Maybe not :)

Have been struggling a bit with the GlassBsdf in Cycles as it doesn't really give the result that I wanted. Especially when working with double sided glass or thick glass.

This is from a test I did, where the glass is casting shadows on whatever is behind it, which is very weird.

This is obviously not going to work.

This is a render of two wine glass from a work in process. The wine glass on the left is using my glass material and the one on the right is using the standard GlassBsdf. The small spice glass in the back also uses my glass. But as seen, the glass on the right wine glass doesn't behave even remotely like you would expect from transparent glass. As it is turned all black, which is the same issue with the test above.

This is the same test as above for comparison. To the far left is a plane with the standard GlassBsdf and then I have increased the thickness as we go to the right, the one on the far right uses my shader, which does seem to solve the shadow issue or at least most of it.

If anyone is interested in the setup this is what I used and seems to be possible to apply to all Glass which have thickness, at least from what I can see.

Anyway, since I have no clue what is going on in the material :D I would like to know if anyone that knows more about it than me have any ideas of how to improve it even further, which would be great, because I would like a standard glass shader that I can use for all kinds of things and not having to make a new one all the time, so would like to make it as good as possible. So please share if you know how to do it? :)

I suggest that you Google bagginsbill and glass see what comes up. He has done some extensive work on glass shaders for both Firefly and Superfly with analysis comments. Given the nature of this topic, I wouldn't be surprised if he chimes in. Now a one-fits-all glass shader might be a tall order to fill unless we're talking background objects as prop fillers to complete a scene. But using glass objects in fore- and mid-ground renders might require a different approach. It's a shame that everything posted at both SmithMicro and RuntimeDNA has completely disappeared from the internet.

hborre posted at 11:29AM Fri, 07 May 2021 - #4418649

I suggest that you Google bagginsbill and glass see what comes up. He has done some extensive work on glass shaders for both Firefly and Superfly with analysis comments. Given the nature of this topic, I wouldn't be surprised if he chimes in. Now a one-fits-all glass shader might be a tall order to fill unless we're talking background objects as prop fillers to complete a scene. But using glass objects in fore- and mid-ground renders might require a different approach. It's a shame that everything posted at both SmithMicro and RuntimeDNA has completely disappeared from the internet.

I did try to search here on the forum, but as you said nothing showed :( But will try to give the big internet a search :D

A few months ago I made refractive optical glasses rendered with Superfly and had a similar problem - the parts behind the glasses appeared darker

I think I finally solved most of it by increasing the bounces, and omiiting IBL lights

You have twice the bounces than you tend to think, cause the ray has to go through both faces and back again through both faces, at least.

Thats at least 5 bounces I think.

Try this one. Similar to the post above, but uses some "LightPath" parameters instead of Normales:

And make realy sure ALL Color is realy white (255, 255, 255). If not you may get those "shadows" :)

If someone is looking for shaders, seek for "Blender Cycles". Most of those shaders can be rewritten for Poser (11 and 12).

As posted above, the number of bounces is very important if you want your glass to appear as accurate as possible.

hborre posted at 1:39PM Fri, 07 May 2021 - #4418665

As posted above, the number of bounces is very important if you want your glass to appear as accurate as possible.Yes! But the goog news is, Superfly renders this surprisingly fast even with a lot of bounces.

Thanks for all the suggestions, have tried to compare three different ones I could find. This is turning into some study of wine glasses :D

1. Is the one that Adp001 suggested and the setup can be seen above.
2. Is mine, is slightly altered later on so will post the new one in the end.
3. Is one I found through the internet from a former post. Which uses this setup:

So the initial render looks like this.

So they give slightly different results. Next I tried to render them on a white background to see how they would react, and given Im not a 100% sure how they are suppose to look, I don't think any of them look particular as one would expect, there are good and bad things about each of them I think.

At this point I tried to mix (2) with (3) which looked like this.

And eventually a new render looked like this.

And again I like certain things about all of them, I like the clarity of nr 1, but something is off with the glass, that makes it look more like what you would expect from a glass sphere like this

But also you would expect to be able to look inside the glass and see the bottom, which doesn't really seem to be the case, so maybe its to reflective or something? This is an image of some real wine glass.

Nr 3, I think looks a bit to metallic especially the foot. Whereas in nr 2, I would like to get some of clarity from the foot of nr 1. So its a mixed bag of some from all of them :D

Here is the final material setup from mine with it being mixed between 2 and 3.

So what do people think, which one do you like the most and can we improve them even more?

Here's bagginsbill's glass

Things to keep in mind, make sure you have enough transparency bounces. make sure the refractive caustics is checked. and if you're working with filled wine glasses make sure the faces of the wine aren't in the same location as the faces of the inside of the glass.

RedPhantom posted at 5:01PM Fri, 07 May 2021 - #4418683

Here's bagginsbill's glass

Things to keep in mind, make sure you have enough transparency bounces. make sure the refractive caustics is checked. and if you're working with filled wine glasses make sure the faces of the wine aren't in the same location as the faces of the inside of the glass. Thanks for that, I have set the maximum bounces to 64 so shouldn't be an issue.

The glass on the far right is with Baggins setup from that thread. As you can see it seems to suffer from the shadow issue as well, so maybe it was meant for a different type of object? like single surface one?

One Problem with your render is: Those glasses are too thick. That's the reason while your photo looks so different ("real"). Cycles (and the shaders) are pysical based. The thinner the mesh of your glasses, the more real the rendered image will be.

adp001 posted at 6:30PM Fri, 07 May 2021 - #4418693

One Problem with your render is: Those glasses are too thick. That's the reason while your photo looks so different ("real"). Cycles (and the shaders) are pysical based. The thinner the mesh of your glasses, the more real the rendered image will be.

I actually thought about that as well and also made another glass as a test which is thinner, so will replace the one in the image later on at some point. :)

No, bagginsbill's shader is for glass with thickness. It doesn't look good for single-sided objects. That shader had an absorption volume of cyan. try changing it to white.

RedPhantom posted at 6:51PM Fri, 07 May 2021 - #4418695

No, bagginsbill's shader is for glass with thickness. It doesn't look good for single-sided objects. That shader had an absorption volume of cyan. try changing it to white.

I tried messing around with it, but couldn't make it work, I could make it transparent but still didn't look like I wanted it. And again, Im not good at cycles so really have no idea what im changing and why :D I know he is very good at it and have his reasons for why he does it, but im not even remotely there so have to rely on what looks correct to me in the renders :D

Anyway I manage to combine a final shader, and also tried adding some red wine as well, which caused other issues especially with the reflection, but think I got that solved as well.

And honestly I think it looks pretty damn good as a glass shader, it doesn't have these crazy reflections, they are nice and clear so you can see through them and pretty nice highlights, so unless someone knows how to improve them even further I think ill go with this one as a default shader for glass. And when they get a bit thinner I think it will be excellent :)

Here is the final setup for those that might be interested in it as well.

I'm loving this glass thread. Er, that's all. Thanks.

Why not try Physical Root node instead of Cycles Root node ...

infinity10 posted at 4:50AM Sat, 08 May 2021 - #4418702

Why not try Physical Root node instead of Cycles Root node ...

Because im pretty sure I saw a video at some point stating that this functionality is the only one lacking from the physical root, it has transparency and opacity, but no real way of dealing with glass and probably not water either. I think you can use it to make quick fake glass as long as it is thin by messing around with the transparency settings.

I actually did a test of this in the thread where I got the (3) glass setup above. Here you can see the difference between physical node glass and just a standard cycle node glass. But the physical one doesn't work well with thick glass as can be seen in the last image.

And obviously this thread is all about fixing the standard cycle glassBsdf, and to solve the shadow issues.

VedaDalsette posted at 5:05AM Sat, 08 May 2021 - #4418701

I'm loving this glass thread. Er, that's all. Thanks.

Yeah its fun :) But also extremely useful, If this shader manage to hold up when I test it some more I different settings, that is a huge obstacle that is removed from me at least, because always have to spend a lot of time on this, given that im not really good at cycles. so very nice to have a basic shader to use.

3D-Mobster posted at 5:19AM Sat, 08 May 2021 - #4418703

infinity10 posted at 4:50AM Sat, 08 May 2021 - #4418702

Why not try Physical Root node instead of Cycles Root node ...

Because im pretty sure I saw a video at some point stating that this functionality is the only one lacking from the physical root, it has transparency and opacity, but no real way of dealing with glass and probably not water either. I think you can use it to make quick fake glass as long as it is thin by messing around with the transparency settings.

Oh. Interesting. Very interesting.

My 2 cents. apd001 is correct. Both in shader and glass thickness observation. I use a Cycles shader similar to his/hers with the addition of a Fresnel calculation so I can use it on anything from water to crystal in addition to glass. The Glossy roughness controls how frosted the glass surface is and is the place to add maps for etching and the like. Adjusting the Transparency colour will change the colour of the glass itself so as to make water and other liquids thus the Fresnal.

I believe IOR in PBR is only indirectly changable via the setting the value of the specular channel (F0 calc) and it does not support refraction.

Could caustics be an issue here, I know there is a button to tick but does it do anything.

Edit: Cross post with parkdalegardener

parkdalegardener posted at 6:24AM Sat, 08 May 2021 - #4418706

My 2 cents. apd001 is correct. Both in shader and glass thickness observation. I use a Cycles shader similar to his/hers with the addition of a Fresnel calculation so I can use it on anything from water to crystal in addition to glass. The Glossy roughness controls how frosted the glass surface is and is the place to add maps for etching and the like. Adjusting the Transparency colour will change the colour of the glass itself so as to make water and other liquids thus the Fresnal.

Ill give it another test on a new thinner glass, lets get to the bottom of this glass shader :D

So this is another test just with different shapes.

The first image the cube she is sitting on is using the standard GlassBsdf straight up, which gives some hard reflections.

This one it uses my material, which give some more soft reflections.

I also looked up some references for glass spheres and think it does pretty well, I think my glass is very clear compared to those in the image and also not sure how much photoshop went into these, but overall I think it does well, if some roughness is added, a more intensive light is shot on them from the back.

Very useful thread!

Found something at blenderguru.com under the Ray Depth section. So I compared 3 glass figures, each with their own balls placed in the backdrop pink square. Fig 1 is the blenderguru glass, fig 2 is the parkdalegardener glass listed earlier in this thread, and fig 3 is the adp001 glass listed earlier in this thread. I rendered Superfly at Low Adaptive GPU. Skydome light has hdri mountain image and light is Poser's Superfly Loop. For the balls, I turned off Cast Shadows.

Here's the blenderguru setup.

So then I tested it on thin boxes (to the right of the mannequin). Box at top is 1% z scale and box below is 5% z scale. Cast shadows of boxes is off.

I'm no artist, so I'm not a good judge, but I kinda like the blenderguru glass using ray depth. (I forgot to test against 3D Mobster's glass!)

VedaDalsette posted at 8:22AM Sat, 08 May 2021 - #4418713

I'm no artist, so I'm not a good judge, but I kinda like the blenderguru glass using ray depth. (I forgot to test against 3D Mobster's glass!)

I use transparent depth in mine, but again I just plugged it in there because it sounded useful :D

I really like the first glass (The one on the left) the middle one doesn't look correct to me and the third might be fine for some thin glass. The last image looks nice I think. Will test that as well on the new glass I made. But could be because there is not a lot of things for the glass to react to. Like the test I did with the white background.

Compared blenderguru glass with 3D-Mobster's glass. 3D's is better. Maybe for some thick glass blenderguru's could be better.

VedaDalsette posted at 8:47AM Sat, 08 May 2021 - #4418715

Compared blenderguru glass with 3D-Mobster's glass. 3D's is better. Maybe for some thick glass blenderguru's could be better.

Thanks for testing it then I don't have to :D And agree, both seems very useful depending on the glass you are fooling around with.

So I made a new ridiculous thinner glass. :)

And it is probably slightly to thin in my opinion, but should be fine I think. Also it is extremely highres, each glass is 83000 polys, compared to all of V4 which is 66800 poly, but wanted to be sure that there was no mesh imperfections at all.

This is with my material:

I think it looks fine, but it probably have to be more clear to really pass as a clean glass. I tried using Apd001 shader again, and there is something that just look wrong.

I took a wine glass at home to compare with how it reacted when being in front of objects, light etc. And there are some things in this material that just doesn't work correctly in my opinion.

First of all the warping of the olive oil flask is way to extreme, this is not going to happen with a thin piece of glass, there is hardly any warping at all, unless you "move" towards the edge of the glass.

The next issue is the reflection in the glass, where the coffee box is clearly seen in the glass, that is not going to happen as far as I can see, the reflection in a wine glass is almost invisible.

You do seem to get some nice clean glass when looking directly through it. But only when looking through one side. The moment it is both sides, the reflection are way to hard compared to what you would expect.

The last issue as I see it is the foot of the glass which becomes almost invisible. If you place a wine glass on a table you can still clearly see the glass. But here it becomes almost invisible.

To me, it just look wrong, but maybe if the reflections and warping is somehow reduced, it could work.

Just want to add, that these issues are also in mine but are not as severe as in Apd001 shader, so probably have to do a bit of tweaking with the reflections I think,

Well to make life completely easy; the Principled BSDF node has got some love recently it seems. One shader to rule them all. Poser's morphing cylander props and the ball. Thick glass, thin glass; and solid glass (the ball) all from the same shader.

Ignore the Firefly root. It only appears because I stripped out the Posersurface to make it easier too see the material. This isn't a Firefly shader.

parkdalegardener posted at 10:16AM Sat, 08 May 2021 - #4418723

Well to make life completely easy; the Principled BSDF node has got some love recently it seems. One shader to rule them all. Poser's morphing cylander props and the ball. Thick glass, thin glass; and solid glass (the ball) all from the same shader.

Yeah I saw that in blender as well, its really cool :D

I'm glassy-eyed over all this beautiful glass. Now, on to diamonds!

Change the IOR and it will give you a start. I'm not at the machine that I used earlier. No Poser on it. https://www.robinwood.com/Catalog/Technical/Gen3DTuts/Gen3DPages/RefractionIndexList.html

This is will be the last one, and my final glass shader, which I think works fine for my needs at least :D

So I manage to make them more clear and also add a nice more defined edge to them, which I like, especially on the foot.

Seems to also work with the other glasses as well, which is very nice.

And as a final test I tried turning off the light and load in a 16k Hdri with different light settings. And think it looks fine the glass seems to behave as expected, so very happy for that :D

The final shader can be seen here, and yes have absolutely no idea what is going on in it :D

Gotta copy this down fast!!!! But I won't be using any 80K polygon glasses, hehe!

And, parkdalegardener, thanks for the refraction site! I'll be copying your simple but elegant principledbsdf glass, too.

I can't type special characters into the forum software so look at the Fresnal calculation here https://docs.blender.org/manual/en/latest/render/shader_nodes/shader/principled.html to get the numbers for diamond. One thing about this though. Diamonds have dispersion and total internal reflections at certain angles. This material works more like the little speckles we see if you fake dispersion by adding each RGB channel together at a slightly different IOR and then mix the whole thing together with a clear glass material. That will provide the little coloured refraction speckles you expect to see. Search for BagginsBill's thread on diamonds and dispersion here on Rendo.

Thanks so much, parkdalegardener! I'll have to play with this.

on edit: Found it! https://www.renderosity.com/rr/mod/forumpro/?thread_id=2950430&#msg4395920

VedaDalsette posted at 9:36PM Sun, 09 May 2021 - #4418728

I'm glassy-eyed over all this beautiful glass. Now, on to diamonds!

I copied a shader someone made from Blender, which looks like this:

With some decent diamond models and not these quick ones I made :) I think it will work very well.

Link to the blender video:

Blender Quick Tip #4 | Diamonds in Blender

Just remember to increase number of bounces, refraction etc in the superfly settings.

Sorry; I forgot about this. Dispersion for diamond is .044 and I don't understand the multiply node in the shader above, but the idea is basically correct. Calculate the dispersion through the stone and enclose it in a glossy or glass. You can use refract nodes for R,G,and B instead of glass and get the same thing. That is the way it was done by BB in the old firefly threads that are now gone it appears.

The dispersion in a crystal is the change in IOR between the red and blue ends of the spectrum when white light passes through it. All crystal has it but it is only diamonds where we really notice it and that is because diamonds are specifically cut so that the dispersion becomes total internal reflection at the base of the stone. It is the separately refracted coloured light rays reflecting around inside the stone that provides the coloured sparkles. Make your shader with refract nodes and you will will have a firefly shader that provides "fire" in your stone as well.

Here is the refract shader. Essentially the same concept as the glass node version above but transportable to Firefly with minimal changes. I did the math already so the IOR and Dispersion constants are not required. This makes it a little easier to read.

parkdalegardener posted at 7:07AM Mon, 10 May 2021 - #4418888

Sorry; I forgot about this. Dispersion for diamond is .044 and I don't understand the multiply node in the shader above, but the idea is basically correct. Calculate the dispersion through the stone and enclose it in a glossy or glass. You can use refract nodes for R,G,and B instead of glass and get the same thing. That is the way it was done by BB in the old firefly threads that are now gone it appears.

The dispersion in a crystal is the change in IOR between the red and blue ends of the spectrum when white light passes through it. All crystal has it but it is only diamonds where we really notice it and that is because diamonds are specifically cut so that the dispersion becomes total internal reflection at the base of the stone. It is the separately refracted coloured light rays reflecting around inside the stone that provides the coloured sparkles. Make your shader with refract nodes and you will will have a firefly shader that provides "fire" in your stone as well.

Don't know why there is a multiplier in it either, I think it has something to do with changing the IOR automatically for the blue. But then again, I have no clue how people even figure out how to combine the nodes correctly in the first place. But happy that some people does and the rest of us can make use of them :D

parkdalegardener posted at 7:11AM Mon, 10 May 2021 - #4418890

Here is the refract shader. Essentially the same concept as the glass node version above but transportable to Firefly with minimal changes. I did the math already so the IOR and Dispersion constants are not required. This makes it a little easier to read.

I think the guy added the dispersion value and the IOR so you can easily change them one place.

Thanks, parkdalegardener!!! Now, I have another beautiful diamond mat.

Bookmarking. Very interesting thread.

VedaDalsette turn on caustics when you render in order to get the "fire" in the stone. Give me a second to toss a quick render on the other machine.

OK. This machine has Poser. This is just quick. 4k HDRI and a single spot because it's a work light and I forgot to turn it off.

Crap. jpg conversion took most of the "fire" out but you get the idea.

**parkdalegardener, thanks for the tip! ** I never even gave caustics a thought (I'm not the sharpest knife in the drawer).

The medium adaptive gpu render was going so fast, I decided to invest the five minutes in a high adaptive gpu render, and I used a colorful hdrihaven environment for light.

Poser 12 with denoising