Sparse Volume Rendering with Per-Pixel Linked List

Abstract:
With the improved capabilities in data acquisition, scientific visualization using volume rendering has to deal with more and more complex situations. These data contain fine and intricate structures with a large number of holes that pose huge challenges for efficient empty space skipping. Inspired by OIT (Order-Independent Transparency) algorithm, this paper proposes an object-order empty space skipping approach based on the per-pixel linked list. We innovatively combine the independent ray-casting of each volume brick with the OIT algorithm to achieve efficient rendering of sparse volume data. Experiments show our method has better performance than skipping all of the outer empty space via rasterization of tight-fitting bounding geometry.