I'm trying to get a similar material to the plastic of a 3D print which is largely dominated by subsurface scattering (SSS):

enter image description here

However, a very simple SSS test in cycles produces this:

enter image description here

There's an awful lot of lot of blue (I assume back-scattered light) at the corners that I can't get rid of. The same happens at smaller detailed bits of geometry in other tests.

A simple answer is I've buggered up the properties and just need to change some numbers. If this isn't the case,

  1. Is this simply a failure case of the current SSS method in cycles or the lack of some other physical phenomenon that isn't being modelled?

  2. Can this be avoided easily? If it's necessary to set up a complex network of nodes to hide the effect then maybe it's not worth the effort.

OK, it doesn't look like forward/back scattering has anything to do with it. Small objects simply have the wrong colour, as in the following image. I'm pretty sure if I shaved off a chunk of my model it wouldn't turn blue/green.

enter image description here

Objects on the left have 0 sharpness while objects on the right have 1 which seems to alleviate the problem a little. I would have thought this is the wrong way round given the look (maybe sharpness should be roughness instead).


Without seeing your current setup (what you've already done), it's a little hard to say exactly what is going on and how to change it..

Here's what I came up with after a bit of fiddling (just trying to match the reference image by eye):

enter image description here

enter image description here

  • $\begingroup$ This looks pretty similar to what I've seen, though using the layer weight is a nice idea. I still find the blue on the spikes of the dragon's head looks quite odd. This effect increases with larger radius/scale values. Maybe this colour is physically correct but gets cancelled with a high orange surface reflectance but to me it still looks unnatural. $\endgroup$
    – jozxyqk
    Jun 18 '14 at 11:07
  • $\begingroup$ The reference image in your question also has a bit of a greenish appearance around the spikes.. Also keep in mind that lighting will have a huge affect on the result. Part of the greenish hue in my example is caused by the lighting (there is a blueish meshlight). $\endgroup$
    – gandalf3
    Jun 18 '14 at 11:26
  • $\begingroup$ Yes, my reference is in a Cornell box with red/green walls but the colour is subtle compared to the SSS issue. I've added a better example image where the problem occurs in small objects. $\endgroup$
    – jozxyqk
    Jun 18 '14 at 12:16

sss bluish band artifact Note about scatter radius in Cycles:

Because of the way this scattering is calculated, when using large radius values, you will notice fringing artifacts that appear as the complementary color to the predominant color of the scattering. Above, you see in the last image a bluish band in the illuminated area. This is an unfortunate limitation. A way to lessen this effect is use multiple passes with different scatter radii, and average them.

for more info: visit:http://wiki.blender.org/index.php/Doc:2.6/Manual/Materials/Properties/Subsurface_Scattering

Example of multiple passes:Brecht's Six Layers SSS Six Layer SSS Six Layer Shader Six Layer SSS Shader Group Values

For the default SSS shader node to generate it's correct effect,set the radius RGB levels to low values to get a accurate results,EDITit seems that the scale to ratio is off,the formula should look like this s1.00 = bu1.00,but it's more like s1.00 = bu100.00 or s.100 = bu1.00,just try it with a low level of scale .100 and left the radius level at R1.00G.500B.250 and it gave the same exact results as leaving the scale at 1.00,but lowering the Radius RGB values R.100G.050B.025,no more blueish tint:


Scale value:

scale = 1.00

Radius Values:

Red(R) = .100

Green(G) = .050

Blue(B) = .025

low radius sss value enter image description here


Scale value:

scale = .100

Radius Values:

Red(R) = 1.00

Green(G) = .500

Blue(B) = .250

scale of 1 scale 1 close


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.