skin shader a la Alsurface pleaseeee

[sharktacos]

The SSS colors are interpreted as distances, so I think they shouldn't be gamma-corrected, but it might be better to ask Anders himself what is his original intention.

So I asked Anders, and he said that "the sss colours are the desired diffuse albedo, *not* distances," so they would need inverse gamma correction just like the diffuse color would. Good to know.

Not quite, they should be very similar in speed. What is the speed difference?

In the tests I did, I'd say the ALsurface was a bit less than 2x longer than the SSS2. So the speed difference seemed significant. Note that this is in Maya 2015. I did not test it in Max.

[jstrob]

In the tests I did, I'd say the ALsurface was a bit less than 2x longer than the SSS2. So the speed difference seemed significant. Note that this is in Maya 2015. I did not test it in Max.

Hi sharktacos, for me in 3ds max the render times are similar for SSS2 and Alsurface. I did a small comparison of all the SSS methods in V-Ray to see the render time difference and I got these results:

Alsurface: 28m07s
SSS2: 28m41s
skinMtl: 7min6s
scatterVolume: 30m7s (but full of artifacts and probably more than an hour to get it clean so this one is just too slow)
V-Ray translucency in standard V-RayMtl: 4m50s

it was with this turntable test scene:

[RichardBlank]

The new alshader for maya is supernice! Is the alhair for maya also going to be included any time soon?

[vlado]

Yes, I'm working on the al hair shader.

Best regards,
Vlado

[Morbid Angel]

Yes, I'm working on the al hair shader.

Best regards,
Vlado

Excellent news! the alsurface has been single nicest feature in vray 3 since a long time! I look forward to alhair as well!

[Bigguns]

i totally second that !

[sharktacos]

Vlado, That's great to hear about the hair shader!

Any chance that we might get a fabric shader too? I see that Marschner has quite a few papers on the topic, including a shader by Irawan & Marchner. Any chance any of this might make its way into a new VrayFabricMtl?

It would also be great to have a procedural texture for generating common weaves such plain, twill and satin



Maya has the cloth texture, but it only has one type of weave, so it would be useful to have more types (there are really only three, (plain twill & silk) with everything else being variations on these common types. It would also be important to have ways to randomize them a bit. This is possible to do with filters in Photoshop, so perhaps something similar could be done procedurally inside of Maya or Max?

[jstrob]

Yes, I'm working on the al hair shader.

Best regards,
Vlado

Wouhou!!! Cool! Can't wait to play with it!

Those fabric shaders looks cool too!

[^Lele^]

The siggy '11 paper requires a measurement of the cloth to be parametrically remodelled, it looks to me.
At that point, why not use VRScans?
The Irawan thesis and approach, instead, is wholly procedural (well, not really. texturable in principle, and the shader in mitsuba comes with some pre-measured files.), and can do a very decent job with front-facing clothing and patterns, but fails to capture the fuzziness of the stray fibers completely, resulting in what to me looks too plasticky, and the paper doesn't provide for any shadowing function for its BRDF, so a straight implementation wouldn't really look any good.

Ultimately, for stuff which has the volumetric complexity of cloth, nothing beats a VRScan today, as far as i can tell, and i cannot imagine what kind of mathematical model would be able to cater for more than a couple of very specific scenarios.

[sharktacos]

The siggy '11 paper requires a measurement of the cloth to be parametrically remodelled, it looks to me.
At that point, why not use VRScans?

Yes if one wanted a specific fabric patterns then Vray scans would be the way to go. Note that the link I gave has papers from 2011-2016. I was thinking more of an artist driven approach. Thus the 2012 paper "Structure-aware Synthesis for Predictive Woven Fabric Appearance"
looked promising:

"Procedural techniques are manually intensive, and fail to capture the naturally arising irregularities which contribute significantly to the overall appearance of cloth. Techniques that acquire the detailed 3D structure of real fabric samples are constrained only to model the scanned samples and cannot represent different fabric designs. This paper presents a new approach to creating volumetric models of woven cloth, which starts with user-specified fabric designs and produces models that correctly capture the yarn-level structural details of cloth."

but perhaps there is a more recent paper that has an even better approach...

[^Lele^]

Yes if one wanted a specific fabric patterns then Vray scans would be the way to go. I was thinking more of an artist driven approach. Thus the 2012 paper "Structure-aware Synthesis for Predictive Woven Fabric Appearance"
looked promising:

"Procedural techniques are manually intensive, and fail to capture the naturally arising irregularities which contribute significantly to the overall appearance of cloth. Techniques that acquire the detailed 3D structure of real fabric samples are constrained only to model the scanned samples and cannot represent different fabric designs. This paper presents a new approach to creating volumetric models of woven cloth, which starts with user-specified fabric designs and produces models that correctly capture the yarn-level structural details of cloth."

"We create a small database of volumetric exemplars by scanning fabric samples with simple weave structures. To build an output model, our method synthesizes a new volume by copying data from the exemplars at each yarn crossing to match a weave pattern that specifies the desired output structure."
Only insofar as it decouples the large texture from the weave scanned data.
Without the right set of scanned inputs for weave and volumetric structure, i highly doubt the results would come any close to the real thing.
Besides the fact that the scans are volumetric datasets of several hundred of megs, requiring many tens of gigs to render for the simplest of patterns (see samples. Mitsuba also has a scarf samplemade such, with detail inferior to that of similar cloth with vrscans...) though BF volumetric path tracing, you'd get a much better result using the VRScan of the type of cloth you want (say, white satin) and drive the color with whatever you please, leaving the BTF, and its intricate microscopic details, intact when reacting to lighting.


As for sheen (before your edit), i guess you realised by now that low GTR gloss (0.1 and lower) produces sheen (EP within the shader, layerable with a blend with the usual EP trick for mutiple spec layers.).
Changing IoR (or mapping the falloff manually between 0 and 1) and Tail Falloff allows for a more precise shaping of the effect intensity.
Below a shader with only a GTR spec @ gloss 0.05, and IoR 1.6 and Tail FallOff at 5.0, lit by a white dome.
There is some energy loss inherent to the very very low gloss, so avoid stacking Glossy Fresnel on the effect, lest it disappears.

[sharktacos]

You may be right about the Marschner stuff, but I'd still like to hear Vlado's thoughts as well.

[vlado]

You may be right about the Marschner stuff, but I'd still like to hear Vlado's thoughts as well.

https://github.com/vidarn/ThunderLoom

Would you like to see this bundled with V-Ray? It's a cloth shader developed for IKEA, with a V-Ray implementation. The only caveat is that I have no clear idea how the loom files are produced.

Best regards,
Vlado

[^Lele^]

https://github.com/vidarn/ThunderLoom

Would you like to see this bundled with V-Ray? It's a cloth shader developed for IKEA, with a V-Ray implementation. The only caveat is that I have no clear idea how the loom files are produced.

Best regards,
Vlado

uses a weaving pattern given by a WIF file. Many types of patterns are freely available from sites such as http://handweaving.net.

I'd be very curious to see if it can get anywhere near this, to be honest.



EDIT: and most importantly, how it'd deal with much rougher clothing, like the following two (you ain't seeing things, the fibers are "true geo", correctly displaced, as the overlayed logo shows. VRScans can also capture heightmaps.):

[^Lele^]

a few more.
In each the only things changing are the type and scale of the loaded VRScan, which provides for light reaction coefficients' values (because saturation is set to 0.0), while the filter texture provides the color values, and the exposure, per image, as we currently lack a gain control for the incoming VRScan (getting there.).
Some UV distortion is due to the mesh being a triangle network for cloth simming, which i didn't relax or post-produce well enough: