Hey folks...

I was trying to take the sample hdr_mapping.pov scene and use it as a 
base for my own HDR experiments.

So, I set up a quick scene with just a camera looking at a sphere using 
the generated map from hdr_environment. Looked good, nice smooth shading 
and lighting.

Then I swapped the pigment statements in the HDR sphere to the mapping 
type (7) that is supposed to work with Paul Devebecs spherical HDR maps. 
All the maps of his that I have tried have had very strange effects on 
the surfaces, almost like "light noise". Using the HDR file from the 
Rendering with Natural Light demo, you can see that the sphere is very 
spotty. Please see the attached picture for an example. I've also 
included a pic done with the "grace" probe. (yuck!)

Considering that all that is different between these scenes is the type 
of HDR, I can't figure out what is going on.

Has anybody else been able to succesfully get these types of maps to 
work correctly?

Thanks!

Arlo


// CODE:

#include "colors.inc"

#version unofficial MegaPov 1.10;

// radiosity (global illumination) settings
global_settings {
   max_trace_level 15
   radiosity {
     pretrace_start 0.08
     pretrace_end   0.01
     count 800
     //count 250
     nearest_count 8
     //nearest_count 5
     error_bound 0.07
     //error_bound 0.2
     recursion_limit 2
     brightness 1

     normal on
     randomize on
   }
}

camera
{
    location <0, 2, -3>
    look_at <0, 1, 0>
    angle 90
}

sphere
{
    <0, 0, 0>, 1
    translate <0, 1, 0>
    pigment
    {
       color White
    }
    finish
    {
       ambient 0
       diffuse 0.7
    }
}

box
{
    <-2, -0.1, -2>, <2, 0, 2>
    pigment
    {
       color White
    }
    finish
    {
       ambient 0
       diffuse 0.7
    }
}

sphere {
   0,200
   pigment { image_map { hdr "c:\grace_probe.hdr" once interpolate 2 
map_type 7 } }
   //pigment { image_map { hdr "c:\hdr_env.hdr" once interpolate 2 
map_type 1 } }
   finish { ambient 0.8 diffuse 0 }
   scale <-1,1,1>
   hollow
   rotate 120*y
}

Post a reply to this message

Attachments:
Download 'isosurface.jpg' (41 KB) Download 'isosurface2.jpg' (39 KB)

Preview of image 'isosurface.jpg'

Preview of image 'isosurface2.jpg'

<cries>

Post a reply to this message

Hi,

> All the maps of his that I have tried have had very strange
> effects on the surfaces, almost like "light noise".

I tried your scene, with some other maps, too... and I got the
same splotchy look. I fiddled with the radiosity parameters for
a while, and I came up with the following:

Use max_sample (somewhere in the range 5..10), it makes the whole
scene a bit darker (which would be ok for the grace probe, it's a
rather dark place after all)... but you can adjust that with the
brightness parameter if you want. The result looks a bit different
and I don't know if it's "realistic" any more, but at least it looks
relatively smooth and it didn't look too bad when I tested it.

  radiosity {
    max_sample 5
    count 1600 error_bound 0.05
    pretrace_start 8/100 pretrace_end 1/100
    recursion_limit 2
    brightness 1.5
  }

Cheers,
Artur.

Post a reply to this message

A.Brinkmann wrote:

> I tried your scene, with some other maps, too... and I got the
> same splotchy look. I fiddled with the radiosity parameters for
> a while, and I came up with the following:

Thanks, I'll try it and play around. :) Considering that map type 7 is 
stated as being specifically for Paul Debevec spherical light probes, I 
wonder what exactly is going on in the renderer....

Thanks again,

Arlo

Post a reply to this message

> Considering that map type 7 is stated as being specifically
> for Paul Debevec spherical light probes, I wonder what exactly
> is going on in the renderer....

I don't think it's a problem of this map type... I used Debevec's
HDR Shop program to convert the map to a spherical map, and mapped
it with map_type 1... same problem. I think the problem is that this
map has some very small, very bright spots. As far as I understand
the concept of radiosity, it's not very well suited to simulate point
light sources. After applying a gaussian blur to the map, the effect
disappeared almost completely.

Post a reply to this message