well the base thing is: smothed uvs adapts the uvlayout to the actual subd level of your object…
If you have lets say lips or an eye and you paint a texture for it, It is depending on your workflow what you use. I am ususally forced to use morph targets in zbrush and a low res mesh as base for paralel scene enhancements while I am zbrushing, or animation is done at the same time.
This is case A, faster, but inaccurate and you will not get the best result from the nice zbrushing…
Case B means: you have the freedom to zbrush, and start texturing AFTER it and you will be able to use subd lvvl 2 or 3 (depending on what you do) as base.
I case A Textures and the maps are calculated based on lowres mesh, meaning if you paint the edge of an eyelid or lip based on an Uv snapshot, you usually use an edge as referense, meaning you will paint a lip that has an somewhat edgy character. If you subdivide this, an uvs are smoothed, the edge of the lips you painted so accurate will wander around, meaning it is shifted somewhat due the smoothing.
turning smooth uvs off (in 3D app) ensures that it tays where you want it.
You could still use smooth uvs and do some guessing and texting with textures, but if the zbrushmaps are not in place, it will easily give artifacts (borders and edges have this nice zebra pattern)
If you are lucky and have case B you can simply work with the smoothed uvs for the final level you use, meaning you can paint a more round lip eyelid or whatever and be happy with it. the worst shifting happens during the first 2 Subdivisions, so if you are working on 2 or 3 and have to go up for some passes, it will not matter that much.
I personally prefer case B, cause who wants to paint edgy textures that ought to be round??
please correct me if I am wrong