There seems to be a problem with normal maps in the V-Ray production renderer. If I change the bump intensity of a normal map, always the whole reflection layer is offset. With RT everything is fine.
I´ve attached a testscene which explains it a little bit better. Just render the image with different bump values and compare the results. Take a look at the reflection of the Vray light, the position changes with different bump values.
So it´s impossible to make layered material with different values, because the reflection will be offset.
If you try the same with RT, it looks correct.
Nomalmap_problem.zip (1.24 MB)
Thank you very much for the feedback and the attached scene and screen-shots.
We’ve reported this issue into our bug-tracking system and we will let you know when we have some updates about it.
Hi,
I may be of help here since I had the exact same issue. In the Bitmap loader of the Normalmap make sure your Gamma is set to 1.0 (or was it 2.2?). EIther one, change it from one to another and it should work fine.
Thanks Art48, this was my first thought too, but changing the Gamma doesn´t fix this. Maybe you could try the attached scene to double check this.
That’ss really strange. That exact highlight shifting was fixed for me by doing the right input gamma setting.
Yes this definitely looks like a bug to me, especially because it renders correct with RT.
Ah okay, that is strange. Maybe it is connected with the legacy Gamma settings. They still seem to be slumbering inside max, from time to time the gamma input/output settings come up even though the settings are no longer visible. Did you render out the normal map by yourself?
Yes I´ve created the map with the Nvidia normal map filter in Photoshop.
This happens because of the non-uniform scale for the object. Will try to fix it, but in the meantime, if you apply an Edit Mesh modifier and scale the vertices instead, it should be fine.
Best regards,
Vlado
Thanks Vlado, thats interesting, because I usually never scale objects non-uniform. I´ve created this object only for testing various materials. 
Thanks for the info - I dont usually scale non-uniformly, too, that’s propably why I didnt have this problem in that context yet.