Silhouette Extraction for Shadow Volumes Using Potentially Visible Set
Josef Kobrtek, Tomáš Milet, Adam Herout · Journal of WSCG · 2019
Figure 1: left: Wireframe representation of a given model.middle: We voxelize the space around the model.One voxel on the lowest octree level is selected, based on the light position, and all potentially-silhouette (need to be tested) and silhouette edges (guaranteed to be silhouette) can be collected by ascending the octree hierarchy.right: Red coloured edges are those that are a part of the silhouette after testing the set of potentially silhouette edges (all red and black ones).Only a small subset of model edges need to be tested, which considerably reduces the computational complexity.