6
Create Character from Image
Upload an illustration, click each body part, and get a layered + rigged character. Uses SAM2 for precise segmentation when WebGPU is available.
Click on the head
Quick-Start Presets
Bone Hierarchy
No skeleton active
Poses
No poses saved
Shape Keys
Capture per-item visual deltas (size, color, segments) and blend them at any weight. Useful for facial expressions (blink, smile, surprise) on rigged characters.
No shape keys saved
Vision Puppeteer (webcam)
Drive the active skeleton from your webcam. Loads MediaPipe Pose
Landmarker on first use (~3 MB). Bones auto-map to landmarks by
name — name your bones left_arm, right_thigh,
spine, etc. for best results.
0.0s
Custom landmark mapping (advanced)
Override the auto-detected landmark pairs per bone. Each bone's angle is computed as atan2(landmark_to − landmark_from).