Disney’s Generative Motion Rig Designs Movement with Just a Few Poses
Disney Research Studios presents a Blender-based generative motion rig for authoring, varying, editing, and extending character motion from sparse artistic controls.
Paper 《A Generative Motion Rig for Artist-Driven Motion Authoring》
Authors Jakob Buhmann · Dhruv Agrawal · Dominik Borer · Luca Vögeli · Robert W. Sumner · Martin Guay
Research Disney Research Studios · ETH Zurich
Presented at SIGGRAPH 2026
Release 2026-07-16
Video 6 minutes 9 seconds
Turning a Generative Model into an Animator’s Rig
Disney Research Studios has unveiled the ‘Generative Motion Rig,’ which allows generative motion models to be controlled directly inside Blender. Artists need only specify a few key poses and handles for a character, then can generate complete movement between them while adjusting the segment length and noise sampling.
The research team calls this workflow ‘generative keyframing.’ Rather than creating every frame by hand from scratch or accepting a model’s output as-is, it was designed to let artists quickly explore multiple motion candidates while retaining the familiar controls of conventional rigging and keyframing.
The tool can be used not only to create new movement but also to modify and extend motion-capture clips. The video shows an artist changing poses and trajectories, comparing multiple variations, and completing a short animation in the actual Blender plug-in interface.
The key idea is for automation to expand repetitive work and the scope of exploration, rather than replace the artist’s decisions. The paper also addresses the control and repeatability required when connecting generative models to existing animation pipelines, as well as their integration with traditional rigs.