Primal Extended Position Based Dynamics for Hyperelasticity

被引:3
作者
Chen, Yizhou [1 ,2 ]
Han, Yushan [1 ,2 ]
Chen, Jingyu [1 ]
Ma, Shiqian [3 ]
Fedkiw, Ronald [4 ]
Teran, Joseph [2 ]
机构
[1] Univ Calif Los Angeles, Los Angeles, CA 90025 USA
[2] Ep Games Inc, Cary, NC 27518 USA
[3] Rice Univ, Houston, TX USA
[4] Univ Calif Davis, Davis, CA 95616 USA
来源
15TH ANNUAL ACM SIGGRAPH CONFERENCE ON MOTION, INTERACTION AND GAMES, MIG 2023 | 2023年
关键词
Position-based dynamics; physics simulation; constrained dynamics; ANIMATION; MUSCLE; MODEL;
D O I
10.1145/3623264.3624437
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The Extended Position Based Dynamics (XPBD) approach of Macklin et al. [2016] addresses the issues with iteration-dependent behavior in the original Position Based Dynamics [2007] (PBD) which itself is a powerful method for the real-time simulation of elastic objects. However, it is limited in its application to hyperelastic solids. It can only treat models with a strain energy density that is quadratic in some notion of constraint. Furthermore, we show that even when applicable the formulation does not always lead to convergent behaviors with hyperelasticity. We isolate the root cause in the approximate linearization of the nonlinear backward Euler systems utilized by XPBD. We provide two fixes to these terms that allow for convergent behavior. The first (B-PXPBD) is a small modification to an existing XPBD code, but can only be used with models addressable by the original XPBD. The second (FP-PXPBD) is a more general formulation that extends XPBD (and our residual correction) to arbitrary hyperelasticity. We show that our modifications allow for convergent behavior that rivals accurate techniques like Newton's method when the computational budget is large without sacrificing the stable and robust behavior exhibited by the original PBD and XPBD when the computational budget is limited.
引用
收藏
页数:10
相关论文
共 62 条
[1]  
Baraff D., 1998, Computer Graphics. Proceedings. SIGGRAPH 98 Conference Proceedings, P43, DOI 10.1145/280814.280821
[2]   Real-time subspace integration for St. Venant-Kirchhoff deformable models [J].
Barbic, J ;
James, D .
ACM TRANSACTIONS ON GRAPHICS, 2005, 24 (03) :982-990
[3]  
Batty C, 2007, ACM T GRAPHIC, V26, DOI [10.1145/1239451.1239551, 10.1145/1276377.1276502]
[4]  
Bertsekas D.P., 1995, NONLINEAR PROGRAMMIN, DOI [10.1057/palgrave.jors.2600425, DOI 10.1057/PALGRAVE.JORS.2600425]
[5]   A 3D model of muscle reveals the causes of nonuniform strains in the biceps brachii [J].
Blemker, SS ;
Pinsky, PM ;
Delp, SL .
JOURNAL OF BIOMECHANICS, 2005, 38 (04) :657-665
[6]  
Bonet J, 2008, NONLINEAR CONTINUUM MECHANICS FOR FINITE ELEMENT ANALYSIS, 2ND EDITION, P1, DOI 10.1017/CBO9780511755446
[7]   Projective Dynamics: Fusing Constraint Projections for Fast Simulation [J].
Bouaziz, Sofien ;
Martin, Sebastian ;
Liu, Tiantian ;
Kavan, Ladislav ;
Pauly, Mark .
ACM TRANSACTIONS ON GRAPHICS, 2014, 33 (04)
[8]   Distributed optimization and statistical learning via the alternating direction method of multipliers [J].
Boyd S. ;
Parikh N. ;
Chu E. ;
Peleato B. ;
Eckstein J. .
Foundations and Trends in Machine Learning, 2010, 3 (01) :1-122
[9]   A Simple Geometric Model for Elastic Deformations [J].
Chao, Isaac ;
Pinkall, Ulrich ;
Sanan, Patrick ;
Schroeder, Peter .
ACM TRANSACTIONS ON GRAPHICS, 2010, 29 (04)
[10]  
Chen Y., 2023, Supplementary technical document