AFAIK there is no direct correlation between the size and quantity of texture maps you feed Poser and the subsequent render speed and memory usage. As Stewer posted in another thread, Poser coverts all textures into tiled mipmapped 32 bit OpenEXR's prior to rendering. In the one test I did, the difference between rendering with an additional 4k 8 bit normal map or not was: no change in render speed (averaged over 3 renders) and an increase in RAM use of 20-30Mb when the extra map was used. 

You cannot derive every map type required from a single diffuse map without compromising a lot on quality. Each map type should be considered separately for what it adds to the final render - specular, colour, diffuse, height (bump or displacement), normal, opacity - all have specific purposes, and each should be made with that purpose in mind. Converting an 8 bit JPG diffuse map (as that's what seems to be most commonly used in Poser content) into a specular or height map in the material room can be done, but the results would be better if those maps were made properly in the first place.

On UV's - the best thing from my POV would be if Poser had a means of swapping UV data without affecting anything else, eg. being able to associate UV data with a material collection. Sometimes a single map is best, other times multiple tiles is the way to go. Then you've got the tradeoff between efficient use of UV space vs seams vs texture distortion ....

There is no one right way in 3d, period. The only consideration is what works best for the intended end use - what works best in a realtime engine for games is different to what would be best in an offline render for print or film; having multiple seams to maximise UV space and reduce distortion is a bad choice if you're using normal and/or displacement maps etc - it's all about the context.

----------------------------------------

Not approved by Scarfolk Council. For more information please reread. Or visit my local shop.