As I promised in binaries.images the source is in binaries.scene-files.
The source and this info should be enough for experiments.


   The radiosity settings are not critical for realistic results.
   More important are assumed_gamma, fade_power, fade_distance and
   light source intensity distribution.

   1. Use assumed_gamma 1.0 and set Display_Gamma in povray.ini
      to a value that you measured using a test image. Adjust
      monitor brightness and contrast using
      http://hammer.prohosting.com/~kkivisal/adjust.gif
      
   2. Use fade_power 2 and try 1.4*light_source_diameter as the
      value for fade_distance first. Don't set it too high though.
      
   3. If the light source is a flat panel use      
      spotlight radius 1 falloff 90 tightness -0.8.
      For other shapes you'll have to improvise. Remember that large area_lights
      don't produce correct direct illumination, you'll have to use several
      area_lights or point sources instead to cover the area.
      
      
   Tips.
   
   If the shadows are too bright multiply light source colors by a factor
   and divide diffuse by the same factor. Do the inverse if there is not
   enough diffused light. Generally low diffuse values are more realistic,
   try 0.5 first.
   
   If you extract a color from an image or from some color selector remember
   to inverse gamma correct it ( pow(color.x,Display_Gamma) ) before using it
   in a scene file. This way the color looks more or less the same. The same
   goes for image maps also. In PSP for example set gamma to 1/Display_Gamma.


_______________________________________________________________________
Kari Kivisalo                                          www.kivisalo.net

Post a reply to this message

About matching the Cornell box image.

I tried first with white light source but the match required too
high diffuse value. Then I red the data page
http://www.graphics.cornell.edu/online/box/data.html and changed
the main surface color to white and adjusted other colors from there.
The diffuse value 0.75 came out from trial and error cycle. It can't
be a coincidence that it's pretty close to the maximum reflectance
of the white surface.

I was a bit worried that my pseudo-scientific approach would produce
information with limited usability but now I can say that the methods
in my previous post apply to all (indoor) scenes.

Next I test photons with this scene.
http://www.graphics.cornell.edu/online/box/measured.jpg

_______________________________________________________________________
Kari Kivisalo                                          www.kivisalo.net

Post a reply to this message

Kari Kivisalo <kar### [at] kivisalonet> wrote:
:       spotlight radius 1 falloff 90 tightness -0.8.

  Shouldn't the radius be 0? If not, you will get a cone of light (of radius 1)
with constant light.

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):5;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message