It is difficult to master this shader and you need several render tests before having the wished result.
To understand its way of working and its nuances, you have to imagine the SSS as a classical material (blinn, lambert) with one more possibility: the transluscent effect.

Actually, this shader enables getting several successive layers, and having more or less light going through them so as to create depth.

The shader can also simulate the color variations of the model according to the lighting, the insides of the model and its irrigation.
The SSS acts as a set of layers with modulable opaqueness.

This is an interpretation of this shader:


First of all, construction of the Shader and first basic parameters:

Update: Construction of the shader with the previous versions of Maya (before 8.5). From the 8.5 version, the shader is already made, you don't have to assign it to the object.

Go in Hypershade : create mental ray nodes, in Materials, created a ‘’miss_ fast_skin_ maya’’.

Double click on the icone.

In attribut editor, open Lightmap, click on ‘lighmap’ (we will set the parameters of this new window further)

You should get a new icone in the hypershade : mentalrayTexture1, linked to miss_fast_skin_maya.

 

Then, select the miss_fast_skin_maya icone and when unrolling its content , double click on miss_fast_skin-maya1SG
In attribut editor : open the section mental ray_ in ‘custom Shaders’ clik on ‘light Map Shader’

The window "create render node" opens. Unroll it and on the bottom in ‘light Maps’, select ‘Miss_fast_lmap_maya’

            

In hypershade, a new icone shoud appear.

After having put it, the ‘miss _fast_ lmap _maya’ is selected ; in the attribut editor, open the section ‘lightmap Write’, click and drag the 'mentralraytexture1' of the hypershade in ‘lightmap’.

The shader is ready, and can now be linked to the model.
A few settings and you can see the effects with a first render.

Select the miss_fast_skin_maya.

In the section lightmap : the number of Sample defines the quality of rendering, the higher it is, the more details (and rendering time) you will get. The number must be a mutiple of 2. 256 is a good quality / time compromise.
By clicking on Lightmap, the window ‘’mentalrayTexture’’ opens.

Same panel, but in maya 2008:
a script is assigned to 'file size width' so  that the entered number = 2 times the length of my render (set in render setting, here 720).

In the same way, a script maintains the file size height = the height of my render.

       
Here, we define the quality of the SSS and of the transluscent effect. ‘Writable’ should be ticked. The size width should be 2 times the length of my render. For instance, for a 720 / 576 render, you have to enter 720x2= 1440.
The size height should be equal to the height of my render.


The size depth should be 32 bits, so that the SSS works.

Finally, you have to write the path where the SSS settings are saved.
In ‘image name’ :'' \'' followed by a name, for instance: \dog, this means that the SSS will save its data on my D drive, under the name "dog".

A last setting:


In the section Algorithm control:
So as to avoid having some noise on the model when rendering, and so as to have a convenient dispersion of the transluscent effect, you have to increase the number of ''scale Conversion''. 25 gives me good results.


Skin and translucent effect, understanding and settings:

With textures and settings for the Vrex skin for instance.

Diffuse layer 

The diffuse layer is a complete layer. It enables, as a classical shader, to put a texture on a model (diffuse color and Overall color ) and to set its intensity (diffuse weight).
Le diffuse layer defines how the skin receipts the light, lets it more or less enter and reflects it. The shaders of the model depend directly from the diffuse layer: the more opaque the diffuse layer is, the clearer the shadows are.
It is necessary not to decrease the intensity of the diffuse layer too much, otherwise, your model will lack of reflected light and the texture and the shadows would consequently be quite non-existent. However, so that the transluscent effect of the SSS is visible, you should not increase it too much so that a minimum of light can enter under the skin and reflects the Subsurface textures.


SubSurface Scattering
layer

The Subsurface scattering layer groups 3 other layers, hierarchically under the diffuse layer.

The first two layers:

The Epidermal scatter and the Subdermal scatter have almost the same function, they are used to reflect the light, and consequently the color relative to everything just under and on the skin, to create a depth belief.
The epiderm (represented by the epidermal scatter) is indeed the thin superficial part of the skin and the derm (Subdermal scatter) is a deeper one.

 

Actually, technically, the only difference between those two layers is linked to the ''scatter radius''.
The biggest it is, the more the layer will looks deep under the skin and the less the shadows will be visible.
Because even the layers under the skin have the ability of flattenning the shadows, which gives the relief effect, and consequently the depth effect.
This is why the first subsurface layer, epidermal scatter, should have a quite low radius so as to simulate the skin. Concerning the opaqueness, it should be over the subdermal scatter, which is only used to get a further deeper layer, and consequently a less clear one (which explains the high radius). The subdermal scatter is not indispensable, but it brings a contrast, particularly if the used color is darker that the base texture.

For my renders, I only use one texture, that I modify a little for each layer.
I create a generic texture, I use the same version for the overall color and the epidermal scatter color. I lighten it for the diffuse color.
For the subdermal color, I do a darker, yellowish and contrasted version of the texture (of course, this modification depends on the model (human being, animal, creature).

For the back scatter color, the generic texture will get completely red and dark.

The back scatter color:

This layer is the hierarchically last one, and is essential to the SSS function.

Exactly as when you put your hand over a flame: the nearest it is from the warmer part, the more the light will go through it. A mulitiude od photons, filtered by blood, will reflect a redish color... or greenish if the blood is green!

The back scatter color is used to tell the SSS the color of the blood it will have to reflect.
It is not essential to put a texture to see the effects. In that case, a color red would be enough.
However, I use a texture so as to get some variances of color, because the light is meant to go through a non homogeneous body, and is consequently irregularly filtered. That's the interest of using the texture and to make it red.

Note that when you want the SSS to make some parts of your model opaque (bones, metal...), just paint these areas in black on the texture.

Concerning the Back Scatter settings:


The highest the weight is, the more visible the blood is: so that it becomes true, you should have the adequat lightning, because the red color doesn't depend only from the SSS but also from the intensity and from the axis of the light going through the skin.
This is why I put the light in the back of the model and in front of the camera: you'll get this effect every time.

    

Note that by unticking Illuminate by default, I get exactly the transluscent effect I wished, without the reflects generated by this light.


The lowest the back scatter depth is, the more the red color will be on the fringe of the body and the less it will be visible in the middle of the body.
Interesting if you want only the ears to blush for instance.
(It seems that the radius has exactly the same function as the weight here, so no need to touch it.)

This was how to set the absorbed light, the color, the depth, the opaqueness, and consequently the transluscent effect of the SSS.

 

Now let's see how to set the light reflected by the model.


Specularity:

The SSS specular is not similar to the other classical shaders.
Its particularity is that it enables the management of two types of reflects, so as to get a more detailed reflection.
The easiest way is to set them singly, and to combine them after.

I think that the settings of the specular should be done before you put the textures (diffuse), because it will be on a homogeneous surface which will give closer reflect effects.

Furthermore, so as to clarify the settings, I always let the overall weight (controls the global intensity of the specular) at 1.
The edge factor determines the softness or the hardness of the global specular.
Contrarily to a blinn for instance, the highest the number is, the harder the reflects are.


Now just see the two types of reflection:

The primary Specular: The goal of the example here is to explain the gross parameters, without added texture.

To set it, I cancel the secondary specular (secondary weight at 0).
Here, this reflect type will give a dynamic to the general volume, it will be soft, wide and of a low intensity. I set the primary specular so that it creates the effect of a satin-finished skin, the light has to spread on the model.
So as to do it, the Shininess is essential: it determines the hardness of the reflects. Here, it has to be very soft (between 1 and 5).
I let the default value of the primary edge weight, even if when I increase it, I notice a little change in intensity..
Finally, I determine the intensity of the Primary Specular (primary weight), this means the brightness of the first type of reflect, on the model.

The Secondary Specular:

It will be harder, it will give more dynamics to the model details, it will emphasize the displacement map effect.
It will anyway be less present, and its reflects will be seen on the end of the nose, on the ear for instance.

Here a lower intensity than on the first one (secondary weight) and a higher Shininess.

Panel with textures and parameters for the Vrex skin for instance.


The bump shader:

To be activated, you just have to create a 2D node bump in the hypershade, then drag and drop it in the slot bump of the SSS.
Note that when you delete it and launch a render, it bugs. In the unrolling bump panel du SSS, just check that the 3 values are at 0, so as not to have this problem anymore.

This is the gross Maya render, after final settings.

Should you have any question: carcaryas@gmail.com