Greater fresnel values on inverted normals

I don’t know why it happens, but it looks like there’s more reflections on the inverted normals side of geometry (I’m not using solidify modifier because I don’t want to increase render time). I calculated all fresnel values, and combined it properly, but still, there’s something wrong. I’m using normal map that I also inverted using RGB separation node, inversion of red and green channel and combining them back again which should work. I guess it’s all because normal maps doesn’t work properly on inverted normals geometry. Do you have any idea what I could do here, exept using solidify modifier?
Here you can see some screenshots of how it really looks like on that node tree preview:
Positive normals:

Inverted normals:

A) you have cropped the nodes, so we can’t even see them
b) if the wing has no thickness at this viewing angle there are virtually no fresnel angles anyways
c) stop inverting normals and separating rgb values , you’ve got some crazy stuff like feeding fresnel into color. Your jacked up noise texture as a mix factor is most likely making all your fancy node stuff irrelevant.
d) post the blend, packed

you want a butterfly wing with a normal map? Or is your goal to use fresnel for a blue irradiance? Either way you only need a handful of nodes.

if you’re trying to have the same IOR on the backface as you have on the frontface, then your IOR is wrong… as the inverted of 1,650 is 0,606 and not 0,350 (1/1,65=0.606).

Thanks for help with fresnel value. It worked. Now I have only one question: How to make normal map work properly on geometry with inverted normals?

depends on how you want the inverted normals to work. you could simply invert the normal map or just invert the blue channel. try it with lights and choose the one you need

mapping > normal > scale z=-1