glass thickness

[momoslav]

I try two types of glass - simple plane and cube. (same material - basic glass from corona site, but "thin" option check with plane)
Which one is correct from physics point of view. Thanks.

[FrostKiwi]

"from physics point of view"
Well, obviously perfect copy of the real world glass (two panes of cubes with thikness) with thin not checked. Because in real life everything has a thickness.
But the difference is so small, that for windows dev's included a non physically correct version specifically for windows. You just make a simple plane with thin checked, then you get no refratction calculation, which especially at a distance can be ignored and thus speeds up render times.

What some archviz ppl do is make double thin planes for the double reflection of real world windows, when camera is near the windows.

More infos here on the corona help desk "what glass type should I use" post.
https://coronarenderer.freshdesk.com/support/solutions/articles/5000501661-what-type-of-glass-should-i-use-

[maru]

"from physics point of view"
Well, obviously perfect copy of the real world glass (two panes of cubes with thikness) with thin not checked.

And caustics. And absorption. ;)

Another useful source:

[kv3t1n4c]

I experienced pretty big differences between these two approaches, ofc one of them is render time, the next is strenght of reflections. It seems that 2x single plane is lacking some contrast compared to 2x box. See attached images.

[steyin]

I always model windows with glass as it is actually constructed, normally two 1/4" panes with a 1/2" air gap, then use the non thin wall material. In this day and age of processor speed/memory availability and the continuing advancement of render engines I find there is very little to no speed difference between the two methods.