Research On Dynamic Cloth Simulation

Realistic simulation of textiles is one of the most fascinating areas of computer animation. During the past decades of years a considerable amount of research has been carried out by mathematicians, physicians and engineers. Although much research has already been done, fast and realistic cloth simulation still remains a challenge. Focusing on the research of the physically based modeling with which to model clothes, the author proposes an improved mass-spring model based on rectangle grid and develops a new collision detection and response method based on local adaptive subdivision.The author first proposes an improved mass-spring model developed for rectangle meshes based on the survey of cloth modeling methods, by which the simulation of cloth is formulated unifiedly. The dynamic system of modeling and simulation is derived and solved using time differentiate method, with given composition and expression of internal and external cloth forces in the formulation. To overcome the serious weakness representing cloth properties simply of previous model, the improved mass-spring model considers mechanical properties of cloth such as stretching, shearing and bending with different stiffness of spring, which can improve the realistic effect.This thesis then gives the general framework and some typical methods of collision problem. With the need for fast cloth simulation, the author proposes an efficient solution to the collision problem, where the surfaces are triangulated. The idea is to develop a general solution applicable both to contact of fabric with external objects and to cloth self-intersection. The approach to checking for collisions is to evaluate the possible configuration of the collision: node-to-triangle, edge-to-edge and triangle-to-triangle. At the same time, the curvature of the triangles is introduced to minimize the number of self-intersections detection of the cloth.To generate realistic cloth simulation at a reduced computational cost, in this thesis, the author presents a fast cloth simulation method using local adaptive refinement and merging. A simple mesh refinement scheme named with Catmull-Clark is employed to refine the coarse mesh into the fine mesh. During the growing phase of cloth simulation, several new particles are added to the system if the cloth grids collide with rigid objects. The cloth grids are merged later when coarse grids are sufficient. Using the local adaptive refinement method, the system can solvethe problems such as super-elasticity, cracks and subdivision levels.Finally, detailed algorithms are also presented and the exploitation of SIMD instructions allows us to achieve nearly 100% FPU utilization. Results demonstrate that the proposed algorithm supports tradeoffs between simulation time and realistic results.