DrawMotion: Generating 3D Human Motions by Freehand Drawing

Researchers introduced DrawMotion, a diffusion-based framework for generating 3D human motion from text plus freehand stick-figure sketches.

It uses a Multi-Condition Module to fuse text and sketch inputs, and supports training-free guidance for more flexible control.

The system can automatically create stickman sketches across datasets and keeps motion fidelity high in experiments.

User studies reported about a 46.7% reduction in user time, making motion specification faster and more expressive than text alone.