What do you guys use for realistic window glass?

[kosso_olli]

Not really, i only recall a convo with Vlado from a long time ago.
I've asked again to make sure it's still current.

I see. Any other fog color different from white should work, right?

[^Lele^]

Well, this was some refuse i carried around all these years.
Fog is *not* needed to get proper total internal reflections, just the "reflect on back side" switch is.

I'm ever so sorry to have led you astray.

[kosso_olli]

Understood.

[glorybound]

What makes this wrong?

[seandunderdale]

Cheers for the info Lele. Its been bugging me for long time. Visually I always got better results with it on, but Id been told by others it was for single sided geo only.

[Macker]

What makes this wrong?

The index of refraction being unlinked and set to 1.1, and reflect on backside disabled.

The most basic glass setup should start out with pure black diffuse, pure white reflection and refraction (VRay takes care of not breaking the laws of thermodynamics internally) and an IOR of around 1.5-1.6. From there on out it's about tweaking colour (fog) and reflectivity (ior).

[Richard7666]

The index of refraction being unlinked and set to 1.1, and reflect on backside disabled.

The most basic glass setup should start out with pure black diffuse, pure white reflection and refraction (VRay takes care of not breaking the laws of thermodynamics internally) and an IOR of around 1.5-1.6. From there on out it's about tweaking colour (fog) and reflectivity (ior).

So we no longer have to worry about reflections at 255 white, as V-ray under the hood will 'clip' anything that's too crazy?

[seandunderdale]

Vray wont clip anything (as far as I know) if you're working fully linear. You'll need to use some tone-mapping / colour mapping to control those highlights. But using 255 white in the reflections means that at fully glancing angles your shader will return the values in your scene or hdri dome. If your lights source or dome is using realistic values, that isnt a huge problem. Obviously nothing is a perfect surface so maybe shouldnt return the full value of the light course, but Id argue that we are talking about such small differences if you use 255 or 250 white that it wont make or break the image. I usually find people obsessing over issues like 255 white or not for reflections, but that's the least of the problems when you see the shot they're working on.

[glorybound]

The index of refraction being unlinked and set to 1.1, and reflect on backside disabled.

The most basic glass setup should start out with pure black diffuse, pure white reflection and refraction (VRay takes care of not breaking the laws of thermodynamics internally) and an IOR of around 1.5-1.6. From there on out it's about tweaking colour (fog) and reflectivity (ior).

I did the things suggested, but the glass turned out bad. When is V-Rays material library coming out?

[seandunderdale]

Does your glass have thickness? Have you got an example of the before and after? I've always used the settings that we recommended for my car stuff.

Black diff / 255(ish) reflect col / reflect on backside on/ 255 refract col / tweak fog colour based on tint / play with bias and multiplier to get the look I want. I also sometimes put a wavy bump map in to distort the glass a bit too. Crisp clean reflections can look a bit fake on cars.

[tmbarker531]

I also sometimes put a wavy bump map in to distort the glass a bit too.

I've always done this when working on larger buildings, using a multisubtex with different noise. Recently though I've been having a lot of issues with the the refraction also being wavy when all I want is the reflection to be wavy. When you look at most buildings the reflections are fairly wonky...you never see perfectly clean straight reflections, but i've never seen that effect in the refraction of the glass in those buildings, you'd only ever see it in very old buildings where the glass has slumped over a long long time.

I'm assuming in real life this might have to do a lot with the coating on the glass, but does anyone have a good process for reproducing this in the reflection only? I usually model two layers of glass with a small gap between them so you get that realistic effect of the double reflections that are slightly mismatched due to the separate bump maps on each panel. To get around the refraction issue I've been lowering the IOR dramatically and that reduces the effect...but it doesn't fully take it away and it seems like there's probably a better way.

Does anyone have advice?

[seandunderdale]

Yeh, I had the same wonky refraction issue with car window glass. If I can get away with bumping the shader as a whole subtly, I will. If the refraction is noticeably distorted, Ill make a blend shader. Duplicate the glass shader so I have two glass shaders the same. Turn reflection off on one, and turn refraction and fresnel off on the other. (Basically a chrome) I drop the first glass (with no reflection) in the base of the blend, then put the chromey shader in the first blend slot, and use a fresnel to blend it over the base glass. You can then drop your bump map into the chrome glass shader and bump it till it looks right, and your glass refractions wont be affected.

A side effect is that the glass doesn't render exactly the same as if it was one shader. It doesn't look wrong, just different. So as long as you're ok with that, then I've found no issues with this method yet.

[Macker]

I did the things suggested, but the glass turned out bad. When is V-Rays material library coming out?

By what measure was it bad? I've only just seen this; reflect on backside was *meant to say enabled*.