Glass is getting too dark...

Use this forum for general user support and related questions.
Forum rules
Please upload a testscene that allows developers to reproduce the problem, and attach some images.
kintuX
Posts: 809
Joined: Wed Jan 10, 2018 2:37 am

Glass is getting too dark...

Post by kintuX »

Dispersion example (0-1 in 50 frames):
The attachment dispersion_LC-21a3-BiDir.gif is no longer available

Roughness example (0.01-1 in 50 frames):
roughness_LC-21a3-BiDir.gif
roughness_LC-21a3-BiDir.gif (2.57 MiB) Viewed 6833 times
To me it simply feels as too dark and I think (IIRC) that glass had much better behavior in previous LuxRender 1.7 (classic&core) version, at least it wasn't loosing so much energy. :? Any ground truth examples around?
Attachments
cornell_LuxCore2.blend
(1.08 MiB) Downloaded 214 times
User avatar
Sprocket
Posts: 20
Joined: Fri Mar 09, 2018 12:31 pm

Re: Glass is getting too dark...

Post by Sprocket »

I changed material in your example file a bit...
Capture.PNG
User avatar
lacilaci
Donor
Donor
Posts: 1969
Joined: Fri May 04, 2018 5:16 am

Re: Glass is getting too dark...

Post by lacilaci »

Please correct me if I'm wrong, but I think this is why glass needs multiscatter ggx.
https://blender.stackexchange.com/quest ... catter-ggx
User avatar
Sprocket
Posts: 20
Joined: Fri Mar 09, 2018 12:31 pm

Re: Glass is getting too dark...

Post by Sprocket »

Also the issue with clear glass might just be the values that you are using. The tool tip for dispersion says:
"Realistic values range from 0.00354 to 0.01342"
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Glass is getting too dark...

Post by Dade »

Sprocket wrote: Thu Aug 16, 2018 8:48 am Also the issue with clear glass might just be the values that you are using. The tool tip for dispersion says:
"Realistic values range from 0.00354 to 0.01342"
Yes, you go out of visible spectrum out of an healthy range.
Support LuxCoreRender project with salts and bounties
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Glass is getting too dark...

Post by Dade »

Dade wrote: Thu Aug 16, 2018 9:06 am
Sprocket wrote: Thu Aug 16, 2018 8:48 am Also the issue with clear glass might just be the values that you are using. The tool tip for dispersion says:
"Realistic values range from 0.00354 to 0.01342"
Yes, you go out of visible spectrum out of an healthy range.
See https://en.wikipedia.org/wiki/Cauchy%27s_equation (upper right image) for values on visible spectrum.
Support LuxCoreRender project with salts and bounties
kintuX
Posts: 809
Joined: Wed Jan 10, 2018 2:37 am

Re: Glass is getting too dark...

Post by kintuX »

Sprocket wrote: Thu Aug 16, 2018 5:12 am I changed material in your example file a bit...
Capture.PNG
Yes, i know.
The scene shared is set as i left it, after few more experiments to brighten the glass... but got tired and also ran out of time (am slowly recovering from an accident).
Sprocket wrote: Thu Aug 16, 2018 8:48 am Also the issue with clear glass might just be the values that you are using. The tool tip for dispersion says:
"Realistic values range from 0.00354 to 0.01342"
Dade wrote: Thu Aug 16, 2018 9:06 am Yes, you go out of visible spectrum out of an healthy range.
Dade wrote: Thu Aug 16, 2018 9:08 am See https://en.wikipedia.org/wiki/Cauchy%27s_equation (upper right image) for values on visible spectrum.
Thanks, all that got me thinking again...

Roughness (0-1 in 50 frames) with Cycles 'MultiscatterGGX' Glass
CyclesMGGX_GlassRoughness.gif
Using "Multiple-Scattering Microfacet BSDFs with the Smith Model" in Cycles is pleasant, comfortable, feels very natural & got me really spoiled :mrgreen:

So, especially in cases of physically accurate light transport, IMHO, this much darkening shouldn't happen. M-GGX is clear proof of that and I also believe similar would apply for dispersion - even when using extreme values, light transport within medium should go without such strong loss of energy (Diamond is also very nice example). Here's where artists add to scientific & technical limits.

Anyways, if anyone knows how this kind of result can be accomplished with LuxCore2 would be really grateful. Also, since there's no additive material layering it's really hard to get nice frosted glass, milky glass, snow, gems & precious stones...
Experimenting & exploring further.

And Thank you all again for your help and explanations 8-)
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Glass is getting too dark...

Post by Dade »

kintuX wrote: Thu Aug 16, 2018 1:52 pm
Dade wrote: Thu Aug 16, 2018 9:06 am Yes, you go out of visible spectrum out of an healthy range.
Dade wrote: Thu Aug 16, 2018 9:08 am See https://en.wikipedia.org/wiki/Cauchy%27s_equation (upper right image) for values on visible spectrum.
Thanks, all that got me thinking again...

Roughness (0-1 in 50 frames) with Cycles 'MultiscatterGGX' Glass
CyclesMGGX_GlassRoughness.gif
We were talking of the first image, of dispersion and Cauchy's equation: how can be Cycles used as reference if it doesn't support dispersion at all :?:

P.S. Cycles is also not rendering any caustic at all in your image.
Support LuxCoreRender project with salts and bounties
kintuX
Posts: 809
Joined: Wed Jan 10, 2018 2:37 am

Re: Glass is getting too dark...

Post by kintuX »

Dade wrote: Thu Aug 16, 2018 3:00 pm We were talking of the first image, of dispersion and Cauchy's equation: how can be Cycles used as reference if it doesn't support dispersion at all :?:
Can simulate dispersion in Cycles with shader nodes, separating RGB chanels using different IORs and adding, summing it back.

& Here are examples with Cauchy @ 1 using Classic Lux BiDir.
Clear Glass
ClassicLux_C1-clearGlass.jpg
Rough Glass (U-R=0.01)
ClassicLux_C1-roughGlass.jpg
:D it become an emitter

I know engine is spectral & values unreal, i am ignorant in detailed science and math behind, but from observing matter/medium in nature, this results still feel more likely to be realistic than results with present engine. As for Cycles matter, i'm also able to accomplish similar results easier and faster, while differences are almost never noticeable to observer's eye.
Dade wrote: Thu Aug 16, 2018 3:00 pm P.S. Cycles is also not rendering any caustic at all in your image.
Had disabled caustics (both/ reflective & refractive) for the sake of clearer image, since second example is mainly about roughness and multi-scattering. With caustics on, i assume, rough glass would just be a fraction brighter since all rays get scattered thus caustic effect can't even be produced.
User avatar
Dade
Developer
Developer
Posts: 5672
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Glass is getting too dark...

Post by Dade »

kintuX wrote: Thu Aug 16, 2018 4:27 pm
Dade wrote: Thu Aug 16, 2018 3:00 pm We were talking of the first image, of dispersion and Cauchy's equation: how can be Cycles used as reference if it doesn't support dispersion at all :?:
Can simulate dispersion in Cycles with shader nodes, separating RGB chanels using different IORs and adding, summing it back.
That is not even vaguely dispersion.
Support LuxCoreRender project with salts and bounties
Post Reply