The two tone mapping algorithms I used are the one presented in class (with a k value of 0.003) and Ward's (in Pete's Tiger book). The one on the left is Erik Reinhard's (i.e. the one given in class) and the other is Ward's. Note that I did include the 683 multiplier when computing X, Y, and Z - the k value would need to be considerably higher otherwise.

I compiled -Ofast on burn.sci.utah.edu and used 14 threads (it is a 14 processor machine). Times given are for that configuration. All calculations were done using floats. All rays took 6 bounces. I used 15 spectral bins.

Note: The down sizing caused the images to look better than they actually are. Please click on them to see the true results.

100 rays/pixel; 3:58.41s 979 sequences of 25 samples used	1024 rays/pixel; 32:58.74 97 sequences of 256 samples used
10,000 rays/pixel; 5:19:24 47 sequences of 2500 samples used	65,536 rays/pixel; 34:50:23 31 sequences of 126^2 samples used

I'd like to do the following. When I do I'll post some images:

• Explicit Direct Lighting
• Russian Roulette (for comparison)
• Gamma correction
• Ray Traced Image (for comparison)
• Image with no direct lighting (for comparison - to show how much indirect lighting adds)