AI Motion Capture

Start by creating a new chat with Kapwing's AI assistant, Kai. Upload a video with a movement you want to recreate, then select Seedance 2.5 from the model selector in the prompt box. Enter a prompt requesting a simplified motion capture video. We recommend starting with the prompt linked above and found in the FAQ just below this, though you can also write a mocap prompt from scratch.
Your source footage will guide the subject’s pose, timing, direction, and overall movement. Once your video is generated, you can click to download it as MP4, request changes with additional prompts, or use it as a motion transfer reference video by uploading a character image.

Here is a prompt you can use to generate an AI mocap video:

Transform the input video into a simplified 3D motion-capture visualization. Preserve the original subject’s pose, timing, direction, and overall movement, but replace their appearance with basic geometric forms: spheres for joints and simple cylinders or tapered capsules for limbs and torso. Reduce details and smooth out small, incidental movements while keeping the main actions easy to read.

Use a clean, consistent 3D style with uniform matte materials, simple lighting, and a plain background. Keep the subject fully visible and centered. Maintain the original camera angle and framing, and keep the geometry attached to the body as it moves. Add facial features and expressions but do not add clothing, textures, props, extra limbs, or new actions. The result should look like a basic animated mocap rig made from simple 3D shapes, not a realistic person or a static diagram.

Note: You can adjust it to request no facial expressions or a different look, like a sketch, wooden block figure, or geometric 3D rig. You can adjust the style and level of detail using your prompt, even as you keep the main action recognizable.

Use AI motion capture to visualize full-body skeletal movement, including walking patterns, athletic actions, dance moves, and martial arts. It can also carry hand motion and gestures, such as waving or pointing, from the source clip.

Seedance 2.5 can recreate facial expression cues too, including eye and lip movement. The result gives you a simplified way to show body language and expressions in a video.

Kapwing uses the latest AI models, including Kling, Veo, Happy Horse, Minimax, and Seedance. For the best results, select Seedance 2.5 from the dropdown model selector menu in the Kai prompt box.

Choose a clip where the subject and their movement are easy to see. Clear framing and fewer cuts or obstructions can help Seedance follow the action and keep the subject visible in the result. The best reference videos feature one or two clear human subjects.

For hand motion, facial expressions, or detailed footwork, use footage where those movements are visible. A clear reference makes it easier to recreate the gestures and timing you want to show.

Yes, you can use your generated mocap video as a movement reference in AI Motion Transfer, then pair it with a character, model, or influencer-style image. The reference clip can help guide new performances across different subjects.

Save clips for movements you may want to reuse, such as a walk, dance, gesture, or athletic move. That gives you a growing library of motion references for future videos.

Create mocap-style references for game character concepts, choreography, sports demonstrations, martial arts sequences, or instructional videos. The simplified look can help make a movement easier to follow or give a video a distinct visual treatment.

You can also use the generated clip as a starting point for character videos, social content, or creative variations. Since the result is a rendered video, it works as footage you can edit and share.

Yes, you can adjust your prompt to create different motion capture styles from the same source footage. Try sketch lines, wooden block figures, or simple shapes like spheres for joints and cylinders for limbs.

You can also describe materials, the background, and how much detail to include. This gives you a way to make the movement clearer or fit the video’s visual style.