The Simulation Of Fluid And Cloth Collision Based On GPU Acceleration

The simulation of fluid and cloth has attracted pervasively attention in the virtual realityand the physically-based simulation field. Especially the study of the fluid and cloth is a hottopic in the field of physically-based simulation. However the current fluid and clothalgorithms are too complex, which lead to high cost of the simulation time for each frame arewhere ten seconds or more are required. As a result we need to find a fast and efficientalgorithm to solve the problems.In this dissertation we use introduce relevant techniques to optimize the traditionalalgorithms of fluid and cloth coupling, and employ the programmable graphics hardware toaccelerate the algorithms. In addition, we also propose a parallel computing algorithm basedon the multiple processors of graphics hardware to enhance the efficiency and accuracy of theprevious methods.Specifically, we mainly research the simulation of the motion of fluid and cloth, buildingreasonable model for cloth and fluid system. Among them we suggest a mass-spring modelfor cloth, and introduce smoothed particle hydrodynamics for fluid simulation. At the sametime we build grid index and neighbor table for each fluid particle. The neighbor table isemployed to quickly search the particles around each particle, to calculate the force of theparticle. We mentioned that our algorithm detect collisions between fluid and cloth accuratelywithout missing any penetrations by dilating continuous collision detection algorithms.Studing stream mapping method, and accelerating correlation algorithm on the GPU. Afterdetecting the collision, we propose to use a new algorithm to calculate the collision responseforce. And its superimposed force respect to the previous force, so as to calculate the finallocation of the fluid particles and cloth particles.