Texture mapping volumes using GPU-based polygon-assisted raycasting

In this paper, we present a fast GPU-based raycasting technique that utilizes polygonal meshes to accelerate raycasting and texture map two-dimensional image detail onto volumetric surfaces, independent of their resolution. The method operates by extracting one or more meshes to enclose the visible region of the volume and allow surfaces to be textured. These meshes determine ray start-end positions and store the information that will be mapped onto the volume. Using a multi-pass approach, we first peel depth layers from the rendered geometry, storing color and depth in a series of corresponding buffers. Second, a GPU-based raycasting pass is performed that traverses rays between the outermost mesh layers. Ray-layer intersections are determined so that the color texture mapped onto the mesh is composited in the correct order with volume samples. The proposed method can be used to enhance volumetric objects in games and other applications.