Manuel Kasten wrote:
> I just rendered me a new background image for my desktop PC. I tried it 
> with radiosity, but even with rad_def.inc's Radiosity_Final there were 
> artifacts left (and it is still not finished, after 2.5 days). So I 
> decided to give MCPov a go and try it unbiased. 10 hours later, this is 
> the result.
> 
> Original render 1680x1050. About 2000 passes in MCPov in 10h. Levels 
> retouched an resized with Gimp. Full size hq image in next message.
> 
> Manuel Kasten

Nice render. I remember doing a couple renders of these some time ago. I 
can't find the link. It's interesting to see how Pascal's Triangle works 
in three dimensions.

If you would like another way to achieve high-quality radiosity in the 
official version of POV, I have one that I use sometimes. It's even 
faster in some cases. The idea is to add extra virtual radiosity samples 
for each surface. I do this by adding a slight surface normal to all the 
scene's objects, like so:

  #default{ finish{ambient 0} normal{bumps .5 scale .0001} }

You can then get away with lower radiosity settings:

  #declare global_illumination = on;
  global_settings{
   #if(global_illumination=1)
    radiosity{
     count 30 error_bound .002
     nearest_count 1
     recursion_limit 1
     normal on
     brightness 1
     gray_threshold 0
     adc_bailout 0.1
    }
   #end
  }

Render with high antialising, and it should look fairly smooth:

  +a0.03 +am2 +r3

It's good to have more than a couple ways of accomplishing the same thing.

Sam

Post a reply to this message