Mixer AI › Knowledge Base › How to › Transfer motion from video to a character

Transfer motion from video to a character

Motion Control needs compatible sources more than a long prompt: a character image and a performance clip. The model needs to see the body and relate the poses.

Try it on Mixer AI

Step by step

Choose footage without cuts, occlusions or the performer leaving frame. Use full-body framing for dance and match the photo's crop for gestures.

Upload one photo and one clip. Choose video or image orientation, noting their different duration limits.

Assign appearance to the photo and movement to the video. Test a short passage before a long run, inspecting hands, legs and head turns.

Which model to use

Kling 2.6 Motion Control

A photo supplies the character; a video supplies the performance, from dance to conversational gestures.

Kling 3.0 Motion Control

Transfer a recorded performance to your character and choose which source supplies the background.

FAQ

Can I describe a dance instead of uploading video?

No. This workflow needs a recorded performance.

Why are hands cropped?

The photo may lack space, or the footage already crops them. Prepare a wider frame.

Should I set the background separately?

Kling 3.0 Motion Control can use the photo or video's background. Avoid changing the entire costume and lighting at the same time.

Try it on Mixer AI