 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> > and before, with megapov 0.5, I got dark circles even with a high count
> > !
> >
>
> I didn't know radiosity changed from 0.5 to 0.6, anyone who knows ?
>
> Christoph
>
I don't know if it changed but it's just to say I rendered it using two different
versions. As I downloaded the 0.6 just this morning and I didn't have time to check
what changed or not. I think it came from my parameters I modified just before
rendering the scene...
David
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
gemelli david wrote:
>
> hum... strange title but this scene is a WIP. I'm using it to test some
> radiosity parameters and technics, and it will be a comprative scene in
> POV and LightScape when the 3D Studio MAX model will be done.
>
> Story:
>
> I saw yesterday a photo in one of my brother's ebenist books of a studio
> like this. I decided to do it with POV.
> today, at work, while my computer was unused, I launched a render. A
> graphist wanted to test the same scene in LightScape (we are using it
> since a few days). So, he will do the model in MAX, and he will import
> it in LightScape, I will convert it to a POV scene and we will do
> tests...
>
> My question:
> I used this parameters in megapov 0.6.
> pretrace_start 0.04
> pretrace_end 0.01
> count 300
> recursion_limit 2
> nearest_count 3
> error_bound 0.3
> I read the "Christoph's Page", I tried various values for radiosity but
> I still have this dark patches in the corner of the window.
>
> and before, with megapov 0.5, I got dark circles even with a high count
> !
I need to write a tutorial on low-error radiosity one of these days so I
can stop typing this stuff out... >_<
One of the problems with using an error_bound of around .3 is that the
radiosity tends to be blotchy; the effect is almost as if you were
shining a light through a screen and getting a lot of holes, or
something like the multiple-shadow effect of area_lights, only more so.
I've found that increasing error_bound will make the blotches larger and
softer, and at the default error_bound of .6, they blend together well
enough as to all but disappear.. but of course, with your error_bound
this high, you won't get the neat shadows.
If you want to use these settings more or less as they are, it's crucial
that you set the pretrace_end low enough that you're getting 2-pixel
radiosity blocks (try something around .005). This will help smooth the
blotches. Also, your nearest_count is way too low; with an error of .3,
you should leave it at default. (You could even increase it to the max
of 20, but I don't think this will help.) Set low_error_factor to 1,
which should make the pretrace a little faster without hurting quality.
If you still get unacceptable dark artifacts or the smoothing still
isn't good enough, and you have a crapload of RAM, you can use a trick
to make the artifacts smaller than a pixel (which gives extremely smooth
results, if you follow me). Set your error_bound to .03, nearest_count
to 1, low_error_factor to 1, increase your count to, hmm, 500 or so for
this scene I think, and adc_bailout to 1. Pretrace_start and _end can be
set to anything you find convenient; with these settings, they have no
useful purpose other than giving you a mosaic preview and waste a lot of
time that could be used for rendering.
-Xplo
Post a reply to this message
|
|
| |
| |
|
|
From: Nathan Kopp
Subject: Re: radiosity inside ! (WIP, Test, Comparative, 15k)
Date: 6 Nov 2000 22:50:55
Message: <3a077c1f@news.povray.org>
|
|
|
| |
| |
|
|
"Xplo Eristotle" <inq### [at] unforgettable com> wrote...
> I need to write a tutorial on low-error radiosity one of these days so I
> can stop typing this stuff out... >_<
I'd love to see the results of your suggestions. It would be very
interesting to see both the low-quality smooth rendering and the
high-quality-extra-slow version.
-Nathan
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>
> I'd love to see the results of your suggestions. It would be very
> interesting to see both the low-quality smooth rendering and the
> high-quality-extra-slow version.
>
> -Nathan
Let's go !
Params:
pretrace_start 0.04
pretrace_end 0.01
recursion_limit 2
error_bound .03
nearest_count 1
low_error_factor 1
count 500
adc_bailout 1
Render in progress !
David
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nathan Kopp wrote:
>
> "Xplo Eristotle" <inq### [at] unforgettablecom> wrote...
> > I need to write a tutorial on low-error radiosity one of these days so I
> > can stop typing this stuff out... >_<
>
> I'd love to see the results of your suggestions. It would be very
> interesting to see both the low-quality smooth rendering and the
> high-quality-extra-slow version.
I've made a few posts here on the subject; particularly, you might want
to look for "YMFWR" and "Son of YMFWR".
-Xplo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Hi,
gemelli david wrote:
> Params:
> pretrace_start 0.04
> pretrace_end 0.01
> recursion_limit 2
>
> error_bound .03
> nearest_count 1
> low_error_factor 1
> count 500
> adc_bailout 1
Use a recursion_limit of 3 or 4...the darker edges must be brighter and
it will look much better as at 2 :-)
Nathan:
it is right, that the maximum of recursion_limit is 6?? I've rendered a
test-scene with several values for
recursion_limit...after 6 all rendering times (also at 100 and 150) were
the same as at 6.
Paul
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Paul Blaszczyk wrote:
>
> Hi,
>
> gemelli david wrote:
> > Params:
> > pretrace_start 0.04
> > pretrace_end 0.01
> > recursion_limit 2
> >
> > error_bound .03
> > nearest_count 1
> > low_error_factor 1
> > count 500
> > adc_bailout 1
>
> Use a recursion_limit of 3 or 4...the darker edges must be brighter and
> it will look much better as at 2 :-)
I'm not so sure about that.. in a number of my tests, the differences
between 2 and higher levels were noticible, but not especially significant.
> Nathan:
> it is right, that the maximum of recursion_limit is 6?? I've rendered a
> test-scene with several values for
> recursion_limit...after 6 all rendering times (also at 100 and 150) were
> the same as at 6.
I would think that the amount of "light" given off from each bounce
would get smaller and smaller.. perhaps, at that high of a recursion
level, the extra bounces are contributing so little to the scene that
they don't show up at all.
-Xplo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <3A0### [at] alpharayde>, Paul Blaszczyk
<3d### [at] alpharayde> wrote:
> it is right, that the maximum of recursion_limit is 6?? I've rendered
> a test-scene with several values for recursion_limit...after 6 all
> rendering times (also at 100 and 150) were the same as at 6.
I think that sampling stops when it gets to the point that additional
samples won't make much difference. It could be that with your settings,
recursion levels beyond 6 would make such a small contribution that they
are dropped. Something like the ADC settings.
--
Christopher James Huff
Personal: chr### [at] maccom, http://homepage.mac.com/chrishuff/
TAG: chr### [at] tagpovrayorg, http://tag.povray.org/
<><
Post a reply to this message
|
|
| |
| |
|
|
From: Nathan Kopp
Subject: Re: radiosity inside ! (WIP, Test, Comparative, 15k)
Date: 12 Nov 2000 01:55:21
Message: <3a0e3ed9$1@news.povray.org>
|
|
|
| |
| |
|
|
"Chris Huff" <chr### [at] maccom> wrote...
> Paul Blaszczyk <3d### [at] alpharayde> wrote:
>
> > it is right, that the maximum of recursion_limit is 6?? I've rendered
> > a test-scene with several values for recursion_limit...after 6 all
> > rendering times (also at 100 and 150) were the same as at 6.
>
> I think that sampling stops when it gets to the point that additional
> samples won't make much difference. It could be that with your settings,
> recursion levels beyond 6 would make such a small contribution that they
> are dropped. Something like the ADC settings.
Yes. There is an ADC bailout setting specifically for radiosity (separate
from the usual ADC bailout setting). See the MegaPov docs for more info.
-Nathan
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nathan Kopp wrote:
>
> "Chris Huff" <chr### [at] maccom> wrote...
> > Paul Blaszczyk <3d### [at] alpharayde> wrote:
> >
> > > it is right, that the maximum of recursion_limit is 6?? I've rendered
> > > a test-scene with several values for recursion_limit...after 6 all
> > > rendering times (also at 100 and 150) were the same as at 6.
> >
> > I think that sampling stops when it gets to the point that additional
> > samples won't make much difference. It could be that with your settings,
> > recursion levels beyond 6 would make such a small contribution that they
> > are dropped. Something like the ADC settings.
>
> Yes. There is an ADC bailout setting specifically for radiosity (separate
> from the usual ADC bailout setting). See the MegaPov docs for more info.
It doesn't seem to work like the docs suggest, though. I've set it to 1
and had no real difference on a scene except to smooth out the radiosity
a little. In fact, based on my limited experience, I'd suggest leaving
it set at 1 all the time.
-Xplo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
