POV-Ray : Newsgroups : povray.binaries.images : Radiosity - interior scene test (90 kbu) Server Time
20 Aug 2024 00:08:21 EDT (-0400)
  Radiosity - interior scene test (90 kbu) (Message 1 to 10 of 61)  
Goto Latest 10 Messages Next 10 Messages >>>
From: Gilles Tran
Subject: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 15:31:39
Message: <39BE8414.E6634119@inapg.inra.fr>
Disclaimer : this scene has no artistical merit whatsoever. It's ugly,
but it's just a test for a radiosity setup that could be acceptable for
a "real" interior scene.
It uses some of the recently gained radiosity knowledge posted in this
group (Arnold settings and outdoors lighting particularly). There's only
one light source, coming from the window. Everything else is indirect
lighting. Balancing the light was difficult. I ended up raising the
brightness to 2 and lowering the Kari Kivisalo light intensity.
Assumed_gamma is 1. It took 4 hours to render on a PIII 733 at 800*600
aa 0.3.

Some of the radiosity effects are really OK. Vicky's face is nicely lit,
and there's even a backlit effect on her right arm. The shadow of the
chair on the left seems good too. On the bad side, there are quite a
number of radiosity artifacts visible on the right wall and on Vicky's
face and right foot (note : raising the recursion level to 2 had
appalling effects on Vicky so I left it to 1). Also, I had to add
reflection to the chairs to make them look better, which lets me think
that one last thing to improve in Megapov radiosity is it's ability to
react like a regular light with some of  the texture properties
(highlights). I didn't try to "cheat" with light groups or area lights
to fix that, though I'll do it probably on a real pic.

Credits : Victoria is the hi-detail Poser girl from Zygote (dress by
Catherine Todd). The chairs and the library are available on my website
(when it works). The wallpaper is just terrible.
The whole scene (but big Vicky who weights around 15 Mb) is posted on
p.b.s-f.

G.


Post a reply to this message


Attachments:
Download 'testroom7.jpg' (90 KB)

Preview of image 'testroom7.jpg'
testroom7.jpg


 

From: ryan constantine
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 15:42:14
Message: <39BE871A.B3808900@yahoo.com>
i'm not worthy

Gilles Tran wrote:
> 
> Disclaimer : this scene has no artistical merit whatsoever. It's ugly,
> but it's just a test for a radiosity setup that could be acceptable for
> a "real" interior scene.
> It uses some of the recently gained radiosity knowledge posted in this
> group (Arnold settings and outdoors lighting particularly). There's only
> one light source, coming from the window. Everything else is indirect
> lighting. Balancing the light was difficult. I ended up raising the
> brightness to 2 and lowering the Kari Kivisalo light intensity.
> Assumed_gamma is 1. It took 4 hours to render on a PIII 733 at 800*600
> aa 0.3.
> 
> Some of the radiosity effects are really OK. Vicky's face is nicely lit,
> and there's even a backlit effect on her right arm. The shadow of the
> chair on the left seems good too. On the bad side, there are quite a
> number of radiosity artifacts visible on the right wall and on Vicky's
> face and right foot (note : raising the recursion level to 2 had
> appalling effects on Vicky so I left it to 1). Also, I had to add
> reflection to the chairs to make them look better, which lets me think
> that one last thing to improve in Megapov radiosity is it's ability to
> react like a regular light with some of  the texture properties
> (highlights). I didn't try to "cheat" with light groups or area lights
> to fix that, though I'll do it probably on a real pic.
> 
> Credits : Victoria is the hi-detail Poser girl from Zygote (dress by
> Catherine Todd). The chairs and the library are available on my website
> (when it works). The wallpaper is just terrible.
> The whole scene (but big Vicky who weights around 15 Mb) is posted on
> p.b.s-f.
> 
> G.
> 
>   ------------------------------------------------------------------------
>  [Image]


Post a reply to this message

From: Gilles Tran
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 15:53:32
Message: <39BE8937.6485AAA8@inapg.inra.fr>
ryan constantine wrote:

> i'm not worthy

Come on, it's just a box with a hole in it for the light to shine in, radiosity
settings found in this NG, plus some objects I had made before and a mesh I
bought. No big deal. The only thing I did made here is the checkered plane and
the fine tuning.

G.


Post a reply to this message

From: Christoph Hormann
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 16:06:11
Message: <39BE8CF2.87195D4F@schunter.etc.tu-bs.de>
The lighting of the paper walls looks very pleasant, while the other walls seem
to have much artefacts.  I wondered about brightness 2 - i always thought it
should stay near 1. Have you tried brightness 1 and recursion_limit 2 ?

The reflective floor with no visible light reflections on the walls is somewhat
irritating.  

Most other things look extremely good, esp. the chairs (reflectivity works good
on them), but using these high end radiosity settings in a more complex scene
will probably be real pain :-)

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/


Post a reply to this message

From: Gilles Tran
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 16:21:59
Message: <39BE8FE3.CEA94976@inapg.inra.fr>
Christoph Hormann wrote:

> The lighting of the paper walls looks very pleasant, while the other walls seem
> to have much artefacts.  I wondered about brightness 2 - i always thought it
> should stay near 1. Have you tried brightness 1 and recursion_limit 2 ?

Yes. In fact, the lighting setup requires a limit of 2 (at least in Kari Kivisalo
setup). The trouble is that using a limit of 2 I had a terrible answer from the
character mesh (very ugly artifacts), not to mention the exploding render time.

> The reflective floor with no visible light reflections on the walls is somewhat
> irritating.

Photons would certainly be appreciated. Not sure about the render time though. I
also tried a version with media... Cough, cough.

> Most other things look extremely good, esp. the chairs (reflectivity works good
> on them), but using these high end radiosity settings in a more complex scene
> will probably be real pain :-)

That's precisely what I want to know. Right now it's quite acceptable and much
better than any radiosity I've done with Povray, but there's still a lot to test.
Arnold-like things are very nice but if I can't replicate some of the effects in a
real scene with a render time compatible with my lifespan I'll be very
disappointed. Note that at the same time, a complex scene could be more tolerant to
radiosity artifacts so there's hope.

G.


Post a reply to this message

From: 25ct
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 16:24:10
Message: <39be90ea@news.povray.org>
Thanks for showing me, (and others),  just 'what can' be achieved
Gilles. I like this very much.

    ~Steve~


"Gilles Tran" <tra### [at] inapginrafr> wrote in message
news:39BE8414.E6634119@inapg.inra.fr...

   [image]


Post a reply to this message

From: Josh English
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 16:30:19
Message: <39BE9238.42962287@spiritone.com>
Wow. Gilles Tran admits to "no artistical merit" and still produces an amazing
image.
The pink bar in the sky is a little odd though, and the lights are a bit dark,
the floor, I think, should have more light on it in general, because there is
too much shadow.

Josh

Gilles Tran wrote:

> Disclaimer : this scene has no artistical merit whatsoever. It's ugly,
> but it's just a test for a radiosity setup that could be acceptable for
> a "real" interior scene.
> It uses some of the recently gained radiosity knowledge posted in this
> group (Arnold settings and outdoors lighting particularly). There's only
> one light source, coming from the window. Everything else is indirect
> lighting. Balancing the light was difficult. I ended up raising the
> brightness to 2 and lowering the Kari Kivisalo light intensity.
> Assumed_gamma is 1. It took 4 hours to render on a PIII 733 at 800*600
> aa 0.3.
>
> Some of the radiosity effects are really OK. Vicky's face is nicely lit,
> and there's even a backlit effect on her right arm. The shadow of the
> chair on the left seems good too. On the bad side, there are quite a
> number of radiosity artifacts visible on the right wall and on Vicky's
> face and right foot (note : raising the recursion level to 2 had
> appalling effects on Vicky so I left it to 1). Also, I had to add
> reflection to the chairs to make them look better, which lets me think
> that one last thing to improve in Megapov radiosity is it's ability to
> react like a regular light with some of  the texture properties
> (highlights). I didn't try to "cheat" with light groups or area lights
> to fix that, though I'll do it probably on a real pic.
>
> Credits : Victoria is the hi-detail Poser girl from Zygote (dress by
> Catherine Todd). The chairs and the library are available on my website
> (when it works). The wallpaper is just terrible.
> The whole scene (but big Vicky who weights around 15 Mb) is posted on
> p.b.s-f.
>
> G.
>
>   ------------------------------------------------------------------------
>  [Image]

--
Josh English -- Lexiphanic Lethomaniac
eng### [at] spiritonecom
The POV-Ray Cyclopedia http://www.spiritone.com/~english/cyclopedia/


Post a reply to this message

From: Christoph Hormann
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 17:18:00
Message: <39BE9DC5.261C3B4D@schunter.etc.tu-bs.de>
Gilles Tran wrote:
> 
> Yes. In fact, the lighting setup requires a limit of 2 (at least in Kari Kivisalo
> setup). The trouble is that using a limit of 2 I had a terrible answer from the
> character mesh (very ugly artifacts), not to mention the exploding render time.
> 

That's probably true.
I just started a try myself, but can't judge for the mesh of course.  

> Photons would certainly be appreciated. Not sure about the render time though. I
> also tried a version with media... Cough, cough.

There are sometimes things that require waiting for a faster computer :-) Doing
this scene with media would definitely be interesting.  I did not use photons
very often but they would also be useful when adding a glass pane to the window.

> 
> > Most other things look extremely good, esp. the chairs (reflectivity works good
> > on them), but using these high end radiosity settings in a more complex scene
> > will probably be real pain :-)
> 
> That's precisely what I want to know. Right now it's quite acceptable and much
> better than any radiosity I've done with Povray, but there's still a lot to test.
> Arnold-like things are very nice but if I can't replicate some of the effects in a
> real scene with a render time compatible with my lifespan I'll be very
> disappointed. Note that at the same time, a complex scene could be more tolerant to
> radiosity artifacts so there's hope.
> 

I made the experience that combining isosurfaces with radiosity can be very
slow.  I would probably use less slow radiosity settings and fake some things
with additional light sources etc. instead.

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/


Post a reply to this message

From: Xplo Eristotle
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 17:41:54
Message: <39BEA3DA.EB26603F@unforgettable.com>
Gilles Tran wrote:
> 
> Some of the radiosity effects are really OK. Vicky's face is nicely lit,
> and there's even a backlit effect on her right arm. The shadow of the
> chair on the left seems good too. On the bad side, there are quite a
> number of radiosity artifacts visible on the right wall and on Vicky's
> face and right foot (note : raising the recursion level to 2 had
> appalling effects on Vicky so I left it to 1). Also, I had to add
> reflection to the chairs to make them look better, which lets me think
> that one last thing to improve in Megapov radiosity is it's ability to
> react like a regular light with some of  the texture properties
> (highlights). I didn't try to "cheat" with light groups or area lights
> to fix that, though I'll do it probably on a real pic.

There doesn't seem to be anything we can do about the artifacts. I think
it's a flaw in the current algorithm, it's not optimised for really low
error_bound, and there's no way around it except rewriting the code.
Whoever wrote it (Nathan?) would know more about that than I would, though.

The only thing I can think of is to turn the count WAY up, use a tiny,
tiny pretrace_end (values lower than 0.01 seem to be valid on large
images, so you could give that a shot), and maybe increase some other
thing to 20 (I can't remember what it was off the top of my head,
probably was nearest_count.. but whatever it is, it maxes out at 20, so
you can check the dox for it). I doubt that'll get rid of the artifacts,
but it might smooth them a little more.

The artifacts aren't all that noticible to begin with, though. You could
probably ignore them altogether unless they really make your scene look bad.

It might be safe to add photons to the floor; I haven't done extensive
testing, but they didn't have a dramatic effect on my bonsai tree render
(the shiny part of the table thingy had photons). Come to think of it,
they didn't have a dramatic effect on the output either, but.. eh, whatever.

You might add some texture to the plain wall and the ceiling; not only
would it look a little more interesting, but it would probably break up
the artifacting on the right wall.

The wallpaper IS hideous.

The window could use some glass, or something.

The bookcase could probably benefit from a slight application of that
"fake highlight" reflectivity, and the floor might look better if it was
a little more reflective, and blurry.

Some media would be nice too.

Oh, and you could buy a second computer to use while the first one
renders this thing with media, photons, a ridiculously high radiosity
count, isosurfaces and/or normal_on (for the wall and ceiling textures),
and reflection blur... ^^;;

(2006, man. It's only six years away...)

-Xplo


Post a reply to this message

From: David Fontaine
Subject: Re: Radiosity - interior scene test (90 kbu)
Date: 12 Sep 2000 17:52:59
Message: <39BEA492.812D2516@faricy.net>
Gilles Tran wrote:

> ryan constantine wrote:
>
> > i'm not worthy
>
> Come on, it's just a box with a hole in it for the light to shine in, radiosity
> settings found in this NG, plus some objects I had made before and a mesh I
> bought. No big deal. The only thing I did made here is the checkered plane and
> the fine tuning.

Really! Get a grip.

--
David Fontaine   <dav### [at] faricynet>   ICQ 55354965
Please visit my website:  http://davidf.faricy.net/


Post a reply to this message

Goto Latest 10 Messages Next 10 Messages >>>

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.