Real looking glass (yes, again)

[Lupaz]

Does this glass look awful to you?
http://www.itsartmag.com/features/it...l-render-5.jpg

A rendering engine uses science. It's physics (optics) with maths (programming). Actually, the only useful thing that we do is change variables through an interface, because it's still a biased software, to make renderings take much less time than an unbiased rendering engine. Wait 20 years and we won't have a job because computers will do this with no human intervention. Actually, post production is starting to disappear, even in major productions.

Anyway, if anyone has an idea of why it's easier to get good glass with maxwell than with vray, I'd be very interested to know.

[Neilg]

It does, yeah.

The only difference is maxwell computes the caustics as standard, you cant turn them off. vray has them as a separate feature.
Dont even know where to start with you saying post production is starting to dissapear... that's not even close to being true and never will be.

[pokoy]

I have to agree with the OP, Vray's glass was always a bit too 'CG' for me. Coming from Brazil I always thought there's some sort of simplification going on so it misses some of the magic. Not sure if it has been mentioned by someone here before, but make sure you're setting the cutoff threshold to something lower, the default value is way too high and will cut off reflections/refractions too early. Not sure if it helps but give it a go if you didn't already.
FYI, the built in Fresnel map has issues, it's probably better to use the Fresnel checkbox in the VRayAdv material.

[flino2004]

as Cubiclegangster said, reflective materials depend of what is around them...glass looks realistic when everything else look realistic as well in my opinion.
glass in windows always have a distortion, glass panels are not 100% flat for big surfaces because the frames are not 100% straight.
more transparent the glass, less reflective looks like...and in real life glass are something in between 80-95% transparent for clear glass, tint is another factor, CA, thickness...specially if you want to be picky modelling 2 panels with air in the middle for insulated double glass.
.
in your examples the renderings don't look very good besides the glass factor...they are not good examples

[plastic_]

I agree with Cubiclegangster, the problem with this renderings, as well with the Maxwell examples, is everything but the glass shader.
Glass is probably the most simple material of all to set up. The problem is that it is impractical or complicated to set up the geometry and lighting in the complexity that's required to make it look realistic, like architectural photographs.
Look at technicad drawings how glass panels are constructed. It's not a just plane with thickness. There's much more going on and all this parts reflect and refract and this is what you see in glass.
The truth is, it's not only the windows that look bad, it's also the other elements, but you don't care as much about them.

[Lupaz]

This is an image I just did.
It's a box with rounded edges, smoothing groups cleared so we don't get distortion.
It's 100% reflective with vray fresnel.
100% refractive with a slight fog (otherwise the result was worst).
Environment (high resolution (150 MB) HDRI on a vray dome light.

That doesn't look like nice glass to me. I don't have maxwell and I don't know mental ray, to make a comparison.
I don't think the environment is the problem. You can't get better than HDR unless you spend 3000 months building a realistic environment.

BTW thanks for the tips. I forgot about caustics. What cutoff? do you mean the "max depth".
Again I think the biased characteristic of vray is making this task harder than it is with an unbiased render engine. IMHO.

[Neilg]

What about it doesn't look like 'nice glass'? it's just a square over a background image, what are you expecting it to look like? I really don't understand what you're getting at here.

[pokoy]

I don't think this example is saying much and I bet you can't get any renderer to make this look more realistic. After all this looks like a glass pane to me.
I meant the subtle stuff that's going on in inter-reflections-refractions in, let's say, a complicated glass vase.

[Lupaz]

It's not just a background image. It's HDRI spherical background. The photo is a beach or something, the sky has a blue color with clouds and there's a sun, etc.
Also, it's not a square, it's a box with rounded edges (isn't glass like that?).
If I buy a piece of glass and put the glass on the ground in a place like that in real life, and take a photo with this angle, I don't think it'll look like this. Do you agree?

What's my point? I've been doing this for at least 10 years. I never came with a really good glass material. Also, computers are getting fast enough for software like maxwell.

I enjoy the speed that vray gives me by being biased, but I think it's time to move forward a little in terms of realism vs performance.
Is chaosgroup going to do something about it in future versions? because with time an unbiased software will be totally usable.

[plastic_]

I think you're looking for stronger dynamic contrast in the reflections, probably specular highlights. I'd blame the hdri/environment setup, not the glass shader or rendering engine.
I don't think that, with the same assets, it would look different with maxwell or mental ray. I'm curious if it does though.
I agree that managing specular highlights has become a pain in the ass with GI/raytracing rendering though. It was so easy before that.

[Lupaz]

May be the fresnel curve that vray uses is not accurate?

[Neilg]

http://refractiveindex.info/?group=GLASSES&material=BK7

draw your own with the reflection calculator. Use the green line, drop in a perpendicular/parallel falloff map & edit the mix curve so it looks like the green line (non polarized). you can set points and type in values which you can cross reference from that graph. bear in mind the mix curve is 0-100 in the horizontal and that graph is 0-90degrees - so you've got to convert the positions slightly.
good luck!

[Lupaz]

You are joking, right?
In case you aren't, wouldn't it be nice to have a standard vray glass material with all that figured out by default?

[Neilg]

the point I was making is that you should do it and see just how much of a difference a 100% accurate fresnel curve makes.

(i'll give you a hint: very little)

[Lupaz]

Search on the Forum "glass". How many times did you read someone having trouble with creating glass material? I'm not the only one.
A vrayGlassMtl would be very welcome, or it would be great that someone from chaos group sends us a hint on why could it be that glass in maxwell looks so much better.