Transparency and SSS

I’m working on a character (morphs, textures, materials, hair, and clothes) based on a 3rd party mesh, and I’ve encountered a problem I can’t find my way around. I think I’ll be reporting it as a bug if it hasn’t been before (I can’t seem to find it listed on the bug reporting site), but I’d still like to find a way around it. Because the mesh isn’t mine, I have to use transparent lashes. Unfortunately, the skin’s SSS behind transparency seems to get some sort of light bounce from the transparent area when it’s close.

Here’s the problem in action:
transparencyProblemShow

Here’s the same face with the lashes hidden by scaling:
transparencyProblemHide

And this is the material for the lashes:

If anyone can tell me a hack around this, I’d really appreciate it. And if it’s already been reported as a bug, let me know.

Oh, and just to say, that’s just a first try on the brows. I’ll definitely be redoing them until they’re decent. And they eyes aren’t done either.

make sure that the transparent color is full white (1.0)
otherwise it could be diffuse bounces from your mat.hm not sure about this.

what blender version do you use?are the scalings applyed? (ctrl a)

is the mesh overlaping into the skin?

make sure that the transparent color is full white (1.0)

That was the first thing I did, along with checking the transmap. You can see in the images that the lashes have no visibility where the transparency is set to 0.

otherwise it could be diffuse bounces from your mat.hm not sure about this.

Not exactly, I don’t think. For one, it’s not a problem with diffuse shading, only SSS shading. In fact, you can see someone else posting about this problem here showing that the problem goes away as soon as you switch to Diffuse shading behind transparency. I’m guessing the SSS surface is acting like the transparent areas are bouncing light into them, but even bounced light shouldn’t make it supersaturate and glow like that. I’ve tested with a LightPath node, and the error is always exactly where Transparent Depth is 1 or greater. But using any other path testing (Diffuse Depth, Diffuse Ray, etc.) doesn’t correlate, so I’m not sure it’s any kind of diffuse bouncing.

what blender version do you use?are the scalings applyed? (ctrl a)

2.79, but as shown, this has been a problem for a bit now. Yes, though that shouldn’t matter. Nothing should make another surface react to invisible mesh.

is the mesh overlaping into the skin?

Probably a little at the base of the lashes, but nowhere else. And again, it shouldn’t matter what fully transparent mesh does. That’s the whole point of making it transparent. If other surfaces are still reacting to it, then the transparent material isn’t working properly.

Having done a bit more testing, I can say the problem is how SSS reacts to transparent mesh in the same object.

transparencyProblemPrim
The above is a test with a simple primitive surrounded by a visually complex but also simple transparent mesh. The invisible material is just a single Transparent node on white. The SSS material is a Principled node with the SSS on 1, the SSS color desaturated to 0.2, and the radius brought down to 0.01 for all elements and all other settings on default. The one on the left clearly displaying the problem has those two meshes in the same object. The one on the right shading properly has the two meshes as separate objects. There is no other difference between the two.

I’d like to find a hack to avoid the problem for now, but I’m wondering if this is still an issue in 2.8?

wait,have you downloaded the blender version with the old compiler?i dont know if this could be the same bug behavior,but in this developer thread, the new AO shader shows strange results, with the old compiler blender version.with the new compiled version the bug was not there.
maybe you should link your thread there,maybe its a bug.
edit,in the link you are posted seems the same bug behavior,maybe the bug is not fixed yet,so the best would be you report it,otherwise its got no fix if no dev knows it.

OK. So I downloaded the 2.79b version to see if it was any better, but no dice. I didn’t try the 2.8 alpha, because I figured that would be more complex and they seem to only be taking 2.8 bugs that crash right now. I don’t think it’s the same as the AO problem, because I did some tests with the transparent mesh as glass (transmission 1), and not only did I not see the same problem (though it’s hard to tell with all the shadow and such that come into play), the two versions appeared to behave the same. So I submitted the bug with a blend file of the two spheres and their surrounding transparent meshes.

I have to say that my past experience with bug reporting has already been phenomenal. One was fixed within hours, the other well within a day. Just amazing. I know they’re working at an insane rate on 2.8, so I’m not expecting a fix any time soon. I just hope they can figure it out for 2.8.