«
»

Towards Realtime: A Hybrid Physics-based Method for Hair Animation on GPU

Li Huang, Fan Yang, Chendi Wei, Yuju (Edwin) Chen, Chun Yuan, Ming Gao

This paper introduces a hair simulator optimized for real-time applications, including console and cloud gaming, avatar live-streaming, and metaverse environments. We view the collisions between strands as a mechanism to preserve the overall volume of the hair and adopt explicit Material Point Method (MPM) to resolve the strand-strand collision. For simulating single-strand behavior, a semi-implicit Discrete Elastic Rods (DER) model is used. We build upon a highly efficient GPU MPM framework recently presented by Fei et al. [2021b] and propose several schemes to largely improve the performance of building and solving the semi-implicit DER systems on GPU. We demonstrate the efficiency of our pipeline by a few practical scenes that achieve up to 260 frames-per-second (FPS) with more than two thousand simulated strands on Nvidia GeForce RTX 3080.

Towards Realtime: A Hybrid Physics-based Method for Hair Animation on GPU

Posted by christopherbatty on August 10th, 2023 Posted in Uncategorized
(Comments are closed)