Mesh compression and its hardware/software applications

Three-dimensional triangle and tetrahedral mesh are dominant representations of 3D geometric models. However, explosive growth in the complexity of the mesh-based 3D models overwhelms the storage, bandwidth, and computing capability of existing graphics systems. One solution to this problem is to use a compressed mesh representation. This dissertations presents a lossless compression-domain mesh processing paradigm for efficient encoding and manipulation of large triangle and tetrahedral meshes. First, we develop a simple and efficient triangle mesh compression algorithm called BFT. The high compression efficiency of this algorithm significantly reduces the network transfer and disk access time for the resultant mesh. At the same time, the simplicity of the algorithm makes BFT decompression amenable to hardware implementation. The feasibility of BFT decompression in hardware is demonstrated with a prototype VLSI implementation. Addition of the decompressor at the front-end of a graphics processor significantly reduces the bandwidth requirement between the host and the graphics processor during rendering. Then we present an algorithm for on-the-fly clipping of a compressed triangle mesh. This algorithm is essential to keep a triangle mesh in the compressed form throughout the rendering pipeline, even though clipping may destroy the compressed representation by removing the triangles outside the viewable area. The clipping algorithm is further extended to sort a compressed triangle mesh for bucket rendering and parallel graphics architectures. These architectures can now render the compressed mesh directly to reduce the bandwidth requirement, as compared to existing mesh representations. On the application side, a triangle mesh editing tool is developed that can directly operate on the compressed mesh representation. A small amount of meta-data enables this tool to find and modify the edited portion of the mesh by decompressing only a small portion of the compressed mesh. Compression-domain editing reduces the memory requirement and thereby improves the editing performance for large triangle meshes whose size exceeds the physical memory size. Finally, the triangle mesh compression algorithm is extended to tetrahedral mesh. This tetrahedral mesh compression algorithm is specifically designed in such a way that a rendering algorithm can incrementally render the compressed tetrahedral mesh as it is being decompressed. This integration of decompression and rendering considerably reduces the working-set size and rendering time for large tetrahedral meshes.