Topic added Aug 7, 2024
Ai Avatar
Create a short avatar video guided by reference media while preserving the person or character across the clip.
videoA downloadable MP4 animated from the supplied start image, with an optional end-frame target.
Examples
Example outputs for this workflow
Each page shows one source image and one short sample video so you can quickly check what to expect before running.
Source: openverse_flickr

Workflow workspace
Run Ai Avatar
The workflow mode is set for this page. Check the estimated credits before you submit.
Kling 3.0 Pro · Image to Video
Image → VideoDescribe one shot with a subject, action, camera move, light, and sound.
0/1
0/1
8 credits
Your video will appear here
How it works
Ai Avatar end-to-end workflow
- 01
Upload the source file and check the preview.
- 02
Choose only the controls that materially affect this task.
- 03
Run the job and keep the progress state visible.
- 04
Compare the result, revise if needed, then download the final asset.
Inputs
Prepare these inputs before running
- Prepare the required Start image, Motion direction, Aspect ratio, Duration, Resolution before starting Ai Avatar.
- Use a small representative source first; keep an untouched original for comparison.
- State one desired change and a measurable pass condition instead of combining unrelated requests.
Result checks
What a usable result should pass
- Watch the full Ai Avatar result for subject or identity changes between frames.
- Check that the requested action and camera movement are readable rather than implied by a single frame.
- Inspect the first frame, last frame, motion continuity, visible text, audio timing, and export duration.
Limitations
What this workflow may not guarantee
- Generation can produce artifacts; inspect the result before publishing or delivering it.
- Do not upload media or use a person, voice, logo, or copyrighted work without permission.
- Short-video limits and source quality can prevent stable motion, faces, text, or audio.
Publish checks
What must work before publish
- Required inputs are validated before a run starts.
- Progress, errors, and retry state remain visible.
- The result can be inspected before download.
- Failed remote jobs do not consume credits.