No radiosity was used in the making of this image.

I created a bezier patch light source...it's kind of cludgy but proves
the notion that a light source in the shape of a dome can create the
same results as radiosity with only one recursion level in less time.
One thing that stands out is the strange coloration of shadowed areas
where the yellowish sun hits (similiar to the arnold renders).  The
light acting as the sun probably needs to be added to the previous light
calculations - I suspects it gets multiplied by the shadows.  I've been
thinking that a dome area light should be an option in sky_sphere, which
would make the implimentation much easier.

Anyway, render time was 5 minutes on a 400mhz celeron.

-Mike

Post a reply to this message

Attachments:
Download 'norad.jpg' (46 KB)

Preview of image 'norad.jpg'

> I created a bezier patch light source...

Awesome! Sounds wonderfully useful. Just think of it: you can make sPatch
models emit light in their true shape, and all kinds of cool stuff like
that. Now we can also make more realistic lights for lamps and such. Thank
you, thank you. Send it to Nathan so he can add it to MegaPOV 0.6b or
something... :)

Post a reply to this message

Sorry I don't get it... Is it possible to make POV emit light from more
than a single infinite small point? If so, I'll be a happy man... But I
only know the solution of setting up, for example 50 point light sources
in some sort of shape, to simulate a "real" area of light.. Every light
in real life, actually comes from an area.

"Area lights" in POV are nice but they only affect shadows, and not the
light itself. Which is logical. The usual problem on todays 3D programs
are that they use incorrect and simplified models of physical laws.

Hugo.

Post a reply to this message

> "Area lights" in POV are nice but they only affect shadows, and not the
> light itself. Which is logical. The usual problem on todays 3D programs
> are that they use incorrect and simplified models of physical laws.

and were not here using a raytracer for the good of our health now are we
:) - long renders go with the territory!

Rick

Post a reply to this message

> Awesome! Sounds wonderfully useful. Just think of it: you can make sPatch
> models emit light in their true shape, and all kinds of cool stuff like
> that. Now we can also make more realistic lights for lamps and such.

Should be possible to do things like that - theoremetically.

> Send it to Nathan so he can add it to MegaPOV 0.6b or something... :)

It's not there yet.  I've been working out some kinks, and I'm not sure what
the final implementation should be.  I supposed it can just stay part of the
light source, in which case it is ready except for some scaling problems and
how to go about deciding on the center of the light (which should really be in
the center of the patches bounding box, which brings up the question of
whether or not the patch should be inherited by using looks_like on a
patch...my head hurtz)

Post a reply to this message

MikeH wrote:

> Should be possible to do things like that - theoremetically.

theoremetically ?

-- 
Ken Tyler

Post a reply to this message

> Sorry I don't get it... Is it possible to make POV emit light from more
> than a single infinite small point?

> "Area lights" in POV are nice but they only affect shadows, and not the
> light itself.

There's been some confusion on this subject and tinkering with the code I
found something interesting.  An area light does emit light from the entire
area.  What's important is how to define light.  What it does is test if the
point is lit or shaded for every area on the surface of the area light.  It
takes all the resulting light colors and shadow colors, without any surface
shading, and averages them together.  The important detail here is that no
shading is done.

Lambertian shading (diffuse) models the falloff of light as a cos of the
incident angle of the light.  This assumes that most of the light is coming
from the same general direction.  I don't think the same rules apply to a
light covering a large area.  Furthermore, phong and specular are light
tricks that mimic the spreading of a reflection of a pointlight by the
roughness of a surface.  Making the highlight larger would probably be all
that is needed (or using reflection of the surface with a high exponent).

Anyway, after performing the area light calculations, you have a softly lit
representation of the scene without any finish.  The problem I ran into is
that when POV does the diffuse calculation it only uses the center of the
light and renders self shadowing black - it also causes the edges of the
diffuse 'highlight' to render black.  This pretty much kills all the soft
lighting effects from the area light.  So I just skipped that part when
doing the patch area light.   :)

It's not perfect but I think it's a step in the right direction.

-Mike

Post a reply to this message

> theoremetically ?

(shhh, I'm using subliminimable messages in my posts)

Post a reply to this message

MikeH wrote:
> 
> No radiosity was used in the making of this image.
> 
> I created a bezier patch light source...it's kind of cludgy but proves
> the notion that a light source in the shape of a dome can create the
> same results as radiosity with only one recursion level in less time.
> One thing that stands out is the strange coloration of shadowed areas
> where the yellowish sun hits (similiar to the arnold renders).  The
> light acting as the sun probably needs to be added to the previous light
> calculations - I suspects it gets multiplied by the shadows.  I've been
> thinking that a dome area light should be an option in sky_sphere, which
> would make the implimentation much easier.
> 
> Anyway, render time was 5 minutes on a 400mhz celeron.
> 
> -Mike
> 

That looks impressive for without radiosity, but this surely only works in some
special cases (if there were walls in the scene, your technique would not help,
same for a sky with no constant color).    

Christoph

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

Post a reply to this message