Multi-layer Volume Ray Casting on GPU

Abstract

We present multi-layer volume ray casting that integrates various volume rendering enhancements into a unified framework, including empty-space skipping, volume clipping, embedding opaque and semi-transparent objects, volume editing through constructional solid geometry (CSG) operations, etc. The central idea is to consider all these objects as volume-of-interests(VOIs). Each VOI is assigned a priority number to determine how the overlapped regions are handled. All the information of the VOIs are encoded into a ray-layer buffer through depth peeling combined with layer simplification. Each pixel of a ray-layer buffer contains the information of a set of ray segments, including the starting locations and the material IDs. The multi-layer ray caster then renders (or skips) each ray segment with the proper viewing parameters depending on the material IDs. The ray-layer buffer is also used to estimate the gradients of surfaces between the layers to improve shading.