I had a try at rendering the image based on your source (with numerous
changes, including converting to the left handed sytem from moray's
right-handed one). For the lightsource I used an array of point lights
(12x12) rather than an area light, as I couldn't get smooth photons with an
area light.  I also changed the materials to use fresnel reflection.  One of
the most important changes here is the change in the colour definitions for
the walls.  As you can see, the caustics appear white now where they are
oversaturated.  As Nathan pointed out, you need to have non-zero components
(even if minimal) in all three colours:
Red = rgb <0.88039, 0.10196, 0.10980>
Blue = rgb <0.18039, 0.23922, 0.87843>

I found an article relating to light and colour a while ago that had stated
something to the fact that in real-life, pure R G or B colours don't exist,
they are only theoretical (or something like that, I don't have the link
anymore at present, but I'm sure it would be easy to find this information
somewhere again)

Render time 15h 43m 16s
Oddly enough, with the method I used the photons took virtually no time to
parse (<1s) yet they look very resonable.

-tgq

Post a reply to this message

Attachments:
Download 'test room.JPG' (72 KB)

Preview of image 'test room.JPG'