Dynamic cloth simulation based on the adaptive hybrid integration

Abstract

Since the mid of the 80's the cloth animation has attracted more and more attention in computer graphics. Cloth animations are now expected in films, games and virtual reality. In the simulation of flexible cloth models, solving a stiff differential equation is crucial. Therefore some methods have been proposed in computer animation, to build a fast and correct integration method for that task. The simple explicit schemes, though requiring considerably less work per step, are too unstable to satisfy the real-time effect demands for cloth simulation with stiff characteristics. The implicit method is of great stability, but it does not integrate the nonlinear term and is too complex that the results and the real-time efficiency are unsatisfactory. A novel adaptive implicit-explicit scheme is presented to solve the arising differential equations. Considering the nonlinear characteristics of the deformation about a particle system, the forces acting on a particle are split into a linear stiff part and a non-linear non-stiff part. Our techniques take advantage of simulation parameters, locally in both space and time, to minimize the coupling of the system. The linear part is treated implicitly, while the non-linear part is treated explicitly, therefore this scheme further improves the efficiency of the computation at each time step. Meanwhile the number of the nonzero blocks in the sparse structure of the coefficient matrix are reduced by exploiting the physical model, as well as the static constraints in the system. This peculiarity further speed-ups the simulation. Consequently, higher accuracy, leading to fast and stable animation with higher quality, will be achieved.