Interior/ Exterior Classification of Polygonal Models
Geometric Processing / Winter 2021
Summary
This project focused on building off the ideas presenting in a publication by F.S. Nooruddin and Greg Turk. In which the goal is to classify faces of a 3D model where some faces are inside of the model and some outside. These calssifications can then be used to perform tanslucent renderign where the interior features of the part are visible. To accomplish this I created a raytracing based method where a group of rays are generated surrounding the model. Anytime a ray intersects more then 2 faces on its way through the model the interior grouping of 2 (there are always faces intersected in 2s) is marked as interior. In other words as the ray strikes faces it builds a list. The first face in the list and the last face are exterior, all others are interior. This method ended up being slow due to it’s O(n*m) runtime. With n being faces and m being rays. Additionally it was innacurate on very large models without high density of rays which again made it slower still.
Fibbonaci Viewpoint Placement
To place the ray start points evenly across the model I first calculated the bounding sphere and used the fibbonaci sequence to evenly place points across the surface of the bounding sphere. These ray start points only needed to be on the half-dome surrounding the object as rays pass through the entire 3d model.