After Nathan Kopp's great "box with no lights" discovery, I tried to
adapt it to an old scene of mine. Note that it's just an experiment,
I'll turn it into a real scene with good textures and all only if I can
get it to work.
As you can see, overall lighting could be fine, but the result is
horribly patchy on the back wall in spite of "high" radiosity settings.
Any clue on how to fix that ? Lowering the distance_maximum make it
slooow. Having more samples doesn't change anything. I'll post the scene
if somebody wants to have a try.
One funny (logical ?) thing : I had to remove the transparent window
panes for radiosity to work. I guess it didn't like the ambient 0 value
of the glass texture.

Gilles Tran

Radiosity settings :
    count 800
    error_bound 0.2
    gray_threshold 0.4
    distance_maximum 0.4
    low_error_factor 0.6
    nearest_count 9
    minimum_reuse 0.006
    brightness 3.3
    recursion_limit 2

The "light" is background{color White*10}
The box size is approximately 20 pov units wide.
Rendering time at 800x600, no aa, was 7 hours on a Pentium Pro 200, NT.

Post a reply to this message

Attachments:
Download 'heliot99.jpg' (39 KB)

Preview of image 'heliot99.jpg'

Gilles Tran wrote:

> After Nathan Kopp's great "box with no lights" discovery, I tried to
> adapt it to an old scene of mine. Note that it's just an experiment,
> I'll turn it into a real scene with good textures and all only if I can
> get it to work.

Very nice!

I do rather like the blockiness of the back wal.. is this intentional, or an
artifact of radiosity?

The mace ball also looks really good here, maybe even better than the
original.. more depth, somehow.

Simon
http://home.istar.ca/~sdevet

Post a reply to this message

I was going to say the same thing, the wall looks "built" of something
this way.
I have yet to see a really smooth radiosity. You can only try to keep
the distance_maximum pretty small (much less than the 1/3 rule), the
count high (250), and minimum_reuse within a good range (0.01), far as I
know.
Another thing is to have a normal applied to everything, at a small
scale/amount. Tends to even out the radiosity it seems, even if its
granular still.


Simon de Vet wrote:
> 
> Very nice!
> 
> I do rather like the blockiness of the back wal.. is this intentional, or an
> artifact of radiosity?
> 
> The mace ball also looks really good here, maybe even better than the
> original.. more depth, somehow.
> 
> Simon
> http://home.istar.ca/~sdevet

-- 
 omniVERSE: beyond the universe
  http://members.aol.com/inversez/POVring.htm
 mailto:inv### [at] aolcom?PoV

Post a reply to this message

One thing you can try is raising the low_error_factor to 0.8 or .9.  For
some reason the smoothing doesn't seem to work as well when it's lower.
Either that or the samples aren't evenly dispersed at the lower setting.
To be honest I'm not sure why it is, but the blocky patterns are
definitely an artifact of the mosaic preview.

Also try raising the distance_maximum and reduce minimum_reuse. 

-Mike


> Radiosity settings :
>     count 800
>     error_bound 0.2
>     gray_threshold 0.4
>     distance_maximum 0.4
>     low_error_factor 0.6
>     nearest_count 9
>     minimum_reuse 0.006
>     brightness 3.3
>     recursion_limit 2
> 
> The "light" is background{color White*10}
> The box size is approximately 20 pov units wide.
> Rendering time at 800x600, no aa, was 7 hours on a Pentium Pro 200, NT.
> 
>   ------------------------------------------------------------------------
>  [Image]

Post a reply to this message

Mosaic preview at fault for the blocks? Oh boy, that's worse than I
would have ever supposed it to be. But great to hear of that anyhow.

Mike wrote:
> 
> One thing you can try is raising the low_error_factor to 0.8 or .9.  For
> some reason the smoothing doesn't seem to work as well when it's lower.
> Either that or the samples aren't evenly dispersed at the lower setting.
> To be honest I'm not sure why it is, but the blocky patterns are
> definitely an artifact of the mosaic preview.
> 
> Also try raising the distance_maximum and reduce minimum_reuse.
> 
> -Mike
> 
> > Radiosity settings :
> >     count 800
> >     error_bound 0.2
> >     gray_threshold 0.4
> >     distance_maximum 0.4
> >     low_error_factor 0.6
> >     nearest_count 9
> >     minimum_reuse 0.006
> >     brightness 3.3
> >     recursion_limit 2
> >
> > The "light" is background{color White*10}
> > The box size is approximately 20 pov units wide.
> > Rendering time at 800x600, no aa, was 7 hours on a Pentium Pro 200, NT.
> >
> >   ------------------------------------------------------------------------
> >  [Image]

-- 
 omniVERSE: beyond the universe
  http://members.aol.com/inversez/POVring.htm
 mailto:inv### [at] aolcom?PoV

Post a reply to this message

Mike wrote:

> One thing you can try is raising the low_error_factor to 0.8 or .9.  For
> some reason the smoothing doesn't seem to work as well when it's lower.
> Either that or the samples aren't evenly dispersed at the lower setting.
> To be honest I'm not sure why it is, but the blocky patterns are
> definitely an artifact of the mosaic preview.

I'm willing to argue this point.

I think the grainyness looks great in this particular pic..

Simon
http://home.istar.ca/~sdevet

Post a reply to this message

I thought that when you had a mosaic preview, it didn't save the image until
the final pass.  If that's the case how can you have artifacts from the
preview.  Also, I've only tried radiosity once, but it seems to me, if I
remember correctly, that by using the "radiosity=on" statement (or whatever
it was, i may be forgetting the syntax) instead of the +QR I didn't get a
mosaic preview.  If I'm mistaken, I apologize...

Ciao!
Dave.

Post a reply to this message