Okay to use SSS artistically?

Alex_J

The general advice for SSS for skin is a light pink color for Caucasian skin, or as Saurabh Jethani does in his Marmoset breakdown, fully saturates the albedo texture and uses that.

It seems like SSS shader in marmoset absorbs whatever color of light you have set. So if I set red, the surface becomes more green/blue, and vice versa. Is this really how it works (asides from the obvious scattering of the light and exiting at different angles)? I looked up info about it but didn't read anything about the absorption of particular colors, only the process of scattering the light.

The main question though, is it breaking PBR norms to use your SSS color artistically? If I am texturing darker skin, it seems like a bluer SSS color gives the final surface color I want. Seems to hold up well under different lighting conditions as well. If it looks good, is it good?

Thanks for any advice.

musashidan

Yes, of course. SSS shaders have always been used artistically in offline rendering. The shader tech has vastly improved over the years, but the input channels have always been derived artistically. The 3-channel approach to offline SSS skin - epidermal/subdermal/backscatter - has meant that everyone was creating and inputting all sorts of map variations to get the result. Crushing levels, adding HSV effects, blending maps, handpainting.......

Back in the Mental Ray days we used to experiment with all kinds of maps in those 3 channels. The offline shader also uses a 2-channel approach to reflection/specularity - broad and tight - and these maps were no different in that people used all kinds of maps to get the final result. Back then, Vray didn't even use ray-trace reflections, you had to disable the physically inaccurate specular in the skin shader and blend or shellac a reflective Vraymat on top. It was(and still is to a large degree) very much a case of 'if it looks good', because before the advent of scanning, or even polarising - when everything was handpainted/projected from photographs and derived from existing maps, there was a LOT of experimenting and endless hours of look-dev rendering involved. Every artist had their own approach based loosely on how the reflection/glossiness maps should look. This is still very much the case today: deriving maps from the base colour map to extract bump/normal and reflection/glossiness - albeit in Mari instead of Photoshop.

Also, the effect of getting different colours than you expect is physically accurate. This is how it works in the offline shaders as well. Remember that technically the uppermost layer of skin is 'dead' looking. It's grey. The blending and scatter of deeper layers is what gives the 'fleshy' look to skin.

Alex_J

thanks!

Alex_J

just want to update cause i think i may have put out some bad info -- in fact right there in marmoset the mouse-hover tooltip explains about color absorption within the the shader.

So I guess it makes sense -- if you set a red color, that will absorb more of the red wavelength and then scatter it, so overall your skin would appear more blueish.

EarthQuake

BIGTIMEMASTER said:

just want to update cause i think i may have put out some bad info -- in fact right there in marmoset the mouse-hover tooltip explains about color absorption within the the shader.

So I guess it makes sense -- if you set a red color, that will absorb more of the red wavelength and then scatter it, so overall your skin would appear more blueish.

Yep, it can be a little weird at first but as @musashidan points out it's physically based and very important to representing skin well.