What is it about?
This cool new numerical solver adds next-gen physics into virtual muscle and cloth simulations! We use a novel per-node framework to increase the fidelity of (X)PBD while maintaining the speed. Come and check out the cool muscle simulation! We advance speed by 6X as compared to traditional methods.
Featured Image
Photo by Tom Pumford on Unsplash
Why is it important?
The novel physics solver offers a unified solution for quasistatic or dynamics simulations of cloth and solid. It has the long-sought property that the simple iteration count parameters acts as a knob between realtime simulation/reasonable quality to offline simulation/movie-level quality. It is easy to implement and parallelize. It boasts more robustness and better convergence rates than the other existing numerical methods.
Perspectives
I encourage people to try our solver and don't be intimidated by the complicated notations in the paper. The solver was developed with a very practical goal in mind: to bridge the gap between Newton's method and XPBD. Indeed we succeeded. The solver is faster than Newton's method for offline simulations and has better results than XPBD for small fixed budget. The only controlling knob would be number of iterations. I would like to thank Brice Criswell and Ryan Kautzman, along with other co-authors for their invaluable help and advice.
Yizhou Chen
UCLA; Epic Games, Inc
Read the Original
This page is a summary of: Position-Based Nonlinear Gauss-Seidel for Quasistatic Hyperelasticity, ACM Transactions on Graphics, July 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3658154.
You can read the full text:
Resources
Contributors
The following have contributed to this page