Works well
- One person fully in frame, between 3 and 30 seconds
- A static camera, no zooms and no cuts
- The motion is fully visible: steps, arms, body turns
- Even light and a background the figure does not blend into
MOTION CONTROL • KLING 3.0
Transfer motion from video to a photo
The clip sets the action, not the text: show the dance, walk or gesture you need, add a photo of your character, and they repeat the motion. This is how you get scenes a prompt cannot describe.
Transfer motionNo subscription • Trim the clip in the browser • The form shows the price per second
How it works
- 01
Character photo
One person, full height, face and hands visible, calm background. This is the character you will see in the result.
- 02
Motion clip
Your own file or a ready template. An uploaded video can be trimmed with a slider: only the selected fragment goes into the generation.
- 03
Orientation and run
Choose where the character orientation comes from and start. A prompt is optional — the clip already carries the action.
What clip works
The model reads the pose and the trajectory, so the cleaner the source, the closer the result.
Breaks the result
- Several people or a partial frame — the model cannot tell who to follow
- Cuts, changing angles, handheld footage on the run
- Motion hidden behind objects or leaving the frame
- A file over 100 MB — cut the extra before uploading
Character orientation
As in the video
The character repeats both the motion and the turns from the clip. Good for dances and runway walks where the angle changes. The fragment can run up to 30 seconds.
As in the photo
The character keeps the orientation of the source photo and moves within it. The face survives better, but the fragment is capped at 10 seconds.
FAQ
No. The clip sets the action, so the field can stay empty. Text only helps when you want to specify the environment or the style of the shot.
Between 3 and 30 seconds. With the "as in the photo" orientation the fragment is capped at 10 seconds. You do not need to cut a long video beforehand — pick the fragment with the slider after uploading.
The price is counted for every second of the selected fragment and depends on the quality — 720p or 1080p. The form shows the exact amount before the run, and it is charged only when the generation starts.
Yes. Templates come with ready motion clips: pick a template, add the character photo and run it — nothing to upload.
The file is only needed for the generation: the service makes a normalized fragment out of it, and the source is kept for a limited time and deleted automatically.
See also
Ready to create your first AI image or video?
Describe your idea and start generating right now — in the browser or in Telegram.