Rough Specular reduces specularity amount?

[instinct]

Hm, where is it?

I THINK it is part of the BRDF Viewer sources: https://github.com/wdas/brdf/tree/master/src/brdfs

[vlado]

Do you think it's the full one and not just some approximation for GLSL? It's missing stuff like refraction, SSS, emission.

Best regards,
Vlado

[instinct]

Do you think it's the full one and not just some approximation for GLSL? It's missing stuff like refraction, SSS, emission.

Best regards,
Vlado

I have no idea, but that's of course very well possible. The paper also seems to be a bit short if you ask me. It looks like it's requiring quite some knowledge of the previous work like ggx etc.
Then i again i have no real clue about papers and turning them into working software so i may not be a good guinea pig for that heh. I just stumbled over the brdf viewer a while back but have not really looked closer.

Cheers,
Thorsten

[sharktacos]

Do you think it's the full one and not just some approximation for GLSL? It's missing stuff like refraction, SSS, emission.

Best regards,
Vlado

I believe the GLSL on Github is the first DisneyBRDF from 2012. That was later extended to add things like SSS and translucency.

[^Lele^]

Nice find, the source code.
And yes, thorsten, the shader's as poor as the description in the paper ("// 1.25 scale is used to (roughly) preserve albedo" isn't an unfortunate accident, but rather the modus operandi behind this shader construction.)
And for all that, a few effects are just plused on top, regardless of leftover energy in what precedes them (quite easy to test, in fact, with rMan NC).
rMan 21 MAY in fact have a newer version of an ubershader to account for refraction and SSS, and perchance EP, but it's not called "Disney" anymore, but "Pixar Surface", so who knows if it's the same extended version as the one linked.
Regardless, rMan 21 matures their layered shader system from daisy-chained to layer collector based, which will perhaps aid them in maintaining EP at the top.
Since i ain't going to buy them a license, however, i'll have to wait until the NC version comes out "later this fall".

On the extended model, three things jumped to my eye:
1) the same read of an ideal unified sampler for light-matter interaction: "The most general model perhaps would be a specularBSDF for the surface combined with a volumetric scatteringmodel for the interior [...] Such a model can reproduce all physical effects, though it is not always practical to do so; e.g. in this model surface color is derived solely from absorption during volume scattering though it would be sufficient and obviously more efficient to use a diffuse BRDF when the scatter distance is negligible.Arguably, one might consider the BSSRDF as the most general unified representation of scattering since, mathematically at least, it fully describes all scattering between any two points and directions on a surface"
It's interesting to read of the way they found to sidestep the searchlight issue.
2) What they call "Furnace" test, is technically called Ganzfeld Illumination (total-field illumination) and is used (as approximately as humanly possible) in real world optics to study surface properties quantitatively. It's what i use to measure shader: a light field with a constant, uniform radiance of 1.0 (i wrote in the other post: "No GanzFeld, no Party", for a reason.), or your default domelight.
3) They confirm that retroreflection is measurable in real materials, and their BRDFs fail to account for the energy loss too, needing a separate sheen term (outside of EP bounds, at that). That's not an improvement from the original Disney paper, and leaves the shader open to being emissive under certain circumstances.

[davision]

Lele.... when you are talking about retroreflectors -- do you mean like road signage or just the subtle effect on the concrete?

I tried to use lower the GTR and glossiness settings (and no fresnel) but didn't see much happening (for signage). If it's possible can you give an example of actual values?

[^Lele^]

call it sheen, if you want.
Try with a very low ggx gloss, like 0.05, to begin with, and little to no fresnel (high IoR or absent).

[FLP]

Having now spent some time with the VrayALMtl shader it has left me very much impressed. For some time I have been wanting greater predictability when creating, specifically, dielectric materials which utilize glossy maps. For me, either correcting the native VrayMtl or fully implementing the ALsurface shader would be of highest importance.

These test (similarly to the ones provided by adanmq) show the "halo" effect at increased glossy values on the VrayMtl.

[vlado]

I can't open the attachments, can you try to re-post them?

Best regards,
Vlado

[FLP]

I can't open the attachments, can you try to re-post them?

Best regards,
Vlado

That's embarrassing. re-uploaded, cheers.

[vlado]

Thanks! Yes, the VRayAlMtl material handles this better as the Fresnel is included in the BRDF calculations.

Best regards,
Vlado

[M.Max]

Sorry for my ignorance but I heard that VrayAlmtl doesn't support a map in IOR? Is that planning to be added in the future? I am asking because I work with Substance Painter and IOR maps are essential to create thesame material.
Thanks,

[vlado]

It does support IOR maps, of course. I haven't checked if they work yet, but if not then it will be trivial to fix.

Best regards,
Vlado

[Morbid Angel]

Looks much better I want it in maya!

[jasonhuang1115]

Looks much better I want it in maya!

Maya version should be available on Github already. Isn't it?