Hi there folks, I hope I can revive this conversation a bit even though it’s late. I stumbled upon this thread in my search for some of the same info.
In the testing I’ve done I’ve found that the color value (ie value from black to white in HSV) you give for any given SSS layer seems to have exactly the same effect as lowering the sss weight for that layer. For example: I’ve found that by taking sss maps that worked well for a caucasion character and simply adjusting the entire image to half the “brightness” I get the same result as I would get by reducing the SSS weights by half.
This seems inevitably problematic for doing really dark skinned characters, for the following reasons: In my research I’ve found that the epidermal layer of skin is where the pigment (melanin) is. Skin can range from almost black (due to very high concentrations of the dark brown pigment melanin) to nearly colorless (See this link: http://en.wikipedia.org/wiki/Human_skin_color) The link I just directed you to states that nearly colorless skin appears reddish white due to blood in the skin, however another sources states “The epidermis is avascular (contains no blood vessels) and is nourished by diffusion from the dermis” (see http://en.wikipedia.org/wiki/Epidermis_(skin)). The maya documentation on the sss_fast_skin shader states that the subdermal sss layer is supposed to represent collectively all the layers that lie below the epidermal layer. I’ve found that the dermis is very distinct from the epidermis (http://en.wikipedia.org/wiki/Dermis) and so one would expect that the dermis, which contains the blood vessels of the skin is intended to be represented by the subdermal sss layer. However, I have a feeling that this is not the case, that in fact the dermis is intended to be represented in the epidermal layer, and that mental images has simply failed to mention this conflicting detail in the documention that we have. My suspisions are due to my observations of how the subdermal layer shows through, but I wont detail those here.
I’m getting off track… back to the topic - since the darkness of the color you use for a sss layer acts as the weight of that layer the result is that the darker you make the “pigment”, Ie - the epidermal layer of your skin the less influence it has, or rather - the more transluscient or trasparent it is, allowing the subdermal layer to show through more. To me this seems closer to the inverse of what one would expect. The first link I quoted states that black skin is a result of high concentrations of dark brown melanin - this suggests that light skin is a result of low concentrations of Lighter colored melanin. One would expect therefore that lighter skin has less influence or is more transluscient, and that darker skin has more influence because it is more opaque. But, when you increase the weight of the epidermal layer the effect is as though you had simply turned up the “value” (HSV) of the epidermal layer color, effectively turning your african character caucasion, when you’d expect the epidermal layer to take on a more lambertian effect instead. Personally I’d prefer if the color of the layer had no influence on the apparent weight. I kind of expect though that there are some deeper reasons for this, and I keep hoping that I’ll find the “intended” solution.
Right now it seems like the mr sss_fast_skin shader may have been optimized for lighter skinned characters, while the requirements for truly dark skinned characters may have been over-looked. Having said that, I have seen some black characters using sss that look good, but I expect the result is achieved by “cheating” (probably unknowingly), and personally it bugs me when I have to resort to that (though inevitably you have to somewhere or other).
I’ve a few options to explore still such as using the diffuse weight to effectively bring back the influence of the epidermal layer, and a few other things, but I’m out of time to write about those theories.
Anybody else have some thoughts to contribute?
