Space-Time Adaptive Simulation of Highly Deformable Substances

Mathieu Desbrun (1) , Marie-Paule Cani (2)

1 CS CALTECH - Computer Science Department
2 iMAGIS - Models, Algorithms and Geometry for Computer Generated Image Graphics

This report presents an approach for efficiently yet precisely simulating highly deformable substances ranging from solids to liquids. The key idea is to use a state equation for specifying the dynamics of the substance. During a simulation, the material is sampled by particles that derive their interaction forces from this state equation. Since this ensures the same qualitative behavior whatever the discretization rate, an adaptive scheme can be used during simulations: the particle system adapts over space and time according to a given compromise between precision and computati- onal efficiency. The system refines (i.e., particles are subdivided) in areas undergoing large or fast deformations, while it simplifies (i.e., neighboring particles are merged) in stable regions. Meanwhile, the values of the individual integration time steps used for each particle are automatica- lly adapted to avoid instabilities. An active implicit surface is used to visualize the substance. It smoothly coats the particles and filters over time internal changes of granularity.

Mots clés: SIMULATION; DEFORMABLE OBJECTS; PARTICLE SYSTEM; SPACE-TIME ADAPTION

Domaines: Autre [cs.OH]

Mathieu Desbrun, Marie-Paule Cani. Space-Time Adaptive Simulation of Highly Deformable Substances. [Research Report] RR-3829, INRIA. 1999. ⟨inria-00072829⟩