How AI motion transfer works
The model reads the pose and camera movement in each frame of your source clip, then generates a new video where the character from your reference image performs the same motion in the same framing.
Independent AI video editor · keep the motion
Motion transfer takes the movement, timing and camera path from a video you own and puts them on a new character from a single reference image. The choreography stays, the performer changes.
Open the studioPut a new character from your image into the same performance. Movement, camera and timing stay.
ModelBest motion accuracy and character consistency. 720p or 1080p, clips up to 30 seconds.
ResolutionCost: 80 credits
Upload a motion video and a character image, then confirm the rights checkbox to unlock generation.
Video generation is not open yet. You can prepare inputs and download the brief now.
Files upload to secure storage only when you start a render. Credits are refunded automatically if a render fails.
Add your files to preview them here.
The model reads the pose and camera movement in each frame of your source clip, then generates a new video where the character from your reference image performs the same motion in the same framing.
Recast a dance or trend clip with your own character, turn a rough performance into an animated mascot, or preview how a choreography looks on a different body type before a shoot.
Use 4–30 second clips with one visible subject, readable movement and little occlusion. Match the reference image framing to the clip: full body for dance, upper body for talking shots. Fast spins and loose clothing can drift for a few frames.
It is a video-to-video technique that copies the movement from a source clip onto a new character taken from a reference image.
It is a similar workflow on an independent tool. AI Genjutsu is not affiliated with Higgsfield.
Yes. Dance is the most common use. Pick a clip where the whole body stays in frame.
Between 4 and 30 seconds, at 720p or 1080p.