Quote -
when I look into a room with a sliding glass door half open, I don't see this much vertical deflection,
even tho glass IOR is 1.5 or greater, not 1.2.  it appears that the refractive glass is preventing some
lite from getting into the box, and is also preventing it from bouncing back out.  presence of small
amt. of diffuse has no effect AFAICT.

I've been meaning to address this with a diagram but I've been kind of busy.

The thing about thin glass is this: it has the same refraction as any glass - it is glass. But it is thin. When light enters the glass, it bends towards the normal. Then it travels a very short distance through the glass and reaches the other side. From there, the reverse refraction happens, bending the light back to the original direction it had when it entered in the first place. The result is a photon passing though thin glass does not change direction at all. It simply hops off its original path a tiny bit, but its ultimate path is parallel to its oriignal path. This is completely different from thick or curved glass.

Poser's Refract node does not understand that it should be using the specified IOR when passing from outside to inside, and then the inverse IOR when passing from inside to outside. This is why it looks wrong. It is doing a double entry refraction, instead of a proper entry/exit refraction.

The usual hack to deal with this when using a single layer mesh is to simply set the IOR to a really small value, like 1.05.

---

Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)