I'm getting some lighting artifacts when using radiosity.  There is one
light in the scene, an area light, in a light group with some spheres.  On
the blue sphere (Neptune, actully) I'm seeing some bright patches (upper
left picture).  If I change the color of the area light, it doesn't affect
the artifact (upper right).  The artifact isn't present in the pretrace
step (lower left). Also, the artifact changes shape/size if I render a
different portion of the image or change the final render size keeping all
other things constant (lower right, also showing a pretrace step).

Any ideas? Thanks in advance!

Here is the radiosity code:

global_settings {
  radiosity {
    #if (image_width>1500)
      pretrace_start 0.01       // start pretrace at this size
      pretrace_end   0.001
    #else
      pretrace_start 0.08
      pretrace_end   0.02
    #end                          // end pretrace at this size
    count 60                      // higher -> higher quality (1..1600) [35]
    nearest_count 8               // higher -> higher quality (1..10) [5]
    error_bound 1.8               // higher -> smoother, less accurate [1.8]
    recursion_limit 3             // how much interreflections are
calculated (1..5+) [3]
    low_error_factor .5           // reduce error_bound during last pretrace
step
    gray_threshold 0.0            // increase for weakening colors (0..1)
[0]
    minimum_reuse 0.015           // reuse of old radiosity samples [0.015]
    brightness 0.3                // brightness of radiosity effects (0..1)
[1]

    adc_bailout 0.01/2
    //normal on                   // take surface normals into account [off]
    //media on                    // take media into account [off]
    //save_file "file_name"       // save radiosity data
    //load_file "file_name"       // load saved radiosity data
    //always_sample off           // turn sampling in final trace off [on]
    //max_sample 1.0              // maximum brightness of samples

  }
  assumed_gamma 1.0
}

Here is the light group:

light_group {
    #if (arealight)
    light_source {
      0*x
      color rgb <255,255,230>/255
      area_light
      <0.5*Rsun, 0, 0> <0, 0, 0.5*Rsun>
      10, 10
      adaptive 2
      circular
      orient
      //jitter
      translate 1*<8*Rj,1.1*Rsun,-Rsun>
    }
    #else
        light_source {
        <0, 0, 0>            // light's position (translated below)
        color rgb <1, 1, 1>  // light's color
        translate <-30, 30, 30>
      }
    #end

    #if (radiosity_on)
    plane {
      y, 0
       texture { pigment { color rgb 1.8 } finish { ambient 0.00 diffuse 0.4
specular 0 reflection 0}}
    }
    #else
    plane {
      y, 0
      texture { pigment { color rgb 1 } finish { ambient 0.10 diffuse 0.6
specular 0 reflection 0}}
    }
    #end

    object { earth translate -Rn*x translate <-2*Re,Re,2*Rj> }
    object { moon translate -Rn*x translate <-2.3*Re,Re_moon,2*Rj+Re>}
    object { venus translate -Rn*x translate <-4.5*Re,Rv,2*Rj>}
    object { mars translate -Rn*x translate <-6.5*Re,Rmars,2*Rj>}
    object { mercury translate -Rn*x translate <-8*Re,Rmerc,2*Rj>}
    object { pluto translate -Rn*x translate <-9*Re,Rp,2*Rj>}

    object { jupiter rotate 180*y translate <1.90*Rs,Rj2, -Rj>}
    object { saturn rotate -13*x rotate -65*y translate <-1.7*Rs,Rs2, -Rj>}
    object { uranus translate -Rn*x translate <1.5*Rs,Ru2, 2*Rj>}
    object { neptune translate -Rn*x translate <0.5*Rs,Rn2, 2*Rj>}
}

Post a reply to this message

Attachments:
Download 'artifacts2.jpg' (41 KB)

Preview of image 'artifacts2.jpg'

Mike Sobers wrote:
> I'm getting some lighting artifacts when using radiosity.  There is one
> light in the scene, an area light, in a light group with some spheres.  On
> the blue sphere (Neptune, actully) I'm seeing some bright patches (upper
> left picture).  If I change the color of the area light, it doesn't affect
> the artifact (upper right).  The artifact isn't present in the pretrace
> step (lower left). Also, the artifact changes shape/size if I render a
> different portion of the image or change the final render size keeping all
> other things constant (lower right, also showing a pretrace step).
> 
> Any ideas? Thanks in advance!

>     count 60                      // higher -> higher quality (1..1600) [35]

Increase this.  For low end (test) renders, I use 100.  For higher end 
(ie final), I use at least 300 (of course, that's just me).

...Chambers

Post a reply to this message

Ben Chambers <ben### [at] pacificwebguycom> wrote:
> Mike Sobers wrote:

> > Any ideas? Thanks in advance!
>
> >     count 60                      // higher -> higher quality (1..1600) [35]
>
> Increase this.  For low end (test) renders, I use 100.  For higher end
> (ie final), I use at least 300 (of course, that's just me).
>
> ...Chambers

Thanks for the tip!  I bumped count up to 1000, and it only changed the
shape/size of the artifacts.  However, you got me thinking, and by dropping
error_bound from 1.8 to about 0.2 the problem is resolved.  Funny thing
though, it seems to be dependent on the pretrace_end size and overall image
size.  Anyone have any suggestions as to what causes this in the first
place?

-Mike

PS - planets are to scale, Earth is the largest of the small ones

Post a reply to this message

Attachments:
Download '9planets_small.png' (410 KB)

Preview of image '9planets_small.png'

Some balls look electrolized and with magnetic tiny metal splits on them...

Sven

Post a reply to this message