Page 1 of 1

Rough/frosted glass

Posted: Wed May 16, 2018 9:58 am
by DavidF
Hi,

I think I have similar questions as forgeflow asked many years ago on the LuxRender forums (http://www.luxrender.net/forum/viewtopi ... =16&t=3252 ), i.e. why do objects using the rough glass shader appear darker when the roughness is increased?

Basically I am trying to create a material resembling frosted glass and I have done so by enabling the "Rough" option of the glass node. I expected that a higher roughness would generate a brighter surface (e.g. https://goo.gl/images/qEZvhD ), but instead the objects become dark. Any ideas why this happens or how I could create a better frosted glass material? I have attached the blend file I've been using for these tests.

Thanks in advance,
David

Roughness 0.02
Frosted_glass_roughness0.02.png
Roughness 0.2
Frosted_glass_roughness0.2.png

Re: Rough/frosted glass

Posted: Wed May 16, 2018 10:23 am
by Dade
DavidF wrote: Wed May 16, 2018 9:58 am I think I have similar questions as forgeflow asked many years ago on the LuxRender forums (http://www.luxrender.net/forum/viewtopi ... =16&t=3252 ), i.e. why do objects using the rough glass shader appear darker when the roughness is increased?
I have the feeling it may be due to an increase of internal reflections, have you tried to increase the max. path depths ? I will do a test.

Re: Rough/frosted glass

Posted: Wed May 16, 2018 10:56 am
by Dade
Ok, few notes about your scene:

- you are "1 1 1" values for the ground color, it is very unrealistic, you should use something more like "0.7 0.7 0.7" to physically plausible.

- you have a sky light source with gain "0.0 0.0 0.0" (outside a closed box). It is a noticeable waste rendering time.

- your max. depth setting are:

Code: Select all

path.pathdepth.total = 7
path.pathdepth.diffuse = 5
path.pathdepth.specular = 6
6 is definitively to low for glass rendering. You should at least double the values (including "path.pathdepth.total").

However I think there is the same problem I fixed a while ago for (non-rough) glass. Try to switch between roughglass and normal glass; with very little roughness, they should look the same however rough glass is a LOT darker. I will check the code.

Re: Rough/frosted glass

Posted: Wed May 16, 2018 11:19 am
by FiatLux
DavidF wrote: Wed May 16, 2018 9:58 amAny ideas why this happens?
It could be because the glass model.
This is the multi-scattering glass model that would fix the situation.

Re: Rough/frosted glass

Posted: Thu May 17, 2018 2:25 pm
by kintuX
Frosted_Glass_DENOISED.png

Here i made Glass-Volume Homogeneous Volume with Multiscattering on & Upped the Glass Material Roughness to 0.2.
Also, made ground Matte 0.76 White Value, rendered with BiDir (5000samples), denoiser used & added a bit of tonne mapping for better contrast.
Glass-Volume.jpg
Glass-Material.jpg
Frosted_Glass-V.zip
(649.16 KiB) Downloaded 271 times

Re: Rough/frosted glass

Posted: Tue May 22, 2018 6:45 am
by DavidF
Thanks for your comments!
kintuX wrote: Thu May 17, 2018 2:25 pm Here i made Glass-Volume Homogeneous Volume with Multiscattering on & Upped the Glass Material Roughness to 0.2.
Also, made ground Matte 0.76 White Value, rendered with BiDir (5000samples), denoiser used & added a bit of tonne mapping for better contrast.
How does the homogeneous volume with multiscattering work, doesn't that introduce scattering within the volume while frosted glass is mostly a clear volume with a microfaceted surface?
FiatLux wrote: Wed May 16, 2018 11:19 am
DavidF wrote: Wed May 16, 2018 9:58 amAny ideas why this happens?
It could be because the glass model.
This is the multi-scattering glass model that would fix the situation.
That's interesting. How does the current LuxCoreRender glass model compare to the models discussed in that work, i.e. the Smith microsurface model and the improved model including the multiscattering component?