Previous | Next --- Slide 13 of 36
Back to Lecture Thumbnails
OtB_BlueBerry

Why would someone use a non-triangle mesh? To me a triangle mesh seems superior to a non-triangle polygon mesh, except for a slightly higher space usage. In a general polygon mesh, what if a "polygon" does not actually lie in a plane? We would have to constantly worry about this constraint. Triangle meshes never suffer from this issue. They are more simplistic and make life easier.

motoole2

@OtB_BlueBerry This is an interesting question. The nice thing about a quad mesh is that it subdivides better than a triangle mesh, as we show in this slide by applying the Catmull-Clark subdivision algorithm to both quad meshes and triangle meshes (in the following slide). While we may eventually want to convert any mesh into a bunch of triangles that can be efficiently rendered using out rasterization pipeline, a quad-mesh representation is a good representation for when we want to increase the level of details in our models. As described in the paper on the development of Pixar's short titled Geri's game, "Quadrilaterals are often better than triangles at capturing the symmetries of natural and man-made objects; tube-like surfaces — such as arms, legs, and fingers — for example, can be modeled much more naturally with quadrilateral." For this reason, animators may choose quads over triangles.