Accelerating the Visualization of Gaps and Overlaps for Large and Dynamic Sphere Packings with Bounding Volume Hierarchies

Abstract

The Collective Rearrangement (CR) algorithm is widely used for simulating packings of spheres to gain insight into many properties of granular matter. The quality of a CR simulation can be judged with a visualization technique by directly visualizing the gaps and overlaps of the spheres with pixel precision in each iteration. This visualization technique is based on an Graphic Processing Unit (GPU) linked list, which requires that information for each pixel of each sphere to be stored in advance and then be sorted for each pixel independently. Such requirements impose restrictions on the scale of sphere packings that can be visualized. Instead, one can use ray tracing to resolve such problems. However, there is no available practical ray traversal algorithm for visualizing overlaps and gaps based on acceleration structures for ray tracing. This paper provides traversal algorithms based on Bounding Volume Hierarchies (BVH) to address this problem and can generally make the visualization process much faster than before and reduce the global memory requirement on the GPU to render larger scenes.