Physkin plus plus enables real time physics based animation across shapes
PhysSkin++: Real-Time and Generalizable Physics-Based Animation via Self-Supervised Neural Skinning
Computer Vision and Pattern RecognitionGraphics
Summary
Animating 3D objects realistically and quickly is difficult because different shapes behave differently and take diverse forms. The authors introduce PhysSkin++, a system that learns how to deform 3D shapes smoothly by understanding their physical properties without needing many examples or specific training for each shape. It uses smart math and physics rules to predict how shapes move and react to control and collisions in real time. This helps create animations that work on many types of 3D shapes without starting from scratch each time.
What this means in practice
- •For game developers: Create responsive, physics-driven character animations that adapt in real time to user inputs and collisions across various 3D models without retraining per character.$Commercial implications: Enables studios to sell games with versatile, high-quality physics-based animations that reduce development time and improve character realism.
- •For vr content creators: Integrate physics-based muscle and skin deformations into virtual reality avatars that adjust dynamically to user interactions and rig controls efficiently.
Authors
Yuanhang Lei, Baofei Liu, Xingxuan Li, Tao Cheng, Zixuan Wang, Boming Zhao, Yujun Shen, Siyuan Huang, Ruizhen Hu, Peter Yichen Chen, Hujun Bao, Zhaopeng Cui
Abstract
Generalizing real-time physics-based animation across diverse 3D shapes and spatial discretizations remains challenging. In this paper, we introduce PhysSkin++, a physics-informed framework that learns continuous neural skinning fields from static geometries to construct compact deformation subspaces. Specifically, we propose a Neural Skinning Fields Autoencoder that predicts shape-dependent deformation subspaces without per-object network training. To learn these subspaces without simulation trajectories, we develop a Physics-Informed Self-Supervised Learning strategy that combines skinning-field normalization with conflict-aware gradient correction, stabilizing joint optimization under physical and orthogonality constraints. The learned representation supports efficient reduced dynamics with user control and contact interactions, together with direct animation binding to meshes, point clouds, and 3D Gaussians. For prescribed rig animation, we further introduce a self-supervised complementarity objective to learn complementary elastic responses without enforcing complementarity constraints at runtime. Experiments demonstrate cross-shape generalization of the learned skinning fields, real-time reduced simulation, and applications to controllable, collision-aware, and rig-complementary animation.