Ok, I got a little bit longer this morning, and got the Beer-Lamberts law function correctly implemented, and it looked a LOT better when it worked correctly. I had some issues with distance and how the absorbtion was calculated, but finally its working.
I’m still just using the depth buffer to get the distance, using two textures.. I need to render to more depth buffers of the object gets complex( like the spikeball ), and use all of them to get the correct distance. I will post more on this later, but first I want to get the transmittance algorithm working correctly ( Depth peeling ).
I decided that I want to devide the transmittance tutorial into three parts. One for the transmittance, one for reflection on our object and one for refraction. I think I will have the first one ready by today or tomorrow so if you think this is interesting, get ready to get some code to play around with! 🙂
Anyway, here is todays result, and don’t mind the artifacts on the spikeball and abstract model:
Spikeball got some depth buffer artifacts, and a problem because the depthbuffers aint overlapping 100 correct. I need to find out whats wrong..
Also, Depth-peeling can be used to get a good result on the distance the rays are traveling inside the object.
Sphere is correct at the moment.
This is a complex mesh and you can clearly see the depth buffer problem.
Looks much better, but still having the depth buffer problem.
Modern art using the GPU 
Did you look at this article? http://developer.nvidia.com/object/real_time_translucency_gdc2004.html Might help you with the depth problems.
No I haven\’t, thank you very much for this. It looks really promising 🙂