17. GPU Accelerated Surface Reconstruction for Particle-Based Fluids
Authors: Wei Wu (Ocean University of China)Hongping Li (Ocean University of China)Tianyun Su (First Institute of Oceanography)Haixing Liu (First Institute of Oceanography)
Abstract: Animating fluids with particle-based methods has attracted great attention in computer graphics. To render discrete particles, implicit surfaces need to be extracted. But this process is often seen as a bottleneck due to the expensive computation and high memory usage especially for constructing a detailed and artifact-free surface. In this work, a GPU accelerated fluids surface reconstruction method using 2-level grid structure is employed with a scheme of arranging fine surface vertices, which could preserve the spatial locality to facilitate the coalesced memory access on GPU. Meanwhile a parallel cuckoo hashing method is taken to help reduce the memory consumption. A parallel version of the optimized surface reconstruction was performed based on CUDA architecture. In the poster, the whole algorithm was outlined, subsequently with an explanation of the 2-level grid structure. The algorithm efficiency was verified by the performance comparison with other approach.
Two-page extended abstract: pdf