Hi !



i noticed that i could hardly fake "real arealight" present in usual
radiosity scenes (such as cornell box scene or Winosi competition scene for
instance) by using povray arealights.

I tried to add a angle dependant "strength" to fake real arealights (cosinus
law i assume) by using sphere with gradient transparency. (ok, ....only few
lines to add)

It seems very efficient in comparison with pure global illumination
renderings. however it is quite boring to tweak. It can be simple but with
more than 1 light source it is more difficult (because transparent layers
should not be seen by other light sources...)

Is this kind of arealight foreseen in future realeses of povray ? I know
that many renderers have this feature and since global illumination is not
very suitable for small light sources (I always got poor results), it could
be a noticable improvement, doesn't it ?

what do you thing ? is it tricky to code ?

(It is just a question that seems logical to me. Correct me if i am wrong ;
please note that i do not intend to do a stupid blame.)

You can notice though, that in cornell box scene in povray package, light is
faked by using a grid of spots with falloff to simulate cosinus law too (is
it linear in this case ?).

I think It is just another way to cope with global illumination issues with
small arealights.

Please find some experiments :

(1) fakeoriginal : the genuine povray arealight : 10 seconds

http://jcolin.site.voila.fr/galerie/GI/fakeoriginal10sec.jpg

fake : arealight surrounded by a sphere with an angle dependent transprency
texture : 27 seconds

http://jcolin.site.voila.fr/galerie/GI/fake27sec.jpg

real : simple small square with ambient = 1 and GI (i had to set count =
1600 to reduce spots) : 12 minutes

http://jcolin.site.voila.fr/galerie/GI/real12min.jpg

NB : the light size is the same = 1/5 of box length. the modified arealight
seems more accurate.

(2) cornell box scene with faked arealight : 10 minutes with recurtion limit
= 3

http://jcolin.site.voila.fr/galerie/GI/cornell10min.jpg

cornell trick : it shows the transparency layer for the same scene.

http://jcolin.site.voila.fr/galerie/GI/cornelltrick10min.jpg

(3) winosi scene : recution limit =3 (2 in reality + 1 arealight): 4 minutes
!!

http://jcolin.site.voila.fr/galerie/GI/winosi4min.jpg

I tried ith only global illumination and even with 1 day i cannot render a
good picture. Could you confirm ?

Coments are welcome

Kind Regards,

Jerome


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.620 / Virus Database: 399 - Release Date: 12-Mar-06

Post a reply to this message

Hi,

This is an interesting post, and I cannot answer your question really. But
I'm curious regarding the WinOSi rendering: How did you get this result in 4
minutes? The image "winosi4min.jpg" doesn't look like WinOSi. There are
suspicious artifacts around the light on the ceiling. There are no visible
light source. There are hardly any anti-alias... What version of WinOSi are
you using?

Hopefully someone else can answer your question. :o)

Regards,
Hugo

Post a reply to this message

Thanks for you reply Hugo,

The Winosi rendering was made with povray 3.6 b1 and not Winosi after a test
scene from WinOsi homepage. The goal of this scene was to compare various
renderers.
(http://www.winosi.onlinehome.de/Comp1.htm)
(all the picture I posted were rendered with POV)

You noticed it : there is no light source indeed. Actually, there is just an
arealight with tranparency layer (with no_image keyword)

By the way, you can see on winosi home page a render made with povray and an
usual (and not very accurate for this purpose) arealight. I run this scene
to demontrate the use of a different kind of arealight for realistic
results.
I hope someone will be interested in making such a feature more easy to use
without any trick :-)

Bye,

Jerome


"Hugo Asm" <hua### [at] post3teledk> wrote in message
news:4056cdf6$1@news.povray.org...
> Hi,
>
> This is an interesting post, and I cannot answer your question really. But
> I'm curious regarding the WinOSi rendering: How did you get this result in
4
> minutes? The image "winosi4min.jpg" doesn't look like WinOSi. There are
> suspicious artifacts around the light on the ceiling. There are no visible
> light source. There are hardly any anti-alias... What version of WinOSi
are
> you using?
>
> Hopefully someone else can answer your question. :o)
>
> Regards,
> Hugo
>
>


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.620 / Virus Database: 399 - Release Date: 12-Mar-06

Post a reply to this message

> By the way, you can see on winosi home page a render made with
> povray and an usual (and not very accurate for this purpose) arealight.

Yes, in fact I contacted the author of WinOSi some time ago and you will
soon see a rewamped page, including a new and up-to-date POV-Ray rendering.

> I run this scene to demontrate the use of a different kind of arealight
> for realistic results.

It's very interesting, but unfortunately I haven't understood everything
about it. I'm not a whiz-kid when it comes to math.

> I hope someone will be interested in making such a feature more
> easy to use without any trick :-)

I'm sure that we, here at the POV-Ray community, would be very interested to
improve POV-Ray but to get the wise peoples attention, maybe you can post a
more concise layout of your idea. What you posted was fine, but ...for
example, me... I don't get it... Are you saying, the spreading of light from
a flat area-light is not realistic? That it should have a stronger intensity
at the center?

This may be unrelated to your proposal, but are you aware that realistic
light sources should always have the following statement: "fade_distance 1
fade_power 2".

Regards,
Hugo

Post a reply to this message

I agree. I should have given more clarifications :
I took fade power = 2 of course for the experimentations.
Furthermore, your assumption "the spreading of light from a flat area-light
is not realistic. That it should have a stronger intensity at the center" is
correct. (sorry if my explanaison was not clear)

The idea comes from
http://news.povray.org/povray.programming/thread/%3C4044ba89%40news.povray.org%3E/. I
don't know if it has already be used before.

I am looking forward to seeing the new povray rendering on winosi page.

Regards,

Jerome


"Hugo Asm" <hua### [at] post3teledk> wrote in message
news:405812f2$1@news.povray.org...
> > By the way, you can see on winosi home page a render made with
> > povray and an usual (and not very accurate for this purpose) arealight.
>
> Yes, in fact I contacted the author of WinOSi some time ago and you will
> soon see a rewamped page, including a new and up-to-date POV-Ray
rendering.
>
> > I run this scene to demontrate the use of a different kind of arealight
> > for realistic results.
>
> It's very interesting, but unfortunately I haven't understood everything
> about it. I'm not a whiz-kid when it comes to math.
>
> > I hope someone will be interested in making such a feature more
> > easy to use without any trick :-)
>
> I'm sure that we, here at the POV-Ray community, would be very interested
to
> improve POV-Ray but to get the wise peoples attention, maybe you can post
a
> more concise layout of your idea. What you posted was fine, but ...for
> example, me... I don't get it... Are you saying, the spreading of light
from
> a flat area-light is not realistic? That it should have a stronger
intensity
> at the center?
>
> This may be unrelated to your proposal, but are you aware that realistic
> light sources should always have the following statement: "fade_distance 1
> fade_power 2".
>
> Regards,
> Hugo
>
>


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.631 / Virus Database: 404 - Release Date: 17-Mar-04

Post a reply to this message