ccotwist3d
A multi pass composite might work, I've never tried it. That's interesting idea.
If you do try it, post about it here, or start another thread for it.

---

luvver_3d
I have a Master's degree in optics, you don't need to explain to me how raytracing works. There are several main types of raytracing and many variations of each of those. All of them have certain pros and cons, but some are better (i.e. more accurate, more efficient, etc.).
The best raytracers use several different types of raytracing algorithms in conjuction and have the ability to activate or deactivate different 'features' depending on whether you require extreme speed, extreme accuracy, or anything in between. The features (variations) are not always labelled as 'raytracing' but they require that raytracing is used to implement them.
Google "Monte Carlo Method" or "Metropolis Light Transport" if you don't know how those pertain to raytracing. I'd think you can find some information about them that way. You could also probably check out "Bidirectional Refractance Formula" too.

---

You are of the opinion that a non-distributed raytracing is equivalent to a raytracing done using sophisticated randomizing , distribution, and sampling algorithms. As long as it raytracing, it's the same result.
I disagree completely. I think that not all raytracers are created equal.
Just leave it that.
If anyone happens to stumble into this discussion, they can make up their own minds about which one of those ideas is the correct one.

No hard feelings, I hope. I get deeply involved in the technical discussions, because it is something I have been very interested in for a long time.

Friends don't let friends use booleans.