PBRT book: http://www.pbrt.org/

I really was just using the types to make the code read better. At one point I did have trouble confusing the different material constants, but switching to record syntax helped with that.

A progress monitor is on my mental list of improvements, but it would involve a bit of work restructuring the code to make it happen. We’ll see how long I keep interest in it (which is also a reason why I’m avoiding putting up a package for now).

I’m already using Mercurial, so I might see how much work it would be to put up a repo for that. Long term I wouldn’t mind setting this up to do photon mapping, but I’m not sure where things will go.

For instance, rather than:

type Reflectivity = Double type Diffusion = Double type Specularity = Double type Shininess = Double type Ambience = Double

I’d use:

newtype Reflectivity = Reflectivity Double …

That way you can’t accidentally use a Shininess value as a Diffusion etc.

The book is a literate program for a ray-tracer. Much better than the toy ray-tracers that you get in most books which don’t really do global illumination. eg, lot of explanation of Monte Carlo integration, sampling, photon mapping etc.

The luxrender program is an open source project forked from pbrt, I think the gallery on this page has some impressive renders:

http://www.luxrender.net/

A couple of small points:

-If you make a cabal package for it you’ll get lots more people playing with it.

-Some form of progress monitor while you render would be good. I have no idea if this render is going to take 2 minutes or 2 days.

I recently bought the PBRT book which is very good (unfortunately very expensive too). One day I’ll get time to write a haskell version of it.