AI Motion Transfer — Your Character, Someone Else's Performance

Motion control is the mode where a video model takes the movement from a driving video and applies it to a different subject: the performance stays, the performer changes. Motion transfer is the same job under another name. Kling calls its version motion control. Versely runs the job two ways: Copy the moves in Versely Remix, and Kling Video V3 Standard and Pro or DreamActor V2 on a still.

For Copy the moves in Versely Remix: a clip (a trending reel, a link or an upload) and the person or character who should perform it, as a photo, an avatar or a ready-made character. For the still route: a driving video and a still of the character, with framing, distance and orientation as close between them as you can manage.

You give it a driving video and a still of your character, and it re-performs the video's movement as that character at the video's timing. On Kling V3 the driving video must show the head, shoulders and torso and run 3 to 30 seconds, and you can choose whether the character's orientation follows the video (the default) or your image. Standard renders 720p and Pro renders 1080p.

Copy the moves uses 4 to 15 seconds of a clip, from uploads up to 200 MB or a link, with up to 8 photos. Kling V3 takes one driving video (MP4 or QuickTime, up to 100 MB, 3 to 30 seconds, at least 340 px, aspect ratio between 2:5 and 5:2) and an optional prompt of up to 2,500 characters. DreamActor V2 takes a character image (JPEG or PNG, up to 4.7 MB, 480x480 to 1920x1080) and a driving video (up to 30 seconds, MP4, MOV or WebM, 200x200 to 2048x1440).

That depends on the route. Copy the moves re-renders your whole person or character in the clip's motion. DreamActor V2 is a face-reenactment model that carries expressions and head movement from the driving clip. Kling V3 motion control re-performs the movement of a driving clip that shows the head, shoulders and torso. Each model's page lists its inputs.

It depends on the route. Copy the moves in Versely Remix keeps the clip's place, camera and sound unless you add a new place or style. Kling V3 motion control takes the backdrop from the driving video by default and can take it from your image instead. DreamActor V2 moves the face and head in your still, so the rest of the picture is your image.

Image-to-video invents motion from your description. Motion transfer copies motion you have already captured. If the exact timing matters (music, choreography, a gesture that has to hit a word), the transfer route is the only one that gives it to you.

Copy the moves keeps 4 to 15 seconds; a longer clip opens a scrubber so you choose the part. Kling V3 and DreamActor V2 take driving clips up to 30 seconds. All of them bill by length, so cost scales with the performance. Short, deliberate clips transfer more reliably than long ones and cost proportionally less.

No. Every transfer costs credits, shown before you confirm; there is no free generation tier. Copy the moves starts at 30 credits for 5 seconds at Lite 480p. Free on-device helpers under /free-tools can trim a driving clip before you upload it.