Just thrown together full support for the "brilliance" keyword in radiosity.

This scene is illuminated by a yellowish classic circular area light 
(with area_illumination) on the right, an emissive disc of matching 
size, colour and brightness on the left, and a blue sky sphere.

All spheres (except the rightmost in the back row) use "diffuse albedo 
1.0", with varying brilliance.

Front row, left to right: brilliance 0.5, 1, 2, 4, 8

Back row: brilliance 32, 64, 256; the rightmost sphere is a reflective 
sphere (with an albedo 1.0 ultra-low roughness specular highlight) for 
reference.

Note that what looks like blurred reflections in the high-brilliance 
spheres is really just a static blurred projection of the environment 
along the surface normal. It might be useful as a "poor man's blurred 
reflection" nonetheless.

(It should also be noted that illuminating radiosity scenes with 
comparatively small bright emissive objects is generally a bad idea, and 
needs crazy high radiosity count settings; in this scene, I used "count 
100000,1000000".)

Post a reply to this message

Attachments:
Download 'radiosity_brilliance.png' (204 KB)

Preview of image 'radiosity_brilliance.png'

Oh, and did you know that the current implementation of brilliance is 
physically bogus?

For virtually all real-world materials, the function that describes how 
much of any given incoming light ray comes out as any other given 
outgoing light ray is a /bidirectional/ one: You can swap incoming and 
outgoing ray, and the equation still yields the same result.

That's not the case for POV-Ray's diffuse surface when using any 
brilliance value other than 1.

This can be fixed however -- and voila: It does provide for pretty neat 
effects.

Left: brilliance 0.7 (using the fixed implementation); this might come 
in handy to model fluffy stuff, such as tennis balls.

Center: brilliance 1.0 for reference.

Right: brilliance 2.0 (using the fixed implementation); should do great 
for black nylons or the like. Might also come in handy as the basis for 
a pearl material I guess.

Post a reply to this message

Attachments:
Download 'radiosity_brilliance.png' (148 KB)

Preview of image 'radiosity_brilliance.png'

Am 22.07.2014 17:07, schrieb clipka:

> Right: brilliance 2.0 (using the fixed implementation); should do great
> for black nylons or the like. Might also come in handy as the basis for
> a pearl material I guess.

... and frosted light bulbs.

Post a reply to this message

Attachments:
Download 'radiosity_brilliance.png' (146 KB)

Preview of image 'radiosity_brilliance.png'

Le 22/07/2014 17:07, clipka nous fit lire :
> Oh, and did you know that the current implementation of brilliance is
> physically bogus?

Most people, I guess, use diffuse. So it might have been unnoticed for eons.

-- 
IQ of crossposters with FU: 100 / (number of groups)
IQ of crossposters without FU: 100 / (1 + number of groups)
IQ of multiposters: 100 / ( (number of groups) * (number of groups))

Post a reply to this message

That is really great! Offers a number of nice possibilities.

Black nylons, hmmm...?  ;-)

Thomas

Post a reply to this message

> That is really great! Offers a number of nice possibilities.
>
> Black nylons, hmmm...?  ;-)

You know you've been raytracing too long when... you read "black nylons" 
and immediately get an image in your head of a square black lump of plastic!

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Oh, and did you know that the current implementation of brilliance is
> physically bogus?
>
> For virtually all real-world materials, the function that describes how
> much of any given incoming light ray comes out as any other given
> outgoing light ray is a /bidirectional/ one: You can swap incoming and
> outgoing ray, and the equation still yields the same result.
>
> That's not the case for POV-Ray's diffuse surface when using any
> brilliance value other than 1.
>
Yes, ...but I consider this current brilliance has its advantage. It has a
simple mathematics form. For users who want to control their own effect, it is
good.

> This can be fixed however -- and voila: It does provide for pretty neat
> effects.

I heard that there is an oren-nayar model. Is it something like that? If so, it
is very good because difference of diffuse model can creat many variety of
texture. The current pov-ray feature which can do this is not rich. Only
normal{...} on a texture can work with radiosity.


> Left: brilliance 0.7 (using the fixed implementation); this might come
> in handy to model fluffy stuff, such as tennis balls.
>
> Center: brilliance 1.0 for reference.
>
> Right: brilliance 2.0 (using the fixed implementation); should do great
> for black nylons or the like. Might also come in handy as the basis for
> a pearl material I guess.


Brilliance for radiosity is good.

Post a reply to this message

Am 26.07.2014 14:19, schrieb And:

>> This can be fixed however -- and voila: It does provide for pretty neat
>> effects.
>
> I heard that there is an oren-nayar model. Is it something like that? If so, it
> is very good because difference of diffuse model can creat many variety of
> texture. The current pov-ray feature which can do this is not rich. Only
> normal{...} on a texture can work with radiosity.

No... not really. Oren-Nayar started with a physical model of how the 
observed effect might be explained, and then derived an exact 
mathematical formula from this model.

What I did was start with an existing mathematical formula (which 
presumably is just a more-or-less random tweak of the lambertian model), 
interpret it as a modulation of the light intensity based on the 
incoming light angle, and apply that very same modulation based on the 
outgoing light angle as well.

> Brilliance for radiosity is good.

That's why I put it in ;-)

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 26.07.2014 14:19, schrieb And:
>
>
> > Brilliance for radiosity is good.
>
> That's why I put it in ;-)

Thank you.

Post a reply to this message