In the lecture, Kayvon mentioned that we may need a mask to keep the edges sharp throughout the subdivision process. I am wondering what does this mean? For example, I can imagine that we can keep some regions constant without any update. However, how do we achieve this in practice?

@tsk, my understanding is that there's a mask to determine which vertices/edges/faces will be used (or will not be used) during subdivision. So for vertices at the sharp edges, they won't be affected by vertices on the other side of the mask, hence won't be smoothed out.