 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I'm trying to get radiosity working right (as per post in .General) and
have included my radiosity settings below.
However the real reason for this post is the strange lighting effect
that is showing around the architraves in the room shown below.
At first I wondered if there was a tiny gap between the top of the
architrave and the wall, however there's not. I've made sure the overlap.
What else could be causing this weirdness?
I can provide the full source upon request, but there's a lot as the
whole house has been built and is getting furnished :)
radiosity {
pretrace_start 0.08
pretrace_end 0.04
count 85
nearest_count 5
error_bound 1.8
recursion_limit 12
low_error_factor 0.5
gray_threshold 0.0
minimum_reuse 0.015
brightness 1.2
adc_bailout 0.01/2
}
Post a reply to this message
Attachments:
Download 'house.jpg' (136 KB)
Preview of image 'house.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rick Measham wrote:
> radiosity {
> pretrace_start 0.08
> pretrace_end 0.04
> count 85
>
> nearest_count 5
> error_bound 1.8
> recursion_limit 12
>
> low_error_factor 0.5
> gray_threshold 0.0
> minimum_reuse 0.015
> brightness 1.2
>
> adc_bailout 0.01/2
> }
I didn't see your post in .general, but...
Holy crap!
recursion_limit 1 is enough for scenes with a lot of ambient light
(skylight, lots of artificial lighting, etc). 2 will do for most scenes
with less light and lots of shadows, and you should only go to 3 if
you're trying to bounce light around a few corners or you're doing a
fancy final render of a scene that normally gets 2. Beyond 3, the
differences are usually imperceptible.
The more you increase the recursion, generally speaking, the more
inaccurate and artifacted your radiosity gets. I don't know that this is
the problem, but try cutting that 12 down to 2.
Also, you really ought to decrease that error_bound. .3 or less will
resolve small shadow details, but tends to be pretty slow and needs a
higher count for smooth results; .5 - 1 will do for an approximation of
realistic lighting.
-Xplo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
also, it's possible that scene ambience + object ambience maybe adding
unwanted energy to the scene (especially at the corners of the room)
benp
Xplo Eristotle wrote:
> Rick Measham wrote:
>
>> radiosity {
>> pretrace_start 0.08
>> pretrace_end 0.04
>> count 85
>>
>> nearest_count 5
>> error_bound 1.8
>> recursion_limit 12
>>
>> low_error_factor 0.5
>> gray_threshold 0.0
>> minimum_reuse 0.015
>> brightness 1.2
>>
>> adc_bailout 0.01/2
>> }
>
>
> I didn't see your post in .general, but...
>
> Holy crap!
>
> recursion_limit 1 is enough for scenes with a lot of ambient light
> (skylight, lots of artificial lighting, etc). 2 will do for most scenes
> with less light and lots of shadows, and you should only go to 3 if
> you're trying to bounce light around a few corners or you're doing a
> fancy final render of a scene that normally gets 2. Beyond 3, the
> differences are usually imperceptible.
>
> The more you increase the recursion, generally speaking, the more
> inaccurate and artifacted your radiosity gets. I don't know that this is
> the problem, but try cutting that 12 down to 2.
>
> Also, you really ought to decrease that error_bound. .3 or less will
> resolve small shadow details, but tends to be pretty slow and needs a
> higher count for smooth results; .5 - 1 will do for an approximation of
> realistic lighting.
>
> -Xplo
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Xplo Eristotle wrote:
> Holy crap!
Excellent! I've always wanted to extract a 'Holy Crap' from someone :)
OK, I've brought those values back to your recommendation, but then
things started getting really really weird. So I decreased the amount of
light flying around (my sun was at White * 40, and the skysphere was
(some blue)*5).
Now the interior of the room is almost pitch black.
I've also tried adding the tiniest amount of reflection and fairly heavy
diffusion to everything in the hope of getting it looking better.
I'm going to look at light_sys once I get a chance, but I'm really
interested in why this isn't working.
(You'll notice in the corners, there's still a little brightness, and
there's still strange patterns there ..)
Cheers!
Rick Measham
radiosity {
pretrace_start 0.08
pretrace_end 0.04
count 85
nearest_count 5
error_bound 0.5
recursion_limit 2
low_error_factor 0.5
gray_threshold 0.0
minimum_reuse 0.015
brightness 1.2
adc_bailout 0.01/2
}
Post a reply to this message
Attachments:
Download 'house.jpg' (114 KB)
Preview of image 'house.jpg'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
bp wrote:
> also, it's possible that scene ambience + object ambience maybe adding
> unwanted energy to the scene (especially at the corners of the room)
> benp
Thanks ben, I've removed any ambient light from everything so it can't
be that ..
Cheers!
Rick
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
I guess there are a couple of problems.
1: The room you're looking out from is in shadow from the actual
lightsource, so it is lit ONLY by radiosity.
2: The door looks out to the opposite wall of the house. Add that to 3...
3: A count of 85 tends to "overlook" things.
If you'll look at the far room, it gets lit properly. So the reason why the
current room is so dark is because radiosity sampling doesn't find anything
bright! Try moving the lightsource so that some of it enters the room and
you'll notice the difference.
What I'm suspecting is that with the current setup, it will be quite
difficult to get some lighting into the room. Try raising count to something
like 600 or so for starters to see if that changes anything. Maybe try
raising recursion_limit to 3 or 4, though I don't expect it to have much
influence on the scene as it is.
You might even want to try and use the radiosity-settings from "Windows
Room", which I've posted in povray.binaries.scene-files. Or at least have a
look at it, as I've used settings which I've come to like and rely on for a
long time now.
Regards,
Tim
--
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
431d3d4c$1@news.povray.org...
> bp wrote:
>> also, it's possible that scene ambience + object ambience maybe adding
>> unwanted energy to the scene (especially at the corners of the room)
>> benp
>
> Thanks ben, I've removed any ambient light from everything so it can't be
> that ..
What Tim said. This setup is just too hard to handle for radiosity because
there's not enough light. For it to work, it would require extremely
high-quality, impractical parameters to get both good results without
artifacts.
If you don't want to change your basic setup, one possibility would be to
add to the scene, hidden from the view, a bright object like a flat box or a
square mesh or disc with ambient 1 or more (playing the part of a soft box
in photography) to generate more light inside the room, the risk being that
this object could also illuminate other parts of the scene.
G.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Gilles Tran wrote:
> If you don't want to change your basic setup, one possibility would be to
> add to the scene, hidden from the view, a bright object like a flat box or a
> square mesh or disc with ambient 1 or more (playing the part of a soft box
> in photography) to generate more light inside the room, the risk being that
> this object could also illuminate other parts of the scene.
>
> G.
Another idea is a 2-pass method:
Render 1 pass with a large area of high ambient filling the window (just
outside the window (save the radiosity data with save_file). Then render
the second pass as-is, but have the radiosity block replaced with just a
load_file and always_sample off statements.
--
~Mike
Things! Billions of them!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Rick Measham wrote:
> Xplo Eristotle wrote:
>
>> Holy crap!
>
>
> Excellent! I've always wanted to extract a 'Holy Crap' from someone :)
>
> OK, I've brought those values back to your recommendation, but then
> things started getting really really weird. So I decreased the amount of
> light flying around (my sun was at White * 40, and the skysphere was
> (some blue)*5).
>
> Now the interior of the room is almost pitch black.
>
> I've also tried adding the tiniest amount of reflection and fairly heavy
> diffusion to everything in the hope of getting it looking better.
>
> I'm going to look at light_sys once I get a chance, but I'm really
> interested in why this isn't working.
>
> (You'll notice in the corners, there's still a little brightness, and
> there's still strange patterns there ..)
>
> Cheers!
> Rick Measham
>
> radiosity {
> pretrace_start 0.08
> pretrace_end 0.04
> count 85
>
> nearest_count 5
> error_bound 0.5
> recursion_limit 2
>
> low_error_factor 0.5
> gray_threshold 0.0
> minimum_reuse 0.015
> brightness 1.2
>
> adc_bailout 0.01/2
> }
>
>
> ------------------------------------------------------------------------
>
Also, add to your global settings block assumed_gamma 1.0 (some will
disagree) so long as your monitor gamma is set right (defaults to 2.2, I
think, which is typical for a PC) it should get you a better tonal
range. Higher counts are a must, and can add greatly to the scene (at
the cost of render time)
hth.
--
~Mike
Things! Billions of them!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
431da444@news.povray.org...
>
> Another idea is a 2-pass method:
>
> Render 1 pass with a large area of high ambient filling the window (just
> outside the window (save the radiosity data with save_file). Then render
> the second pass as-is, but have the radiosity block replaced with just a
> load_file and always_sample off statements.
Another idea, but one that doesn't work in Povray AFAIK but could be
implemented in a patch (I don't think that it's in Megapov) would be to have
no_image objects able to generate radiosity. I've used that in another
software for this exact type of effect (soft radiosity light coming from a
window that doesn't receive enough sky light) and it's really efficient,
particularly as the light-generating object can be kept close to the target
(which seems to reduce artifacts) and still be invisible.
G.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
