Progressive volume rendering of unstructured grids on modern GPUs

Figure 1: A sequence of progressive volume rendering steps for the SF1 dataset consisting of about 14 million tetrahedra. We have developed a new progressive technique that allows real-time volume rendering of large tetrahedral meshes. Our approach incrementally streams portions of the mesh to a hardware-assisted volume renderer and displays approximations that increase in qual-ity until a full-quality rendering is achieved. The results of previous steps are stored as image data and re-used in each subsequent refine-ment, thus leading to an efficient rendering. Our interactive rep-resentation of the dataset is efficient, light-weight, and high qual-ity. The result is a level-of-detail framework for interactive explo-ration of tetahedral datasets too large to render using conventional approaches.