Which method of SSS is best for a skin shader?

I’ve created four node arrangements in Cycles for simulating skin capillaries and veins, and I’m I’m looking for feedback on which method is the best, or more likely, how to improve one the the materials.

The first material mixes the color of the arteries and veins onto a base skin color for the diffuse, then adds a solid color SSS shader.

The second one uses the color of the veins and arteries as input for the SSS shader and then mixes it with a defuse shader.

The third one is pretty similar to the second, but uses an add node instead of a mix node.

The fourth one is a mix of the first and third.

Given that my explanation of the node setups is so objectively terrible, I’ve included a blend file as well as a render.

The blend is located at:
http://www.pasteall.org/blend/23113

Attachments


I rerendred the models with the radia set to R: 0.2, G: 1.0, B: 0.05. Improvement?

Attachments


None of those is great for true skin. You’d want a multilayer approach where each represents an actual layer affecting the light entering and exiting the skin. You could also look up the GPU Gems method of using a sum of gaussians which is a popular option these days in production. Either way, you’ll likely want little or no diffuse contribution, two specular lobes, and different details on different layers. The render below was done using the first technique and the shader will likely be released in the near future after it’s further refined either through myself or Blender Cookie. There are some big changes coming to the BSSRDF in the near future though which should improve both quality and speed, so you should likely wait for that before doing a lot of research into a skin shader.


I’m guessing that would be a combination of three SSS shaders, one for the skin pigmentation, another for the capillaries, and another for the veins?