vraySkinMtl compared to fastSSS2

[nickgaul]

hello!

was curious what peoples take on the new skinMtl were compared to the sss2 shader? what are peoples approach to setting up the shader and how are they finding its results? After spending some time working with the skinMtl I just cant seem to find that sweet spot to get it actually feeling like skin. Particularly in the area's where shadow bleed occur. This was one thing i felt the sss2 shader out of the box did so well, and why i found it hugely successful. Im noticing with the skinMtl its appearing very "hollow" and the sss either feels as if its glowing or isnt really happening enough. Ive played with the scale to and depth of the scattering but it never quite feels right. here's a quick simple image comparison. pretty basic , just plugging in one map into the scatter color. will post more tests ive rendered later. i got a fairly better result plugging the same map i was using for the shallow scatter into the diffuse and bringing in some diffuse helped get rid of the glow from the deep scatter, but then the sss never quite felt it was looking right.

[vlado]

Well, the skin material is a cheat after all, for people who complained that they don't have complete control over how the material works. It also matches what people are used to from other renderers. Many users found, just like you, that fastsss2 actually works fairly well out of the box in many cases, even if it has some quirks itself, f.e. unwanted blue/green tint in thin parts - which, while physically accurate, is not always desirable.

There is one more addition that needs to be done to the skin material - an overall sss reflectance map (like the sss color map in vrayfastsss2) that would automatically adjust the three components in a way that produces the desired final look. Right now this is difficult to do with the three components being completely separated. Still haven't gotten round to finishing that yet.

Best regards,
Vlado

[nickgaul]

thanks alot for the feedback! here's another render comparison. on the skinMtl im using a map to drive the color for both the diffuse and shallow color, on the sss2 shader just the subSurface color uses this map, and the rest is defaults. ive tried quite a bit of various settings with the skinMtl , and it seems quite tricky to match the color bleed occurring in the sss2 shader , particularly where the shadow cuts off. it feels the deep color also contributes too much to the overall color, even when i put the diffuse all the way to 1. thats why i was curious if others have hit a sweet spot in the shader to produce a similar result, as i cant seem to find it yet. i actually was extremely happy with the results ive been able to achieve with the sss2 shader thus far. the feature in the new skinMtl i was looking forward to taking advantage of was the double specular lobe, and the IOR controls, but i can grab an extra spec from a vrayMtl and blend with my sss2 if needed.

that would be awesome if the skin material had an overall reflectance map like the sss color map. that sounds like it could be a solution to manage the 3 layers in way to get a result more easily.

[griml0ck]

One way to get the bleed effect with the new shader is to use a red color in the medium channel, and then use a green/blue color in the shallow slot. This way the red will bleed more and then green/blue will be adding on top which will give you the skin color. This way you have much more control over the bleeding edge, as you can just tweak the radius values to control how much of the red edge you will see. Personally had more problems with the old shader, I think the new one is better.

J

[nickgaul]

thanks johan! i actually saw thats how other skin shaders from other renders were being setup. using blue and green values in the shallow/ mid slots. im going to give this a try!

[nickgaul]

ive finally come across some of that "greenish" tint occurring in the sss2 shader. u can see just around the lower eyelid. im thinking this may be resulting from the surface being thinner than the scatterRadius? i need to try that out. it made me go back and check out the skinMtl a bit more...



both setups ive tried ive used a "greenish" color in some shallow area, and one playing with using the medium scattering to get more my desired color values by putting it in the medium slot. both dont quite seem to be working as good as id hope. is it physically accurate to be using blue/green maps in a skin shader? is that how the shader is intended to be used? it seems a bit cumbersome to have to deal with all these colors / maps to get what one map should be able to achieve : / and i feel to get more of my actual map colors i have to plug that into the diffuse and add more diffuse , killing the sss in most areas. it would be cool to see more documentation on how the shader should be used with examples of values. even the scale on the shader seems to respond much differently than the sss2. where if you have real world head size, 1 is the scale to use, where with the skinMtl i find i have to turn that down to like 0.3 - 0.5.

[sharktacos]

I'm also getting used to the new skin material. I'm much more familiar with the SSS2 and frankly prefer it. The only issue is with the green edges that appear. As I understand it there is no way to fix this. That's a shame as I find I can get much nicer results with the SSS2 than I can with the new Skin material which seems a bit more "cartoony" to me.

For what it's worth here are my settings for the shaders. This is by no means an official "here's how to do this right" setup. It's just my attempt to get predictable results where the color of your maps correspond to the resulting render rather that doing unintuitive things like having a blue map to get pink skin or having a dark map to get bright skin. I'd love to hear other folks approaches as well.

Lights & Linear:
This is working with a linear setup, so no color mapping since the mode is set to "don't affect colors, only adaptation." I find having a bright light (a directional with the intensity set to 2 for example) helps a lot with getting nice spec and SSS bleed.

SSS2:
overall color: 0.8, 0.8, 0.8
diffuse amount: 0
sub-surface: map of skin color
scatter color: orange-y version of the skin color map.

Since there is no "scatter color amount" (hint, hint), you need to control this by adjusting the value of the map (the color gain on the file node). This has a big impact on the SSS look, and if it's too bright it will blow out.

The overall color is set to 0.8 similar to how this in done on the diffuse amount of a VrayMtl. This allows the colors to be WYSIWYG rather than the darker color values you see on the "skin pink" present.

Skin material
diffuse amount: 0
Shallow color: 0.8, 0.8, 0.8
shallow amount: 0.4
shallow radius: 0.1

medium color: map of skin color
medium amount: 0.5
medium radius: 0.5

deep color: orange-y version of the skin color
deep amount: 1.0
deep radius: 0.5

What's curious about the skin material is that the amount is not a multiplier for the color. So if I want to have less deep color and lower it from 1 to 0.5 the result is brighter. I'd be interested to learn more specifically what's going on here exactly.

I find I need to make the deep color darker and more saturated on the Skin Material than I do on the SSS2 to get a similar nice reddish bleed.

[nickgaul]

thanks sharktacos! ill give your settings on the skin material a try.

i did notice that affect from the scatterColor with the sss2 , where the gain of that scatter color map would have a pretty significant impact on the result.

the greenish edges really seemed to happen on thinner pieces of the mesh. for me it was only around the eyelids. this green color also would transfer to any refractive material i had setup to get some extra spec around the eye / eyelid transition ( a tear layer) which was a bit frustrating. the best way i found to get rid of it , or reduce it , was to paint down the scatterRadius quite a bit in those areas. interestingly i didnt notice this happening with the previous version of vray where the sss2 didnt have the raytraced option but only the prepass illumination map : /

[sharktacos]

In my experience the green error comes from having multiple geo. If the geo is combined the error goes away.