PhysSkin++ learns self-supervised neural skinning fields for real-time physics animation across diverse 3D shapes
Yuanhang Lei · hf · 2026-10-21
PhysSkin++ introduces a framework that learns continuous neural skinning fields from static geometries, enabling real-time physics-based animation across diverse 3D shapes and spatial discretizations without per-object training:
- A physics-informed self-supervised learning strategy combines skinning-field normalization with conflict-aware gradient correction and stabilized joint optimization under physical and orthogonality constraints—no simulation trajectories required.
- The learned subspace supports efficient reduced dynamics with user control and contact interactions, and binds directly to meshes, point clouds, and 3D Gaussians.
- For rigged animation, a self-supervised complementarity objective learns compliant elastic responses without enforcing complementarity constraints at runtime.
Experiments demonstrate cross-shape generalization, real-time reduced simulation, and diverse applications.
More from Research
- Stanford research shows switching agent harnesses mid-run cuts cost and boosts performance — StanfordAILab · 2026-10-21
- Ropedia releases open W0 world model trained on 10,000 hours of egocentric human video — liuziwei7 · 2026-10-21
- ROTATE, an open-ended training method for generalizable teamwork agents, wins NeurIPS 2026 Spotlight — PeterStone_TX · 2026-10-21
- EdgeKernelBench: LLM-Generated Kernels Hit 64.5% Correct but Only 25.5% Fast on Edge Backends — encodelab · 2026-10-21
- MMEnterprise-Bench from KAIST Tests Enterprise Agents on Conflicting, Evolving Multimodal Records — Seohyun Lee · 2026-10-21
- TRACE Organizes Daily Clinical Events into Trajectories for Inpatient Risk Prediction — Minjun Kim · 2026-10-21