|
 |
JSR nous apporta ses lumieres en ce 02-05-2007 03:37:
> Hi,
> I have a problem when I combine radiosity and transparent objects.
> Splotchiness appears behind glass in the second pass.
> I use 2 pass radiosity:
> 1st pass.- No transparency, no reflection, no area_lights, save_radiosity
> 2nd pass.- transparency, reflection, area_lights, load_radiosity
>
> The parameters are:
> #if (RAD)
> radiosity {
> brightness 1.0 //[1.0]
> recursion_limit 3 //[3]
> count 800 //[35]
> error_bound 0.2 //[1.8]
> gray_threshold 0.5 //[0.0]]
> pretrace_start 0.05 //[0.08]
> pretrace_end 0.01 //[0.04]
> nearest_count 15 //[5]
>
> #if(SAVERAD)
> save_file "escena.rad"
> #else
> pretrace_start 1
> pretrace_end 1
> load_file "escena.rad"
> always_sample off
> #end
> }
> #end
>
> I post an image for an example.
>
> Regards,
>
> Joel
>
>
> ------------------------------------------------------------------------
>
In the first pass, the glass doors are present and opaque. Result: there is NO
radiosity sampling behind them during that pass, leaving all the sampling to the
second pass.
Suggestion: REMOVE transparent objects during the first pass.
Your count may be to low, increase it some, somewhere from 1000 to 1600.
A low error_bound is good to deepen shadows. Your shadows are almost black. You
probably can increase it a good deal.
Increasing nearest_count to 20 (max value) can help reducing splotchiness.
If you need the extra samples of a small error_bound, using a smaler
low_error_factor can help by gathering more samples during the last pretrace step.
Some of your dark spots may be caused by rays hiting exactly a corner and geting
radiosity samples from "outside" the scene. A very small change in the camera
position can often remove them, or make them change location.
Also, changing the pretrace_start valus may, in some cases, help.
--
Alain
-------------------------------------------------
A husband is what is left of a man after the nerve is extracted.
Post a reply to this message
|
|
