So why do we want the "good" triangles if they sometimes can be less efficient than long and skinny triangles?

motoole2

@hii Here's a link to the paper referenced at the bottom of this slide. Even though long & skinny triangles may provide a good geometric approximation to a mesh, performing any computations on the same mesh could result in nasty problems. The previous slide gives a good example of this, where the area of the cylinder is poorly approximated by its triangle mesh.

So why do we want the "good" triangles if they sometimes can be less efficient than long and skinny triangles?

@hii Here's a link to the paper referenced at the bottom of this slide. Even though long & skinny triangles may provide a good geometric approximation to a mesh, performing any computations on the same mesh could result in nasty problems. The previous slide gives a good example of this, where the area of the cylinder is poorly approximated by its triangle mesh.

https://en.wikipedia.org/wiki/Delaunay_triangulation