Higgsfield Genjutsu

Take the motion from one video and give it to a character from a picture.

Genjutsu Motion Transfer: Put Any Character into Any Video’s Movement

Take the motion from one video and give it to a character from a picture. Upload a clip of someone dancing, walking, or acting, add 1 to 8 images of the character you want to see doing it, and get back a new video where that character performs the same moves.

It runs on Higgsfield’s Genjutsu model, built for video-to-video motion transfer. The source clip sets the choreography and the camera; your images set who’s on screen.

What can you do with it?

How to use it

  1. Upload the motion video. The clip whose movement you want to copy. One clear subject, full body in frame, works best.
  2. Add character images. 1 to 8 pictures of the character that should perform the motion. More angles help the model keep the look consistent.
  3. Describe the look (optional). A short prompt for outfit, style, or setting. Leave it empty to let the images speak.
  4. Pick a resolution. 720p by default; 480p is cheaper and faster for tests.
  5. Run it and download, or send the result into your next step right here.

What you get

One video at the length of your source clip, with your character performing the source motion. 720p or 480p, MP4.

Settings

Tips

All AI models & tools

See pricing and start creating on Mitte