Glossy reflections behaviour

[studioDIM]

Hey Vlado, I'd like to understand why this behaves as it does:




I am missing something of VRay's behaviour, and this is what made me go completely tits-up trying to auto-expose scenes with glossy reflections...
In fact, i now can safely tone reflection coefficients down by a zillion times, and have the specular highlight still look good, if they are glossy (even by a tiny amount).

Can you shed some light?
Walking in the dark and fearing for an obstacle, here...

Thanks!

Lele

[cpnichols]

Is it a Vray Sun? If so, make it sure it is Visible. I think the moment you put a glossyness in there, vray makes it a specular (not a reflection) and the sun becomes visible again.

[studioDIM]

It was set to visible in both cases, unfortunately
I tried again with the sun set to invisible, so that only the atmospheric glow would be reflected, but it behaves exactly the same: glossies are around 100 times brighter than non glossy reflections, but ONLY where the specular highlight is...
The normal reflection just gets calculated normally, and the variance is very minimal, if there at all.

Lele

[instinct]

hmmm....another wild guess on my side here...i would guess that as soon as there's a glossy reflection the different glossy rays are adding up. And directly around the reflection of the sun it gathers a LOT of light. But that could just as well be complete bullshit :P

Thorsten

[studioDIM]

Hey Thorst, i guess that should be easily verifiable.
If that's correct, wouldn't raising subdivisions in either AA or glossies raise the brightness?
Also, the same thing happens for normally decaying lights, not only in the case of a sun...
Am on it now, anyways.

Lele

Edit: It works PERFECTLY FINE with normal lights. that IS odd.
I'm guessing you're right Thorsten, in thinking it's possibly due to the way the Sun (or sky, it's visibility matters not) iterates with the glossy mats...
Re-Edit: Tried with both standard and physcam, it works the same way with either: glossies tend to soften reflections a wee bit (30% or thereabout a 0. when they are reflecting standard lights...

[studioDIM]

Vlado, any ideas?
This thing is nagging me to no end

I don't understand why, and have cracks starting to appear on the wall by the side of my desk by now. The head's fine of course

Lele

[jacksc02]

drop vlado an email dude

[vlado]

Ok, let me make a few examples....

Best regards,
Vlado

[studioDIM]

ah thanks a LOT man

Lele

[vlado]

So, as far as I understand, your concern is that when you turn glossy reflections on, you get the superbright highlights.

Here is a rendering of a non-glossy reflective sphere:

There don't appear to be any too bright pixels.

Now, let's move a bit closer:

Here you can just see what is causing your problem: there is a superbright group of 2x2 pixels (RGB values are about 60000) which is where the sun is reflected in the sphere. In fact, this superbright dot should have been in the first image as well, however the sphere is curved and the actual probability that a reflection ray hits the small sun directly, is quite low.

If we render a blowup of that region, this is what we get:


Now, if we make the material glossy, this bright dot will spread around and become bigger, while loosing some of its intensity - the more rough the surface is, the less bigger and bright the hilight will be:

In this image, the hilight has intensity only about 660, compared to 60000 for the mirror reflection. Again, this is because the bright area is bigger, and so its intensity must be reduced in order to achieve the same overall result.

If you start to increase the size of the sun, you can make it visible even in the first image. As the sun becomes bigger, it's intensity in the reflection will decrease, so as to preserve the overall illumination. However, the hilight values will not change much, as they are already blurring the sun.

For example, here is an image where the sphere glossiness is 1.0, but the sun size multiplier is 20:

The intensity of sun in this image is only about 180, compared to thousands in the second image above.

I hope this makes things somewhat more clear.

[Note] There does appear to be some issue where the glossy hilight is somewhat (but not too much) brighter than expected with large sun sizes, and I will look at this, but it doesn't change things in the principle.

Best regards,
Vlado

[studioDIM]

That is what i thought would cause the behaviour in the first place; however, even hiding the sun from the reflections behaves in the same way.

Is it because even if it's hidden the sky area by the sun gets to glow extremely hot?

EDIT: that's precisely why. Thanks Nikki/Vlado.
I'm trying different methods to expose the camera now.
I hope i'll be able to manage some smart code...

Thanks again for taking the time, Vlado.

Lele