Motion Sync
Motion sync transfers movement from one video onto a still photo: you supply a driving clip showing the motion and a target image showing who should perform it, and the model animates the photo to follow the clip. Unlike image to video, where you describe the motion in words and hope, here the motion is given exactly — every step, gesture and turn comes from footage you chose.
That precision is the point of the tool, and it is also what makes the pairing rules matter more than the prompt. This route is not connected yet — the controls and prices are real, the Create button is not.
The pairing decides the result
Almost every failed motion sync comes from a mismatch between the driving clip and the target image rather than from the model. The model maps a pose sequence onto a body; if the two bodies are framed differently, that mapping has to invent the difference, and invention is where things go wrong.
The single most useful habit is to choose the driving clip to fit the photo rather than the other way round. You usually cannot reframe the photo, but you can almost always find a different reference video.
| Match this | Works | Breaks |
|---|---|---|
| Framing | Both full body, or both waist-up | Full-body clip onto a head-and-shoulders photo |
| Orientation | Both facing the camera | Profile clip onto a front-facing photo |
| Aspect | Both portrait, or both landscape | Landscape clip onto a tall portrait photo |
| Subject count | One person in each | Two dancers in the driving clip |
| Motion range | Movement that stays in frame | Subject walking out of frame |
What the photo should be
The target image has to show the whole of whatever will move. A photo cropped at the waist cannot produce a walk, because the legs it would need are not in the picture — and the model will either invent them badly or keep the frame static below the crop.
A plain, uncluttered background helps more here than in any other tool in this studio. The subject is about to move across ground that has to stay still; the more structure there is behind them, the more obvious it is when the model shifts it.
- Full body in frame if the motion involves legs. Waist-up is fine for gestures and upper-body movement.
- Arms slightly away from the torso rather than crossed or hidden. Limbs that overlap the body in the photo are the ones that come apart first.
- A simple background. Foliage, patterned wallpaper and crowds all warp visibly as the subject moves against them.
- Sharp focus on the subject. A shallow-depth-of-field portrait with a blurred body animates worse than a plain sharp one.
Motion sync, image to video, or lip sync?
All three animate a subject, and they differ in where the motion comes from. Image to video invents the motion from your written description — the most freedom, the least control. Motion sync copies motion from footage you supply — the most control, and it requires you to have or find that footage. Lip sync drives only the mouth from an audio track, leaving the rest of the frame as it was.
If you know exactly what the movement should look like and can find a video of it, motion sync will get you there far more reliably than any prompt. If you only have a vague idea, a prompt is easier than sourcing a driving clip.
When not to use this
- Making a real person appear to do something they did not do. Use a photo of someone who agreed to this, and do not use it to fabricate someone’s behaviour.
- Photos cropped above the motion. A waist-up portrait cannot be made to walk; the legs are not in the picture.
- Driving clips with more than one person in frame. The model maps one subject.
- Complex hand work. Fingers at contact points are the least reliable part of generative video, and copied motion does not fix that.
- Busy backgrounds you need to stay put. The ground behind a moving subject warps, and a patterned or crowded background makes it obvious.
- Matching a specific face precisely. Identity softens as the subject moves, as it does everywhere else here.
Questions people ask
What is motion sync?
Motion sync takes the movement from a video you supply and applies it to a still photo you supply, so the person in the photo performs what the person in the video did. The motion is copied rather than invented, which is what separates it from image to video.
What kind of driving video works best?
One person, facing the camera, framed the way the target photo is framed, moving within the frame throughout. Match the aspect ratio too. Most failures are a framing mismatch between clip and photo, not a limitation of the model — and the clip is usually the easier of the two to swap out.
Can I use a waist-up photo?
Only for motion that is also waist-up. A dance or a walk needs legs, and a photo that does not contain legs cannot be made to use them convincingly. Crop your driving clip to match what the photo actually shows.
How is this different from image to video?
In image to video you describe the motion in words and the model invents it, which gives you freedom and little control. In motion sync you hand it the exact motion as footage, which gives you precision at the cost of having to find the right clip.
Why does the background move when only the person should?
Because the whole frame is regenerated as the subject moves, not just the subject. A plain background hides this well; foliage, patterns and crowds make it visible. Choosing a photo with a simple background is the fix.
How much does it cost?
The per-clip price of whichever model you choose, shown on the Create button before you spend anything.
Related tools
Checked 20 September 2026.