Looks like I'm doing something right with the 3.7 radiosity code...

This scene shows an unlit room, with an almost-but-not-quite closed door,
leading outside into blazing light.

Well, that's what it *should* show. See for yourself what 3.6 and the
3.7.0.beta.29 make of it.

I admit that I didn't use top quality settings for those shots - but compare it
to what *can* be made out of it. Exactly the same scene. Exactly the same lousy
radiosity settings.

Different code though - but not at all optimized for this scene. Just some
changes I thought would be good. I guess they are...


Admittedly, rendering time was significantly higher as well (talking about CPU
power here; wall clock time was actually still lower on an AMD Phenom Quad due
to the multithreading benefit) - but getting 3.6 to the same quality (even with
those remaining "ligt leak" artifacts) would kill. I wonder what my current
version can do with reasonable settings. I used these:

  radiosity {
    pretrace_start 0.08
    pretrace_end   0.004
    count 300
    nearest_count 10
    error_bound 0.15
    recursion_limit 3
    low_error_factor 0.5
    gray_threshold 0
    minimum_reuse 1/1024
    brightness 1.0
    adc_bailout 0.01/2
  }

I guess I'll try with a lower minimum_reuse and error_bound (to reduce
"crosstalk" between wall sides), a higher recursion_limit (there's no direct
path for any light ray into this room, so this scene lives from multiple
bounces) and a higher count (to get everything less splotchy)

Boy, I *AM* excited about this one...

Post a reply to this message

Attachments:
Download 'secret.jpg' (269 KB)

Preview of image 'secret.jpg'

"clipka" <nomail@nomail> wrote:
> This scene shows an unlit room, with an almost-but-not-quite closed door,
> leading outside into blazing light.

Yeah, I had one of those myself.  POV-Ray gives this creepy feeling to scenes
where light comes solely from the outdoor through a semi-opened door... :)

I think the light coming through the corner of the wall and floor can be
suppressed by having actual walls with thickness, like a scaled box.

> Well, that's what it *should* show. See for yourself what 3.6 and the
> 3.7.0.beta.29 make of it.
>
> I admit that I didn't use top quality settings for those shots - but compare it
> to what *can* be made out of it. Exactly the same scene. Exactly the same lousy
> radiosity settings.

The radiosity settings don't seem so lousy.

>   radiosity {
>     pretrace_start 0.08
>     pretrace_end   0.004
>     count 300
>     nearest_count 10
>     error_bound 0.15
>     recursion_limit 3
>     low_error_factor 0.5
>     gray_threshold 0
>     minimum_reuse 1/1024
>     brightness 1.0
>     adc_bailout 0.01/2
>   }

> bounces) and a higher count (to get everything less splotchy)

That, mainly.  But I believe error_bound should actually go up a little:  it
should also be faster to test.  Try with error_bound .8 first.

Post a reply to this message

"clipka" <nomail@nomail> wrote:
>
> Different code though - but not at all optimized for this scene. Just some
> changes I thought would be good. I guess they are...

Hey clipka (and everyone else who might be interested),

There's a quick workaround for 3.7 radiosity. 3.7's radiosity doesn't seem to
like high count values, so to get more detail you can apply a small-scale
global surface normal to all objects in your scene like so:

#default{
 finish{ambient 0}
 normal{bumps .25 scale .001}
}

You can now get away with lower (<300) count settings. It's a hack--I admit
it--but it's worth it for difficult scenes. This works just fine with custom
pretrace_start/end and low error_bound settings as well. You'll need good
antialiasing settings to go along with it:

+a0.1 +am2 +r3

Objects which require their own surface normals can be specified in the usual
manner. Try it! This method works best where sharp point lights are not
involved, since the global surface bumping will try to visually reassert itself
at any given opportunity. I can't remember where this method came from. I hope
this helps somebody!

Sam

Post a reply to this message

"nemesis" <nam### [at] gmailcom> wrote:
> Yeah, I had one of those myself.  POV-Ray gives this creepy feeling to scenes
> where light comes solely from the outdoor through a semi-opened door... :)
>
> I think the light coming through the corner of the wall and floor can be
> suppressed by having actual walls with thickness, like a scaled box.

Actually, the walls I use *do* have a thickness. It doesn't even matter whether
I use meshes or CSG.

Post a reply to this message

"Samuel Benge" <stb### [at] hotmailcom> wrote:
> "clipka" <nomail@nomail> wrote:
> >
> > Different code though - but not at all optimized for this scene. Just some
> > changes I thought would be good. I guess they are...
>
> Hey clipka (and everyone else who might be interested),
>
> There's a quick workaround for 3.7 radiosity. 3.7's radiosity doesn't seem to
> like high count values, so to get more detail you can apply a small-scale
> global surface normal to all objects in your scene like so:
>
> #default{
>  finish{ambient 0}
>  normal{bumps .25 scale .001}
> }
>
> You can now get away with lower (<300) count settings. It's a hack--I admit
> it--but it's worth it for difficult scenes.

BTW, Samwise is one of our resident wizards -- Jaime and Gilles too but they
seem to be a bit off in other planes of existence -- and is well versed in
povray black crafts.  Heed his magic words. :)

Post a reply to this message

"clipka" <nomail@nomail> wrote:
> Looks like I'm doing something right with the 3.7 radiosity code...
>
> This scene shows an unlit room, with an almost-but-not-quite closed door,
> leading outside into blazing light.
..
..
..

Would you mind making the source available?  I have been experimenting with
radiosity under 3.6 recently, and I've run into an odd, non-intuitive way of
getting certain difficult to render scenes to trace correctly.  I also want to
see how the recursive sampling does under MegaPov 1.2.1.

Since the images is still not very accurate and your settings seem to be fairly
high (in comparison to what I use for most scenes), I think this would be a
good test of some alternate methods.


-Reactor

Post a reply to this message

Reactor wrote:
> Would you mind making the source available?

He can't. Redistribution of 3.7 beta code isn't allowed.

Post a reply to this message

"Samuel Benge" <stb### [at] hotmailcom> wrote:
> Hey clipka (and everyone else who might be interested),
>
> There's a quick workaround for 3.7 radiosity.

I'm actually not interested in quick workarounds, but getting the thing to
ultimately work fine without them :)

> 3.7's radiosity doesn't seem to like high count values,

That's very interesting for me to hear. Can you describe what's wrong with it,
and maybe post a sample scene?

> so to get more detail you can apply a small-scale
> global surface normal to all objects in your scene like so:
>
> #default{
>  finish{ambient 0}
>  normal{bumps .25 scale .001}
> }
>
> You can now get away with lower (<300) count settings. It's a hack--I admit
> it--but it's worth it for difficult scenes. This works just fine with custom
> pretrace_start/end and low error_bound settings as well. You'll need good
> antialiasing settings to go along with it:

Hm - sounds to me like you're actually using the normals to force more samples
to be taken. Does this any good to rendering time?

Post a reply to this message

Nicolas Alvarez <nic### [at] gmailcom> wrote:
> Reactor wrote:
> > Would you mind making the source available?
>
> He can't. Redistribution of 3.7 beta code isn't allowed.

Sorry, I meant the scene source code, not the binary.  My tests will be on the
scene with another method of adjusting radiosity under version 3.6 (official)
and MegaPov 1.2.1

Reactor

Post a reply to this message

"Reactor" <rea### [at] hotmailcom> wrote:
> Would you mind making the source available?  I have been experimenting with
> radiosity under 3.6 recently, and I've run into an odd, non-intuitive way of
> getting certain difficult to render scenes to trace correctly.  I also want to
> see how the recursive sampling does under MegaPov 1.2.1.

Whoops - somehow this post of yours slipped past me...

Tell me about those experiments - it may help me identify some more quirks in
the radiosity code.


> Since the images is still not very accurate and your settings seem to be fairly
> high (in comparison to what I use for most scenes), I think this would be a
> good test of some alternate methods.

Sure, give it a try. But promise to tell me about your results ;)

As it is now, it needs a texture named "Parquetry_plain.png", but I guess that
doesn't really matter, so you may want to replace it with a uniform pigment.
The original texture I use is too large to post it.

Post a reply to this message

Attachments:
Download 'the_secret.zip' (291 KB)