An Interactive Hybrid Renderer using Multi-Layer Depth

Current interactive hybrid (combined volume and surface) rendering techniques are only able to embed opaque polygonal objects and therefore cannot provide accurate depth perception of semi-transparent surfaces. This paper proposes a volume rendering approach based on a Multi-Layer Depth Map (MDM) that enables multiple translucent polygons to be embedded. This novel approach generates multiple depth map layers by ray tracing the polygons and sorting distances to intersections. It then blends colors and opacities of polygons and volume segments together in depth order by referencing the depth layers. Our approach is highly scalable allowing thousands of translucent polygons to be rendered in a volume at arbitrary positions in real time. To further improve depth perception, it also supports embedding smoothly shaded and texture-mapped polygons. As an example, the paper provides details for embedding view-dependent text labels in a volume. We present an optimized OpenCLTMimplementation of the renderer capable of rendering interactively a volume of 256x256x178-voxels on commodity hardware with 8K polygons embedded, and on a netbook with 1K polygons. We also present screenshots of renderings created by our renderer and by an existing rendering technology to showcase the increased depth perception of the embedded polygons.