This test model has 3 sets of columns, in the left column of each set is a geometry model that has a bulge and in the right column of each set is the equivalent but using a normal map instead of geometry. This is equivalent to baking geometry detail into a normal map, and thus the resulting normal map, with the exception of shadow and inter-reflection effects, should look equivalent to the original geometry.
But with V-Ray it doesn't look equivalent.
Reference image from Three.js:
Current V-Ray result - in the normal map versions, the bulge is not as pronounced, you can see it especially in the reflections in the gold column:
This is a test model located here: https://github.com/KhronosGroup/glTF-Sample-Models/tree/master/2.0/NormalTangentTest
This test model has 3 sets of columns, in the left column of each set is a geometry model that has a bulge and in the right column of each set is the equivalent but using a normal map instead of geometry. This is equivalent to baking geometry detail into a normal map, and thus the resulting normal map, with the exception of shadow and inter-reflection effects, should look equivalent to the original geometry.
But with V-Ray it doesn't look equivalent.
Reference image from Three.js:
Current V-Ray result - in the normal map versions, the bulge is not as pronounced, you can see it especially in the reflections in the gold column: