[Radiance-general] Using MinGW Radiance

[David Smith]

I had never tried it, so out of curiosity, I took Francesco Anselmo's
MinGW Radiance 3.8 installation and ran Mark Stock's Radiance
Benchmark on it in Windows XP. Surprisingly, the only real thing I had
to change was to make a copy of xform.exe in the same directory and
call it cat.exe - xform without any transformation arguments is
essentially cat (right?), and none of the cat lines had arguments. I
may have had to rename genrbox to genbox, but I think I did that a
while ago.

After downloading and unpacking the benchmark package (try 7zip), make
a batch file (a text file with a .bat extension) with the following
lines (based on the Makefile):

obj2mesh -n 15 -r 16384 lens.obj > lens.msh
oconv -f -n 6 -r 16384 materials.rad cube2f.rad > cube2f_instance.oct
oconv -f -n 6 -r 16384 materials.rad cube4f.rad > cube4f_instance.oct
oconv -n 6 -r 16384 materials.rad scene.rad > scene.oct
rpict @viewpoint @options -x 2048 -y 2048 -t 60 -o windows.unf scene.oct
pfilt -1 -e +0 -x /4 -y /4 -r .6 -m .15 windows.unf > windows.pic
del windows.unf
ra_bmp windows.pic > windows.bmp

Make sure your PATH variable is set to the MinGW installation's bin
directory, and run the batch file from a DOS prompt in the folder with
the benchmark files. This 2.13 GHz Core 2 got 4614.84 sec, on XP Pro
SP2 (683733972 rays) without Cygwin, XMing or virtualization. (Thanks
Francesco!)

--Dave Smith