Hi --

Finally here are the results for the second test case.
The results showed excellent agreement, despite the
difficulty of computing an image in which all of the
illumination was indirect. There are two images summarizing
the results on tiber.nist.gov under pub/holly/CASE2
Case2a.ppm which has results for the original scaling factor
of 500,000; and Case2b.ppm which has results for a scaling
factor 250,000. In each image r is the minimum value reported,
g the average and b the maximum. Thanks to the following
people for participating:

Nick Holliman <nick@lightwork.co.uk>
   Using radiosity
   (special thanks for setting the problem)

Graham Jones <grj@physiology.oxford.ac.uk>
   Using radiosity type method presented at 
	4th Eurographics workshop

Arjan Kok <arjan@duticg.twi.tudelft.nl>
   Using a variation of progressive radiosity

Christophe Vedel <vedel@dmi.ens.fr>
   Using the technique described in 
       "Improved Storage and Reconstruction of Light Intensities
         on Surfaces", 3rd Eurographics Workshop on Rendering
	  (Bristol)

Greg Ward <gjward@lbl.gov>
   Using Radiance

(I also submitted an entry, using vanilla Monte Carlo
path tracing, allowing up to 102,400 samples per pixel
and still getting noise in the image!!).

Just to give you an idea of the results, I am attaching a
summary of the range (R=max value -minvalue) over each
of the images.

Case 3 will be coming along soon.

-- Holly

===================================================================
===================================================================

CASE2a
Results with original scaling factor of 500,000 :

R = 0 for 579 pixels
R = 1 for 64 pixels
R = 2 for 13 pixels
R = 3 for 1 pixels
R = 4 for 15 pixels
R = 5 for 2 pixels
R = 6 for 10 pixels
R = 7 for 4 pixels
R = 8 for 12 pixels
R = 9 for 7 pixels
R = 10 for 3 pixels
R = 11 for 11 pixels
R = 12 for 16 pixels
R = 13 for 12 pixels
R = 14 for 29 pixels
R = 15 for 29 pixels
R = 16 for 30 pixels
R = 17 for 43 pixels
R = 18 for 57 pixels
R = 19 for 71 pixels
R = 20 for 125 pixels
R = 21 for 112 pixels
R = 22 for 174 pixels
R = 23 for 212 pixels
R = 24 for 186 pixels
R = 25 for 115 pixels
R = 26 for 56 pixels
R = 27 for 51 pixels
R = 28 for 57 pixels
R = 29 for 65 pixels
R = 30 for 57 pixels
R = 31 for 54 pixels
R = 32 for 45 pixels
R = 33 for 28 pixels
R = 34 for 14 pixels
R = 35 for 12 pixels
R = 36 for 6 pixels
R = 37 for 8 pixels

for 37<R<154 the number of pixels is between 0 and 4
for R>= 154 there were no pixels with that large a difference.

Case2b: with scaling factor of 250,000

R = 0 for 36 pixels
R = 1 for 13 pixels
R = 2 for 34 pixels
R = 3 for 94 pixels
R = 4 for 186 pixels
R = 5 for 250 pixels
R = 6 for 292 pixels
R = 7 for 237 pixels
R = 8 for 206 pixels
R = 9 for 154 pixels
R = 10 for 128 pixels
R = 11 for 122 pixels
R = 12 for 86 pixels
R = 13 for 58 pixels
R = 14 for 51 pixels
R = 15 for 67 pixels
R = 16 for 56 pixels
R = 17 for 46 pixels
R = 18 for 60 pixels
R = 19 for 73 pixels
R = 20 for 65 pixels
R = 21 for 51 pixels
R = 22 for 30 pixels
R = 23 for 20 pixels
R = 24 for 19 pixels

for 24<R<96 the number of pixels is between 0 and 5
for R>= 96 there were no pixels with that large a difference