The POV documentation on area lights (3.4.7.5) has what seems to be some
ambiguity.  Here's part of the description:
"...it is approximated by an array of point light sources spread out over
the area occupied by the light. The array-effect applies to shadows only.
The object's illumination is still that of a point source."

The word "object" here is somewhat ambiguous. In one or several places in
the documentation, a light source, in general, is referred to as an object
(like any other object.) But in this context, I read it to mean the object
BEING illuminated by the light...although I'm not sure about that. Either
way, I'm a bit confused about this description of an area light being just
a "point source."

To better illustrate what I'm talking about: If I purposely place a typical
object like a box at the location of the area light, but offset the box
slightly, it appears to occlude part of the area light's illumination on
other objects in the scene. (The overall light gets "grainy" or spotty.)
Which leads me to conclude that the box is blocking some of the area light
array. And that it does occupy a certain spatial dimension, not just a
point.

Or does the documentation actually mean, in a literal way, that the
illumination from the area light emanates from just ONE point--as far as
the objects being illuminated are concerned? In other words, do the
illuminated objects in the scene "see" the area light as just a single
point?

Admittedly, these are rather arcane technical questions, probably of no real
importance. But does anyone besides me see the confusion here?

Ken Walker

Post a reply to this message

Le 01/18/2007 07:14 PM, Kenneth nous fit lire :
> The POV documentation on area lights (3.4.7.5) has what seems to be some
> ambiguity.  Here's part of the description:
> "...it is approximated by an array of point light sources spread out over
> the area occupied by the light. The array-effect applies to shadows only.
> The object's illumination is still that of a point source."

> Or does the documentation actually mean, in a literal way, that the
> illumination from the area light emanates from just ONE point--as far as
> the objects being illuminated are concerned? In other words, do the
> illuminated objects in the scene "see" the area light as just a single
> point?
> 

I believe the document is speaking about phong/glossy/finish effects
generated on the object by the area light: the highlight generated
on a shining sphere would be identical with an <area light> or a
<point light>.

It might be different if instead of an <area light> object, you used
a real collection of point lights... but it might be slower, and the
difference might be small, even too small to notice.


> Admittedly, these are rather arcane technical questions, probably of no real
> importance. But does anyone besides me see the confusion here?
> 
> Ken Walker
> 
> 
> 


-- 
The superior man understands what is right;
the inferior man understands what will sell.
-- Confucius

Post a reply to this message

Kenneth <kdw### [at] earthlinknet> wrote:
> The word "object" here is somewhat ambiguous.

  Yes. "Surface" would be much better.

> In one or several places in
> the documentation, a light source, in general, is referred to as an object
> (like any other object.) But in this context, I read it to mean the object
> BEING illuminated by the light...although I'm not sure about that. Either
> way, I'm a bit confused about this description of an area light being just
> a "point source."

  As the documentation says, the area light parameters are taken into
account only when calculating shadows. When calculating direct illumination
they are ignored (which means that in direct illumination it's just a
point light source, like everything you wrote in the light source related
to area lights was removed).

  (And no, this is not necessarily a good thing. It is just implemented like
that currently.)

> To better illustrate what I'm talking about: If I purposely place a typical
> object like a box at the location of the area light, but offset the box
> slightly, it appears to occlude part of the area light's illumination on
> other objects in the scene. (The overall light gets "grainy" or spotty.)
> Which leads me to conclude that the box is blocking some of the area light
> array. And that it does occupy a certain spatial dimension, not just a
> point.

  The box casts a shadow, which is calculated from the area light
parameters. Naturally if you want the shadow to be any good, with
the box being so close to the light source, you'll have to make the
area light quite dense and use a high 'adaptive' value (or not use
it all, which will naturally make the whole rendering quite slower).

> Or does the documentation actually mean, in a literal way, that the
> illumination from the area light emanates from just ONE point--as far as
> the objects being illuminated are concerned? In other words, do the
> illuminated objects in the scene "see" the area light as just a single
> point?

  Whether the objects "see" the light source is determined by the
shadow calculations, which takes into account area light paramters.
  Once povray calculates that the object indeed "sees" the light
source (even if partially) then it calculates the illumination using
only the point light source part of the light.

  Btw, as a curiosity: If you are wondering whether it makes any
considerable difference whether the direct lighting is calculated
as a point light source or as an area light, the answer is that in
some scenes it makes a big difference:

http://warp.povusers.org/images/arealight.jpg
http://warp.povusers.org/images/lightgrid.jpg

-- 
                                                          - Warp

Post a reply to this message

Le Forgeron <jgr### [at] freefr> wrote:
> I believe the document is speaking about phong/glossy/finish effects
> generated on the object by the area light: the highlight generated
> on a shining sphere would be identical with an <area light> or a
> <point light>.

  No, it's talking about *all* direct illumination.

> and the
> difference might be small, even too small to notice.

  It's easy to construct a scene where the difference is very notable.
As I already quoted in my previous post to this thread:

http://warp.povusers.org/images/arealight.jpg
http://warp.povusers.org/images/lightgrid.jpg

-- 
                                                          - Warp

Post a reply to this message

Le Forgeron <jgr### [at] freefr> wrote:



> It might be different if instead of an <area light> object, you used
> a real collection of point lights...

An interesting idea. I'll do an experiment, to compare this to an area light
(of the same number of point lights), to see just what a visual difference
there might be on a scene's illuminated objects, as well as the shadow
qualities produced.

KW

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:

>   The box casts a shadow, which is calculated from the area light
> parameters.

Ah yes. I had neglected to consider that.

>   Whether the objects "see" the light source is determined by the
> shadow calculations, which takes into account area light paramters.
>   Once povray calculates that the object indeed "sees" the light
> source (even if partially) then it calculates the illumination using
> only the point light source part of the light.

That's quite fascinating. So in essence, POV-Ray is performing a sort of
"backwards" calculation when using an area light--the shadow being the
determing factor.
>
>   Btw, as a curiosity: If you are wondering whether it makes any
> considerable difference whether the direct lighting is calculated
> as a point light source or as an area light, the answer is that in
> some scenes it makes a big difference:
>
> http://warp.povusers.org/images/arealight.jpg
> http://warp.povusers.org/images/lightgrid.jpg
>
Thanks for the images.  It does indeed produce a much bigger difference than
I imagined.

Ken

Post a reply to this message

"Kenneth" <kdw### [at] earthlinknet> wrote:
> Le Forgeron <jgr### [at] freefr> wrote:
>
>
>
> > It might be different if instead of an <area light> object, you used
> > a real collection of point lights...
>
> An interesting idea. I'll do an experiment, to compare this to an area light
> (of the same number of point lights), to see just what a visual difference
> there might be on a scene's illuminated objects, as well as the shadow
> qualities produced.
>

Of course, an area light can use jitter, which would automatically make the
shadows (from the area light's discreet point lights) smoother, compared to
using a similar array of "real" point lights. The latter would simply cast a
number of separate, discreet shadows, all slightly displaced.

It would still be interesting to see the difference in the surface
illumination of an object, comparing both methods.  I.e., the area light's
single "point light" direct illumination vs. an array of real point lights.

Here's a really wacky experiment to try (though I haven't done ot yet):
Start with two copies of the same object. Use the area light to create the
shadows on one copy--but use the no_image tag for that object, so it's
invisible.  Then light the 2nd copy with the array of "real" point
lights--but use the no_shadow tag for that one. So what you end up with is
the object lit by multiple real point lights, with the shadow created by
the area light.  The best(?) of both worlds!

KW

Post a reply to this message

Kenneth <kdw### [at] earthlinknet> wrote:
> It would still be interesting to see the difference in the surface
> illumination of an object, comparing both methods.  I.e., the area light's
> single "point light" direct illumination vs. an array of real point lights.

  Didn't I post links to two images showing this difference?

-- 
                                                          - Warp

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> Kenneth <kdw### [at] earthlinknet> wrote:
> > It would still be interesting to see the difference in the surface
> > illumination of an object, comparing both methods.  I.e., the area light's
> > single "point light" direct illumination vs. an array of real point lights.
>
>   Didn't I post links to two images showing this difference?
>

Hmm. Didn't know that for sure at the time (since your discription of the
images didn't give specific details.)  But I stand corrected. I should have
realized this.

Do the white blocks represent the extent of the light array? I suppose it
should be obvious, but I don't want to assume.

KW

Post a reply to this message

> Warp <war### [at] tagpovrayorg> wrote:
>

> >   Btw, as a curiosity: If you are wondering whether it makes any
> > considerable difference whether the direct lighting is calculated
> > as a point light source or as an area light, the answer is that in
> > some scenes it makes a big difference:
> >
> > http://warp.povusers.org/images/arealight.jpg
> > http://warp.povusers.org/images/lightgrid.jpg
> >
> Thanks for the images.  It does indeed produce a much bigger difference than
> I imagined.
>
> Ken

I should have been more detailed in my follow-up to your images. What
strikes me immediately is that lightgrid.jpg (which I take to be an array
of real point lights) produces *much* smoother illumination on the wall and
sphere than does the area light.  Surprisingly so. And the shadows from the
sphere look smoother than I would have imagined. Do you happen to recall
the difference in rendering times of the two images?

KW

Post a reply to this message