Quote - Hmmm. Not sure what to say. But go check this post in the node cult. This is a new technique that works great for the two-sided detection.

I suspect that we did not turn on normals forward in the GC diffuse node. That's why it would behave differently.

But - given what we just learned in the linked thread, we have a new and better way to do the detection!

bagginsbill does the link work for you?  i can not open it. and  i  also cna not find it on the forum. did they delete  the thread?