PS and finally, if this is a "physically correct" shader, why should I add a layer of transparent glossy varnish to the glass if it is not there?
Technically fingerprint marks are not the part of the glass, it's a layer of grease and dirt on top of the glass. Maybe using clearcoat is not the best solution, but physically it's more correct than making fingerprints as a property of the glass.
Absolutely, but i'm talking about using clearcoat roughness parameter instead of clearcoat amount
I think the map should go into clearcoat roughness (inverted), not clearcoat amount. Clearcoat amount needs to be >0 then, of course.
in this case it is absolutely obvious that the glass surface without fingerprints is not covered with any clearcoat at all
Some differences in roughness are to be expected, the underlying models in Physical material are different than in the legacy material. Diffuse reflection is different based on base glossiness/roughness, reflection is normalized (scaled in order to not lose energy) differently to be more physically plausible.
Comparison 1, HDRI lighting:
https://corona-renderer.com/comparer/NugVgkComparison 2, two light sources:
https://corona-renderer.com/comparer/Uw9dpNIn my opinion, I find the physical model to be more accurate, but I can see how this might need some effort into being get used to, as always if a lot of voices raise concern we will forward your criticism to the devs for further consideration.
(Report ID=CRMAX-71)