Text-to-Skeletal Motion

Definition

Generative machine learning models that synthesize continuous, biologically plausible 3D human skeletal movements and kinematic trajectories conditioned on natural language descriptions.

Key Characteristics

• Diffusion or transformer architectures operating on SMPL-X parameter manifolds.
• Temporal smoothness and foot-contact constraints preventing sliding.

Applications

• Procedural character animation
• Virtual reality avatars
• Choreographic assistants

Mentions in Source

• “Generating 3D SMPL-X skeletal animations directly from natural language prompts.” — kimodo-cpp-ggml-motion-synthesis