It's pretty, but there is an odd singularity when you scrub between zero and non-zero variance. Is zero variance handled as a special-case for some reason?
zero variance is all copies of the same triangle tessellated together. By definition a non-zero variance has to introduce change in at least 1 triangle and in doing so has to recalculate all connected triangles. It's hard to imagine how there couldn't be a strange blip when transitioning from zero to non-zero.
