Is this traditionally implemented as a recursive algorithm or is there some optimization that can be made? Also, is it always the case that the triangles are split into equal-sized triangles?
For re-positioning old vertices, are we only looking at its neighbors which are also old vertices? For instance, those vertices labelled "u", are those all old vertices, or are those the new vertices after we have subdivided?
@rileylyman194 I believe those are all old vertices. Furthermore, the positions used in this calculation are also their old positions (i.e. before the current iteration of subdivision).
Is this true for non-isometric triangles? Here we see the triangles being split using the bisections of each side, but does that change if these sides are not equal, or stretched along a curve?
To clarify this slide, since it confused me: the new vertices are the new vertices added to the midpoint of the old edges. The old vertices are the vertices of the old triangle. We are literally going through every vertex and calculating its new (x,y,z) coordinate.
@mylesdomingo Yes, loop subdivision still splits sides at their midpoints with unequal sides