Adaptive Surface Reconstruction for SPH using 3-Level Uniform Grids

Abstract

The marching cubes algorithm is a popular method for constructing surfaces from SPH data sets. In order to preserve all of the surface details in high curvature regions and to prevent potential temporal coherence artifacts, the resolution of the underlying uniform MC grid should be set up sufficiently high. However, this requirement unnecessarily increases the resolution in relatively flat regions where the surface can be constructed with lower resolutions without changing the quality. Accordingly, excessive number of triangles are generated, the memory consumption increases dramatically, and the performance decreases. In this paper, we present a 3-level grid structure which adapts its cells according to the curvature of the fluid surface. In contrast to widely-used octrees, we propose a simple to construct yet efficient hierarchical uniform grid structure. Mesh blocks from different resolution cells are seamlessly stitched by closing cracks with new triangles which establish only 0.15% to 0.6% of overall number of triangles in average. Experiments show that in contrast to the single level low resolution uniform grid approach, the presented method reconstructs fine details properly with a comparable performance; while it produces similar results with less number of triangles, up to four times better memory consumption and up to 60% better performance when compared to the single level high resolution uniform grid approach.