Orthographic backface culling using nested normal cone hierarchies

This paper presents an efficient dual-space method for culling back-facing triangles of a triangular mesh under directional projection. The proposed method first performs a binary space partitioning on the points that correspond to the image of the normals under the Gaussian Map. Using this bsp-tree a nested pyramid hierarchy containing the triangle-normals is constructed. In the last step, cones are fitted to each pyramid in the nested hierarchy to improve the robustness and efficiency of the back-face culling algorithm. The results of a preliminary performance analysis of the proposed method is also given.