POV-Ray : Newsgroups : povray.beta-test : Radiosity artifacts at low error_bound in 3.7RC3 Server Time
19 Aug 2022 19:45:04 EDT (-0400)
  Radiosity artifacts at low error_bound in 3.7RC3 (Message 1 to 4 of 4)  
From: aaz
Subject: Radiosity artifacts at low error_bound in 3.7RC3
Date: 17 May 2011 18:20:00
Message: <web.4dd2f2fc468c5d60dca745f10@news.povray.org>
Consider this scene of a hollow sphere viewed from the inside:

    difference {
        sphere { 0, 100 }
        sphere { 0, 99 }
        pigment { rgb 1 }
        finish { ambient .4 }
    }

    global_settings {
        radiosity {
            error_bound .1
            always_sample off // default in 3.7
        }
    }

Rendering produces dark splotches at the centers of the pretrace
blocks. Blocks rendered earlier have darker splotches. They also differ
in shape between renders even with +HR (but not with +WT1). Turning
"always_sample" on removes them, as does increasing "count" past 1000.

I wonder if this might be a bug, as 3.6 doesn't produce these artifacts
regardless of "always_sample" and "count".


Post a reply to this message

From: clipka
Subject: Re: Radiosity artifacts at low error_bound in 3.7RC3
Date: 18 May 2011 18:07:29
Message: <4dd44321$1@news.povray.org>
Am 18.05.2011 00:13, schrieb aaz:

> Rendering produces dark splotches at the centers of the pretrace
> blocks. Blocks rendered earlier have darker splotches. They also differ
> in shape between renders even with +HR (but not with +WT1). Turning
> "always_sample" on removes them, as does increasing "count" past 1000.
>
> I wonder if this might be a bug, as 3.6 doesn't produce these artifacts
> regardless of "always_sample" and "count".

Yuck!

The radiosity settings are a poor choice (low error_bound and a low 
number of pretrace passes don't mix well), but it seems there /is/ 
something else fishy. I'll have a look at it. Would you mind filing a 
bug report at http://bugs.povray.org in the meantime?


Post a reply to this message

From: Alain
Subject: Re: Radiosity artifacts at low error_bound in 3.7RC3
Date: 18 May 2011 22:38:13
Message: <4dd48295$1@news.povray.org>

> Consider this scene of a hollow sphere viewed from the inside:
>
>      difference {
>          sphere { 0, 100 }
>          sphere { 0, 99 }
>          pigment { rgb 1 }
>          finish { ambient .4 }
>      }
>
>      global_settings {
>          radiosity {
>              error_bound .1
>              always_sample off // default in 3.7
>          }
>      }
>
> Rendering produces dark splotches at the centers of the pretrace
> blocks. Blocks rendered earlier have darker splotches. They also differ
> in shape between renders even with +HR (but not with +WT1). Turning
> "always_sample" on removes them, as does increasing "count" past 1000.
>
> I wonder if this might be a bug, as 3.6 doesn't produce these artifacts
> regardless of "always_sample" and "count".
>
>
>
Changing pretrace_end to 0.01 make those spots disapear.

count 100 and pretrace_start 0.04 pretrace_end 0.02 also show an uniform 
surface.


Post a reply to this message

From: aaz
Subject: Re: Radiosity artifacts at low error_bound in 3.7RC3
Date: 19 May 2011 07:45:00
Message: <web.4dd5025db1fe0a28dca745f10@news.povray.org>
> Would you mind filing a bug report

http://bugs.povray.org/task/203

> Changing pretrace_end to 0.01 make those spots disapear.

You can still make out their shapes at very high contrast. Actually,
it looks like "always_sample" doesn't remove them completely either.
The render produced by 3.6 is also not entirely uniform, but it doesn't
have this distinct pattern.

Originally I was rendering the shadow of a "no_image" object standing
on a plane and tuned "error_bound" all the way to .01 to get an
accurate edge where the object touched the plane. At this setting these
artifacts appear as random black dots and it's very hard to get rid of
them within a reasonable render time.


Post a reply to this message

Copyright 2003-2021 Persistence of Vision Raytracer Pty. Ltd.