Avatar (OmniHuman 2.0) avatar
video model · provider: bytedance · typical generation time ~300s
Capabilities
| Capability | Value |
|---|---|
| Aspect ratios | — |
| Durations | the audio's length |
| Resolutions | — |
| Modes | — |
Image-to-video (image_url) | ✓ |
First frame (first_frame_url) | — |
Last frame (last_frame_url) | — |
Audio generation (generate_audio) | — |
| Reference images | — |
| Reference videos | — |
| Reference audio | — |
| Extendable | — |
Credits
28 credits per started second of the audio.
Example
bash
curl -X POST "https://api.aiveed.io/v1/generate/video" \
-H "X-API-Key: $AIVEED_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "avatar",
"image_url": "https://YOUR-UPLOADED-PORTRAIT-URL",
"audio_url": "https://YOUR-UPLOADED-AUDIO-URL"
}'The response is a job — poll GET /v1/jobs/{id} or use
webhooks/SSE; when
status is completed, result_url is the hosted output.
Good to know
- The length is the audio's, and so is the price (per started second). The speech follows the audio, in any language.
audio_durationis optional: it only makes a batch's credit estimate exact, and the charge is always the measured audio. - Reference media (
reference_*_urls) is not supported — 422. - Prompt cap: 1,000 characters.