Motion Control: Animate a Character from a Reference Video
Motion control on VEME maps a reference clip onto a still image, so a portrait or character repeats the dance, gesture or full-body action. On Motion Sync you upload the move, add the character you want to animate, and get a new video of that character performing the same action with face and gesture transfer.
- Reference video at least 5 seconds long, with clear motion
- Character image of the person, mascot or illustration to animateoptional on the form
- Motion description to steer the scene, light or styleoptional
Generation uses credits. Sign in and check the cost shown in the tool before running it; account offers and plan details are shown on the pricing page.
AI dance video generator, gestures and full-body action
The same motion control tool covers three kinds of movement. What changes is the reference clip you feed it and how much of the body needs to be visible.
Dance routines
Film a 10-second routine on your phone and hand it to a mascot or avatar. Sproutly, a made-up seed shop, could reuse one trending dance for its cactus mascot instead of animating it by hand.
Hand and arm gestures
Record yourself explaining something and move a course character with your own hand gestures. Keep the upper body fully in frame, because the tool can reject clips where it is not detected.
Full-body action
Turns, kicks, sword forms and walk cycles work best from a clip shot with the whole body in view and a steady camera, so the model can read every limb.
Using motion control on VEME
Pick a model and add your reference video
Open Motion Sync, choose a model, and upload or pick a reference video from your history. The form asks for at least 5 seconds with clear motion, and it can warn you if the effective motion is too short or the resolution is too low.
Add the character and describe the scene
Upload the character image you want to animate, then use the motion description box to say what the background, lighting or style should be. The motion itself comes from the clip, so describe the scene rather than the steps.
Check the settings and generate
Generation uses credits. Sign in and check the cost shown in the tool before running it; account offers and plan details are shown on the pricing page.
How motion transfer AI turns two inputs into one clip
Your character image decides how the subject looks. The reference video decides how it moves. The example below uses an invented clay gardener and a dancer filmed in a living room.
Character1. Character image
A clear, full-body still in a simple pose. Images upload as JPEG, PNG or WEBP up to 10 MB.
Motion ref2. Reference video
At least 5 seconds of clear movement. A steady phone shot with the whole move in view works best.
Result3. Result
The gardener performs the dancer's move while keeping its own hat, apron and boots. Results vary by clip, so check each take.
Motion control models you can choose
Motion Sync lists three Kling models. These options come from VEME's own model settings for this tool on 20 October 2026, and they can change.
| Model | What you give it | Options |
|---|---|---|
| kling-v3-motion-control | Character image and reference video are both required, and a text prompt is optional. | See cost in the tool |
| Kling 3.0 Omni | A required text prompt, with optional reference images and videos. | Standard, Pro or 4K mode; 16:9, 9:16 or 1:1; 3 to 15 seconds; reference type Feature or Motion; optional audio and original sound. |
| kling2.6 | A required text prompt, with optional image and reference video. | Std or Pro; 5 or 10 seconds; 16:9, 9:16 or 1:1; character orientation; negative prompt. |
Credit figures above are VEME's listed rates for that model. The live button always shows the cost for your exact settings.
Getting clean AI motion capture from video
You don't need a capture suit or sensors, only a phone clip. The quality of that clip decides most of what motion control can do with it, so a few minutes of setup saves credits.
- Keep the full upper body in frame, and the whole body for dances and kicks.
- Shoot at least 5 seconds of real movement, not a long pause with one short move.
- Use a steady camera and a clear, well-lit subject.
- Pick a character image with a similar framing, such as full-body for full-body moves.
- Low-resolution or heavily compressed clips, which the tool may reject.
- Very long clips. VEME can ask you to shorten one, and the exact maximum isn't published.
- Crowded frames where several people move at once.
- Expecting a frame-perfect copy. The reference guides the motion, and fine details can drift.
Motion control vs a capture studio or keyframing
| Approach | What it takes | Best for |
|---|---|---|
| Motion control | A phone clip, a character image and credits. | Social clips, mascots, quick tests of a routine on different characters. |
| Capture suit and stage | Markers, a calibrated space, a crew and cleanup time. | Game and film work that needs exact skeletal data. |
| Manual keyframing | An animator posing the character frame by frame. | Stylized moves that no real person can perform. |
Motion Sync outputs a video, not rig or skeleton data you can import into 3D software.
Motion control questions
Is motion control free on VEME?
Generation uses credits. Sign in and check the cost shown in the tool before running it; account offers and plan details are shown on the pricing page.
How long should my reference video be?
At least 5 seconds with clear motion. Output length depends on the model; kling-v3-motion-control is priced for 5 or 10 seconds, and Kling 3.0 Omni accepts 3 to 15 seconds. VEME doesn't publish a maximum for the reference clip, but it will tell you to shorten one that's too long.
Will my character's face and outfit stay the same?
The character image is what the model uses for appearance, so a clear, well-lit image usually holds the look better. It's not guaranteed for every frame, especially on fast turns or when the character's framing is very different from the clip. Review each take and try a closer framing match if the face drifts.
Can I make a character talk or sing with it?
Motion Sync copies body, arm and head movement from video. For mouth movement matched to a voice track or song, VEME has a separate Lip-Sync tool that takes a character video and reference audio.
What image and video files can I upload?
VEME's upload hint for images lists JPEG, PNG and WEBP up to 10 MB. The reference video field accepts common video files, and the form checks length, resolution and whether enough motion and upper body are visible. If a clip fails one of these checks, the error message says which one.
Can I keep the music from my reference clip?
The motion-control models on this page include a keep-original-sound option, so you can carry the reference clip's audio into the result. Turn it off if the clip has background noise you don't want, and add a soundtrack later instead.
Put your character in motion
Upload a reference clip and a character image, then check the result in a few clicks.