Xplo Eristotle <inq### [at] unforgettablecom> wrote:
: How long did that first one take to render, Warp..

  A couple of minutes.

: And why didn't you bother to light it properly? Why did you even render
: it at all? What are we supposed to glean from this image, other than the
: fact that you're apparently using a nightlight to illuminate the scene?

  I didn't use any light at all (in the first two images).

  My goal was to show that:

  1. The recursion level can have a drastical impact in the image if a
higher recursion level is needed to illuminate parts of the scene that
are not directly facing the illuminating object.

  As I undestand it, the second image is brighter than the first one because
the white floor reflects light to the other objects. At recursion level 1
this doesn't happen. Making the floor dimmer would have probably made the
illumination dimmer as well.

  2. When the radiosity calculations make the rendering very slow, you can
actually use area lights to speed up the rendering without any quality
decrease (you may experience even a quality increase since the direct
illumination is better).
  Someone gave the idea that using area lights is of no use since you get
a completely different image. The third image was made to test if this is
true. Well, it isn't.

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

Warp wrote:
> 
> Xplo Eristotle <inq### [at] unforgettablecom> wrote:
> : How long did that first one take to render, Warp..
> 
>   A couple of minutes.
> 
> : And why didn't you bother to light it properly? Why did you even render
> : it at all? What are we supposed to glean from this image, other than the
> : fact that you're apparently using a nightlight to illuminate the scene?
> 
>   I didn't use any light at all (in the first two images).

<snip>

>   As I undestand it, the second image is brighter than the first one because
> the white floor reflects light to the other objects. At recursion level 1
> this doesn't happen.

So why is the floor so dark in the first image, then? If it's getting
its light from some sort of glowing sky or ceiling, which would seem to
be the case, then it should be just as bright as the floor in the second
image (given that light from the objects doesn't seem to have much
influence on the floor).

I'm inclined to think that this is a bug (and Kari's made comments to
that effect, IIRC), but I wonder what one could produce with some
parameter tweaking.

>   2. When the radiosity calculations make the rendering very slow, you can
> actually use area lights to speed up the rendering without any quality
> decrease (you may experience even a quality increase since the direct
> illumination is better).
>   Someone gave the idea that using area lights is of no use since you get
> a completely different image. The third image was made to test if this is
> true. Well, it isn't.

Your conclusion is pretty premature there, Warp. I don't see anything in
this scene that would depend on photons, specular reflections, "simple"
caustics, harsh direct illumination, or scattering media, the first four
of which would probably not sit well with multiple area lights and the
last one which might also be slower rather than faster. I also don't see
this being of much use in confined spaces, such as autowitch's
recently-posted Egyptian chamber.

-Xplo

Post a reply to this message

Xplo Eristotle <inq### [at] unforgettablecom> wrote:
: Your conclusion is pretty premature there, Warp. I don't see anything in
: this scene that would depend on photons, specular reflections, "simple"
: caustics, harsh direct illumination, or scattering media, the first four
: of which would probably not sit well with multiple area lights and the
: last one which might also be slower rather than faster.

  The comparison was: radiosity without light sources vs. radiosity with
light sources.
  If you don't use light sources at all you don't get any of those effects
anyways (or in the case of scattering media it would probably be seen, but
the light will certainly not interact with it as expected), so I don't
understand why would you use any of those in a scene without light sources.
  If you are using scattering media in a scene without light sources I think
that a combination of emitting and absorbing media would get the same
effect faster. If the media is global, fog would probably do it.

: I also don't see
: this being of much use in confined spaces, such as autowitch's
: recently-posted Egyptian chamber.

  If you remember, Kari Kivisalo made his cornell box test using an area
light source. I don't remember the rendering time, but IIRC it was some
hours.
  I made the same test myself, but without any light source (I discussed
this in p.u-p if I remember correctly). If we don't count the artifacts
in the corners (which appear for a reason yet to be explained), the
overall illumination is rather similar. The render, however, took more than
20 hours.
  The difference in speed is notable.

  (Btw, the image in question can be found at
http://www.cs.tut.fi/~warp/cornell/cornell.jpg)

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

Warp wrote:
> 
> Xplo Eristotle <inq### [at] unforgettablecom> wrote:
> : Your conclusion is pretty premature there, Warp. I don't see anything in
> : this scene that would depend on photons, specular reflections, "simple"
> : caustics, harsh direct illumination, or scattering media, the first four
> : of which would probably not sit well with multiple area lights and the
> : last one which might also be slower rather than faster.
> 
>   If you don't use light sources at all you don't get any of those effects
> anyways (or in the case of scattering media it would probably be seen, but
> the light will certainly not interact with it as expected)...

My point is that (a) you *need* light sources for those effects, and (b)
they're NOT going to look the same if you have area lights scattered
everywhere, and therefore (c) your solution is a highly limited one, not
suitable for most scenes.

> so I don't
> understand why would you use any of those in a scene without light sources.

Okay, now you're just being stupid.

>   If you are using scattering media in a scene without light sources I think
> that a combination of emitting and absorbing media would get the same
> effect faster. If the media is global, fog would probably do it.

Um, not from what I understand. Radiosity can be set to affect media -
yes? - which means that a bright object should glow in a scattering
media, which it would not do in any other sort of media, nor in fog.

> : I also don't see
> : this being of much use in confined spaces, such as autowitch's
> : recently-posted Egyptian chamber.
> 
>   If you remember, Kari Kivisalo made his cornell box test using an area
> light source. I don't remember the rendering time, but IIRC it was some
> hours.

Kari used ONE light, positioned in the one place where light would
logically enter the room: the lit panel in the ceiling. That's not even
remotely the same thing as your horde of "trick" fill lights.

-Xplo

Post a reply to this message

Xplo Eristotle <inq### [at] unforgettablecom> wrote:
:> : Your conclusion is pretty premature there, Warp. I don't see anything in
:> : this scene that would depend on photons, specular reflections, "simple"
:> : caustics, harsh direct illumination, or scattering media, the first four
:> : of which would probably not sit well with multiple area lights and the
:> : last one which might also be slower rather than faster.

: My point is that (a) you *need* light sources for those effects, and (b)
: they're NOT going to look the same if you have area lights scattered
: everywhere, and therefore (c) your solution is a highly limited one, not
: suitable for most scenes.

  Besides being odd that you want to use high-diffuse objects as radiosity
light sources and separate regular light sources, there still isn't any
problem.

  Photons? Don't add a photons-block in the area light.
  Scattering media? Turn media_interaction off for the area light.
  Simple caustics? What's the point in using a light source which doesn't
cause caustics? If you _really_ want "fake" caustics and radiosity (from
different light sources), then don't use this trick. But I'm sure that this
case is quite rare (I would not talk about "most" scenes).
  What is "harsh direct illumination"? Does direct illumination look a lot
different in my second and third example images? A bit perhaps, but not much.
I wouldn't call it "useless".

:> so I don't
:> understand why would you use any of those in a scene without light sources.

: Okay, now you're just being stupid.

  I still think that using bright objects and regular light sources in
different places is odd. Usually you attach a light source to the bright
object if you want any light source.

: Um, not from what I understand. Radiosity can be set to affect media -
: yes? - which means that a bright object should glow in a scattering
: media, which it would not do in any other sort of media, nor in fog.

  Have you tested how differently an area light source affects a scattering
media compared to a bright object without a light source?

:> : I also don't see
:> : this being of much use in confined spaces, such as autowitch's
:> : recently-posted Egyptian chamber.
:> 
:>   If you remember, Kari Kivisalo made his cornell box test using an area
:> light source. I don't remember the rendering time, but IIRC it was some
:> hours.

: Kari used ONE light, positioned in the one place where light would
: logically enter the room: the lit panel in the ceiling. That's not even
: remotely the same thing as your horde of "trick" fill lights.

  You talked avout "confined spaces, such as <example image>". If I remember
correctly, that image had one light source?
  Usually confined spaces have specific small light sources.

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message