skin shader a la Alsurface pleaseeee

[sharktacos]

With colors that are close the the presets in SSS2, it works very well. But the problems arise when you try to get precise colors other than those.

The trick I've found to get SSS2 to work well with precise colors is to begin with a preset (skin pink for example) then
1. set overallcolor to 0.8... this is the most important part
2. keep diffuse off
3. plug your map into the subsurface color
4. make sure your scatter color is not too saturated. Also be aware that higher values here in the color act like a multiplier of the scatter effect.

That gets you wysiwyg colors from sss2.

[jstrob]

The trick I've found to get SSS2 to work well with precise colors is to begin with a preset (skin pink for example) then
1. set overallcolor to 0.8... this is the most important part
2. keep diffuse off
3. plug your map into the subsurface color
4. make sure your scatter color is not too saturated. Also be aware that higher values here in the color act like a multiplier of the scatter effect.

That gets you wysiwyg colors from sss2.

Thanks for the tips. A good example of problem with saturated SSS2 scatter color is this:

If for example I have this blue skin character with green blood and I change the saturation of the green scatter color of only 1 unit (from 253 to 254), the colors all suddenly change for red, pink and yellow...

I once got this kind of problem with a food product and that's when I changed for the translucency in Vrmtl for food.

[ian_barter]

Not yet. I was not sure if people will find it useful. Do you want us to add it?

Best regards,
Vlado

I'd like to see it in Maya. Seems to be pretty useful.

[Bigguns]

Valdo : Did you finished the alsurface, did you added the ID raytrace ? and also samples for the SSS ? and one bump slot for each reflection layer?

[sharktacos]

Valdo : Did you finished the alsurface, did you added the ID raytrace ? and also samples for the SSS ? and one bump slot for each reflection layer?

On Github, under the "issues" tab, you can see which ones are still open:
https://github.com/ChaosGroup/vray_al_surface/issues

[sharktacos]

Thanks for the tips. A good example of problem with saturated SSS2 scatter color is this:

If for example I have this blue skin character with green blood and I change the saturation of the green scatter color of only 1 unit (from 253 to 254), the colors all suddenly change for red, pink and yellow...

Wow.

By translucency in the VrayMtl, do you means something like this:
http://richardrosenman.com/portfolio...tion-tutorial/

[jstrob]

Wow.

By translucency in the VrayMtl, do you means something like this:
http://richardrosenman.com/portfolio...tion-tutorial/

I remember that Richard Rosenman page, but this one is just using the fog, the explanation about how to do SSS with translucency in the V-RayMaterial is over there (By Svetlozar Draganov):

http://forums.chaosgroup.com/showthr...t=translucency

The only problem with Svetlozar tutorial is that you end up with the thinniest parts not being transparent and with some black artifacts.
like this:



To avoid this:

1-I put the refraction glossiness to something like 0.4.
2- I teak my material without translucency adding more refraction
3-I set the fog color to add some SSS in the transparency (but before adding translucency, all will take that fog color)
4-Then I add translucency and the back color will mixed wiht the diffuse to tint the more opaque parts while the fog color will tint the more transparent parts.

It took me so much time to understand all this. But now with the Alsurface it's much more intuitive. But the translucency can be useful when you need a faster render (we are talking about around 7.5 times faster) and looks still very good for stuff other than skin.

Here is an example:

[sharktacos]

"But now with the Alsurface it's much more intuitive"

Have you experimented with the translucency in the ALsurface (i.e. transmission) in the full (Arnold) version of the shader? I don't think you could get the "green slime" look with the SSS alone in the ALsurface...

[Mnietek2707]

What about transsmision layer? Is this planned?

[cpnichols]

Hello guys... I have been following this tread and have been working on this for the wikihuman project and getting ready to do a post about it on the site. I figured I would share some of the early results I have been getting. It has been a bit tricky to reverse engineer the inputs from each shader so I had to guess several parameters and tweak a few variables as they don't match one to one. Fresnel as one example.

In these results you will notice a few things, which, based on this thread, will not surprise you. The issue of slightly glowing lips and nose seem better with the ALShader. On the other hand, the ear has a bit more back scatter in the original wikihuman shader which matches the reference better. Also The wikihuman shader seems to pick up better GI in the shadow area under the chin compared to the AlShader. Much of these issues could be user error so I will continue to investigate.

One clear benefit to the ALShader is that we no longer have to tweak the reflections by adding the reflection on top of the SSS to preserve the SSS at the glancing angles. The ALShader seeme to do it on it's own, albeit I am curious as to what math it uses to make sure it has the full light preservation.

[jstrob]

Hello guys... I have been following this tread and have been working on this for the wikihuman project and getting ready to do a post about it on the site. I figured I would share some of the early results I have been getting. It has been a bit tricky to reverse engineer the inputs from each shader so I had to guess several parameters and tweak a few variables as they don't match one to one. Fresnel as one example.

In these results you will notice a few things, which, based on this thread, will not surprise you. The issue of slightly glowing lips and nose seem better with the ALShader. On the other hand, the ear has a bit more back scatter in the original wikihuman shader which matches the reference better. Also The wikihuman shader seems to pick up better GI in the shadow area under the chin compared to the AlShader. Much of these issues could be user error so I will continue to investigate.

One clear benefit to the ALShader is that we no longer have to tweak the reflections by adding the reflection on top of the SSS to preserve the SSS at the glancing angles. The ALShader seeme to do it on it's own, albeit I am curious as to what math it uses to make sure it has the full light preservation.

[ATTACH=CONFIG]32657[/ATTACH]

Cool thanks for sharing your tests!

It seems to me that the ear is glowing like lips in the wikihuman shader. BTW what is the wikihuman shader (I've heard of the project but not all the details) is this rendered in V-Ray too?

The alsurface may need a bit ore scatter in the ear but you could surely take control of this with a map?

would be could to see the same comparison but with a strong backlight.

[cpnichols]

The Wikihuman project is available here: http://www.wikihuman.org/ It is part of Chaos Group Labs, which I am the head of.

If you want to get the data, you can use it as well: http://www.wikihuman.org/index.php/off-topic/emily-2-1/

The shader is an exposed layered shader which was done that way so that anyone can recreate the shader using any rendering engine. It is, however, done in V-Ray. The math of the shader is based purely on the ICT data set, so it tries to minimize as many of the arbitrary decisions that an artist would make.

Much of this structure fits directly into the ALSurface shader.

[jstrob]

The Wikihuman project is available here: http://www.wikihuman.org/ It is part of Chaos Group Labs, which I am the head of.

If you want to get the data, you can use it as well: http://www.wikihuman.org/index.php/off-topic/emily-2-1/

The shader is an exposed layered shader which was done that way so that anyone can recreate the shader using any rendering engine. It is, however, done in V-Ray. The math of the shader is based purely on the ICT data set, so it tries to minimize as many of the arbitrary decisions that an artist would make.

Much of this structure fits directly into the ALSurface shader.

Ah ok it's you Chris Nichols! I know who you are! It's a honor! I saw you yesterday on a video about THU with the Zbrush summit interviewer dressed like a drag queen! Seemed like a lot of fun! LOL!

Thanks for the explanation. I'm downloading the project right now. I did it a long time ago but didn't really explore it in depth, it's time to do it!

[JeffG2]

I see a vrayalmtl for Max 14, 16 & 17.

What happened to 15?

[sharktacos]

One clear benefit to the ALShader is that we no longer have to tweak the reflections by adding the reflection on top of the SSS to preserve the SSS at the glancing angles. The ALShader seeme to do it on it's own, albeit I am curious as to what math it uses to make sure it has the full light preservation.

Hi Chris,

Rather than computing Fresnel once, based on viewing angle, the ALsurface computes the Fresnel for each microfacet. This has the benefit of avoiding both edges that are too dark (which is what you are referring to I believe) and also over-bright spec at glancing angles. Anders discusses this here. If you wanted to look at the math, the shader is open source, so you can see the c++ code here (as you can Vlado's implementation of it on Github).

My understanding is that Vlado is planning on changing the VrayMtl to work with per microfacet Fresnel for the next service pack. However, if one is doing the technique of using a blendMtl to combine SSS from one material with the spec/reflections from another material (which is how the shader for Emily2 is made) the Fresnel will not work properly since to be energy conserving one would need to subtract the inverse spec Fresnel from the sss. That's not possible to do per microfacet outside of the BRDF. What's therefore nice about the ALsurface is that it has 2 GGX spec lobes together with SSS and diffuse, so the Fresnel mix can happen inside the shader.

One thing I'm not quite sure about is how to get the Lambertian diffuse that you called "single scatter" to be part of the ALsurface.

I noticed that your render has hair (including eyebrows and "peach fuzz" on the face). Is that available as part of the 2.1 update?