A discussion of how to properly handle liquid inside glasses in a raytracer made me look up the scheme that I had implemented once before, which I’m now stashing a pointer to just for safekeeping.

Journal of Graphics Tools – Papers – Simple Nested Dielectrics in Ray Traced Images

The reason that this is difficult (or can be difficult) is that often a glass of liquid is modelled as two objects (the glass and the liquid) with different refractive indices, and with surfaces that are largely coincident. This avoids the coicident surface problem by using a technique reminiscent of a scheme that I used to implement constructive solid geometry efficiently in a raytracer.

Anyway, some good ideas, I thought worth mentioning.