the point is a basic setup within Cycles for realistic glass:
edit:
specular roughness
transmission roughness
colour
three most basic characteristics of glass.
I am not sure there is such thing. It depends. There is no single definition of what glass is in this context. It’s not even a single material. there are glass recipes. It’s like wishing to have a single basic setup for a cake. Very similar situation. all sorts of forms, all sorts of ingredients, all sorts of looks.
The expression of glass is much trickier with the blender than with the V-Ray.
There are basic settings, but to make the glass you want, you have to search a lot of data and you’ll need a lot of ideas.
It seems that the current results are being structured similarly.
What do you think is lacking?
Try using a node to fill in the gaps.
So far the first solution looks like it works better, the one with BDSF and volume absorption, i just need to tone down emission.
It may be better to make the amount of the Emission emission visually desired and install the lights.
Alternatively, you can set the actual emission and the amount of emission that will be shown in the rendering differently.
It would be hard to equalize the actual amount of emission to the amount of emission you want to show in the rendering. It’s always either strong or weak on one side.
Without knowing the size or geometry (glass thickness in Blender) of the model, it’s complicated.
If you know the thickness (in mm) of the glass:
ln(Min(Color,0.999))/(-thickness) = Absorption coefficient in mm-1
Multiply the result by 1000 to feed the volume node
If it seems too dark, it is because the color is passing through two glass walls: take the square root of the color
Suzanne, Solidify 1 cm (10mm)
For the logarithm node, in the “base” input, simply type e instead of the number.
…or just put the light in your scene and then visually adjust the values until they match the reference. It’s helpful to have the reference always visible with something like PureRef. The problem is that whatever reference you have will most likely be in different environment so you have to take that into consideration and imagine how it might look in your scene and with your lighting. You generally cannot atchieve good results only trusting some numbers and you certainly cannot get the numbers for various products - there is simply no information like that anywhere for the specific products you may encounter in real life. Best to forget number altogether if you ask me. Of course keeping them in plausible ranges is a good idea.
for reference - this is 1000 lumens “lightbulb” in Cycles:
2.062 is matching VRay, Blender’s docs claim 2.9 - pick any number between the two and stick with it. pbr shaders expect pbr lights - just to make everything simple.
Tinted reflections are often seen at facing angles on glass with certain coatings. A green tint due to anti reflection coating on regular glasses. I haven’t been able to mimic the effect properly using thin film, so I’m staying with fresnel mixed refraction and glossy, with a custom way to handle shadows.
I did just that, and works a bit better, but when i get to sertain angle glass gets super weird still, and losses reflection, idk whats happening, but once you put that glass in interior context it gets all wierd. Looks like this theme is lot more complicated than i thought
will try this one too, thank you
exactly what happens with my case, looks like environment itself tweaked look of the glass, and makes some wierd looking results.
I think that handles internal reflections differently than glass shader(I mean it doesn’t handle them at all really 
). Refraction and glossy mix might not work well in some cases.
Sometimes the glass may have a color but specular reflections are white, sometimes specular reflections may have color and it’s a weird mix, like those metallic kinds of glass:
In that case just mixing with glossy shader might work:
But I am not really sure how to just make glass that is completely clear and has colored specular reflections to be honest… On the other hand, I am not sure I have seen such glass in reality, they seem to be tinted in that case. I think in those cases the reflection color is coming from non-transparent additives so mixing with another shader should be right. Coating on glasses is different I think.
Somebody is working on units for lights in Blender. We might get them some time in the future hopefully. At the moment it’s quite complex to set up lights to match lumen values in Blender. It depends on the color of the light as well, but even manufacturers of light sources don’t bother with that and measure lumens in specific wavelength(green I think 
)… or just any way they happen to feel like that day(inaccurately). If you wanted to set Blender’s point light to match some lumen values, one naive approach would be to divide your lumen value by 683 and you are good to go(spoiler alert: not really). So that would be around 1.464 for point light to match 1000 lumens if it’s white(but it’s a rough approximation, because light spectrum would actually need to be considered for something less naive). Efficacy should probably be calculated for RGB per channel taking working color space into account especially if you want to have colored lights, which of course you do, because you often want warm or cold light sources not just white… Then there is the problem of all the light types in Blender working differently. For spot lights their strength is the same as point lights, but the output is limited by their cone, so you enter a value as if it was a point light, but the actual amount of light going to your scene is absolutely not that… and we then have mesh lights, that would require taking surface area into account if you want to match them to lumens not like regular area lights, I don’t even remember what you have to do with those…
I think it’s better to wait until it’s actually finally implemented.
Not really… Any light is PBR… Strengths start to matter when you want plausible ratios for different light sources. Other way, just adjust the exposure as needed and it doesn’t interfere with any PBR ideas…
I’m glad I looked at this post because I had no idea glass had so much variety until I did.
Yeah… It’s as if they are really trying to mess with us. I imagine a light fixture designer starting a day in the office with quiet contemplation: “OK… Let’s see now… How could I ruin a 3d artist’s day today?..”
I am joking off course, it’s nice… Some of them are really beautiful and unique, but for us archvis people this means a problem with no single solution.
anything to support the 1.464 number?
Not much. I might be incorrect here.
The way I understand you need to get efficacy to calculate that, so that leads me here:
Efficacy is different for different light sources, but the way I understand Blender lights deal only with visible light measured in radiometric units, which is not something one does usually the way I understand and the values in the table there are for total radiation, not for visible light part of that radiation of the sources there and so they can probably be ignored and only the ideal one is probably OK. …sort of. It’s still only for green light, but the way I understand that’s how light sources are measured anyway and that’s what we see on light bulbs. Plenty of space to be wrong somewhere there though.
This is clearly off-topic here though. Maybe splitting the thread to a separate one about lighting would be a good idea if we want to discuss that more. The guy working on light units for Blender also has some sources, maybe looking into them might be useful. I haven’t though.
Yeah, he seems to be using the same value in his code. The funny thing is: I don’t think this really confirms that it’s correct beyond any doubt 
…and he doesn’t really take working color space into account. I wonder how that works. If it’s an approximation, maybe it doesn’t matter and maybe that’s why there is a task for custom efficacy function. Which would probably make sense, because if you are using a value that is written on a light bulb, then probably it’s approximate anyway, if you know the spectrum of the light for some reason you can then maybe somehow use custom efficacy function?.. But then why are you using RGB renderer why does that matter at all and what do you expect?.. This is all very confusing.