PhysSkin: Real-Time and Generalizable Physics-Based Animation via Self-Supervised Neural Skinning

TL;DR

PhysSkin is a novel neural framework that enables real-time, physics-based animation with high generalization across diverse 3D shapes using self-supervised learning and neural skinning fields.

Contribution

Introduces PhysSkin, a physics-informed neural skinning method with a transformer-based autoencoder and self-supervised training for generalizable, discretization-agnostic animation.

Findings

Achieves real-time physics-based animation with broad shape generalization.

Outperforms existing methods in neural skinning tasks.

Demonstrates effective self-supervised training for complex deformation tasks.

Abstract

Achieving real-time physics-based animation that generalizes across diverse 3D shapes and discretizations remains a fundamental challenge. We introduce PhysSkin, a physics-informed framework that addresses this challenge. In the spirit of Linear Blend Skinning, we learn continuous skinning fields as basis functions lifting motion subspace coordinates to full-space deformation, with subspace defined by handle transformations. To generate mesh-free, discretization-agnostic, and physically consistent skinning fields that generalize well across diverse 3D shapes, PhysSkin employs a new neural skinning fields autoencoder which consists of a transformer-based encoder and a cross-attention decoder. Furthermore, we also develop a novel physics-informed self-supervised learning strategy that incorporates on-the-fly skinning-field normalization and conflict-aware gradient correction, enabling effective balancing of energy minimization, spatial smoothness, and orthogonality constraints. PhysSkin shows outstanding performance on generalizable neural skinning and enables real-time physics-based animation.