The results are in and I am (happily) surprised to get a second place! 
Thanks to all!! And well done everybody!

I would like to take the opportunity to answer to some of the more technical 
comments to my scene, mainly because they address problems and puzzles that 
I do not know how to solve properly.

There are several comments about the rather flat shadows and the flatish 
contrast of the towers in the background. I know. And I wonder how to solve 
this. The truth is that there is only one (repeat: one) light source in this 
scene: the Sun. The luminosity (in Lightsys) has been multiplied by 3.5. 
There is no ambient finish used anywhere. No ambient occlusion. Radiosity 
uses the following settings:

    radiosity {
      //saving radiosity data:
      #if (RadSave)
        #debug "saving...\n"
        save_file "TS.rad"
        pretrace_start 0.1           // start pretrace at this size
        pretrace_end   0.0025           // end pretrace at this size
        error_bound 0.5                // higher -> smoother, less accurate 
[1.8]
        count 1000                     // higher -> higher quality (1..1600) 
[35]
        always_sample on            // turn sampling in final trace off [on]
        media off                     // take media into account [off]
        normal off                    // take surface normals into account 
[off]

      //reading radiosity data:
      #else
        #debug "reading...\n"
        load_file "TS.rad"
        pretrace_start 1
        pretrace_end 1
        error_bound 0.9                // higher -> smoother, less accurate 
[1.8]
        count 300                     // higher -> higher quality (1..1600) 
[35]
        always_sample off            // turn sampling in final trace off 
[on]
        media off                     // take media into account [off]
        normal off                    // take surface normals into account 
[off]
      #end

      //common settings:
      brightness 1                  // brightness of radiosity effects 
(0..1) [1]
      recursion_limit 2             // how much interreflections are 
calculated (1..5+) [3]
      nearest_count 6               // higher -> higher quality (1..10) [5]
      adc_bailout 0.005
      low_error_factor 0.8           // reduce error_bound during last 
pretrace step
      minimum_reuse 0.015           // reuse of old radiosity samples 
[0.015]
      gray_threshold 0.5            // increase for weakening colors (0..1) 
[0]
      //max_sample 1.0              // maximum brightness of samples
    }

I believe that better shadows would have been obtained with a higher 
recursion_limit. I experimented a bit and that seemed to be the case, but it 
increased render time a lot. Overall, I think radiosity is the culprit here 
and my trouble with using it correctly. Any comment on this is more than 
welcome.

Something that puzzles me is the bright *light* in the shadow on the right 
tower, as if the left tower reflected much light. Maybe that is the case? 
This did not happen in a no radiosity image.

Also, I am not happy about the rendering of the cobblestone texture in and 
out of the shadows. The shadowy parts are *motly*, which they are not in a 
no radiosity scene.

I wonder if I should not have used different lightgroups in this scene, 
separate for the foreground and the background...

Again, any comments on these issues are more than welcome!

Thomas

Post a reply to this message