effective when using the q2 model
Passthrough parameter
Not processed; data transfer only
Note: Max 1048576 characters
Off-peak mode; default false; allowed:
- true: generate video in off-peak mode
- false: generate video immediately
Note 1: Off-peak mode uses fewer credits; see product pricing
Note 2: Off-peak tasks are processed within 48 hours; unfinished tasks are auto-cancelled and credits refunded
Note 3: You can manually cancel off-peak tasks
Whether to add a watermark
- true: add watermark
- false: no watermark
Note 1: Watermark content is fixed (AI-generated); off by default
Note 2: Use watermarked_url from the query task API for watermarked video
Watermark position on the image:
1: top-left
2: top-right
3: bottom-right
4: bottom-left
Default: 3
Watermark image URL
If omitted, default watermark text: AI-generated content
Metadata identifier (JSON string), passthrough; custom or example format:
{
"Label": "your_label","ContentProducer": "yourcontentproducer","ContentPropagator": "your_content_propagator","ProduceID": "yourproductid", "PropagateID": "your_propagate_id","ReservedCode1": "yourreservedcode1", "ReservedCode2": "your_reserved_code2"
}
If empty, Vidu-generated metadata is used by default
Callback URL (POST). Set when creating the task; Vidu sends task status updates to this URL. Body matches the query task API response.
Callback status values:
- processing: in progress
- success: completed (retried up to 3 times on delivery failure)
- failed: failed (retried up to 3 times on delivery failure)
Vidu uses callback signature authentication; see callback signature docs
First-frame image
The model uses the image(s) in this parameter as the first frame to generate the video.
Note 1: Supports image Base64 or image URL (must be accessible)
Note 2: Only 1 image allowed
Note 3: Formats: png, jpeg, jpg, webp
Note 4: Aspect ratio must be less than 1:4 or 4:1
Note 5: Max image size 50 MB
Note 6: HTTP POST body must not exceed 20 MB; encoding must include the correct Content-Type, e.g.:
data:image/png;base64,{base64_encode}
Whether to use direct audio-video output; default false; allowed:
- false: silent video (no direct audio-video)
- true: video with dialogue and background audio
Note 1: voice_id applies only when this parameter is true
Note 2: When model is q3-pro or q3-turbo, default is true
Note 3: Off-peak mode is not supported when this parameter is true
Voice ID
Determines the voice timbre in the video; if empty, the system recommends one. See voice list: https://shengshu.feishu.cn/sheets/EgFvs6DShhiEBStmjzccr5gonOg
Whether to use recommended prompts
- true: system recommends prompts (count=1) and uses them to generate the video
- false: generate from the input prompt
Note: Recommended prompts add 10 credits per task
Responses
Success