Octree Pruning for Variable-Resolution Isosurfaces

Cell-based polygonal isosurface generation is a common and useful technique for visualizing sampled scalar fields. Typically, however, the number of polygons in the surface model can become quite large as the size of the data set increases. In this paper, we present octree pruning techniques designed to compress the isosurface model while maintaining an accurate representation. We show results for several data sets, and demonstrate speed improvements of over two orders of magnitude over our previous results. A key feature of this work is the use of a volumetric error criterion to quantify model accuracy.