One other point to consider is the final render size and how many of the pixels in that final render are going to be used for your object/texture.  In RedPhantom's example above the higher texture is visible in the output as most of the screen is filled with just that image.  On the other hand, if those skirts were 20 feet back from the camera and only occupied an area of 100x130 pixels in size then both would be scaled downwards and little to no detail would be readily visible.  Think about a leaf if you did a close up of that leaf then a 2K image would be in order to convey the full details present.  However, if you rendered a forest and the trees had that texture on each leaf and there were thousands of leaves then you would get maybe 2 pixels to convey the leaf and all the detail in that 2K image would be compressed down to 2 pixels.  So it comes down to a balancing act between visible detail and render resources.  

As far as UV to image maps, A UV map (Normally) is in the range of 0-1 both up & down and left to right.  A value of .5, .5 in the UV map would be dead center of the image map.  If the image map was 500x500 then the center would be 250,250.  If the map was 2000x2000 then the center would be 1000.  The center point is the same just how many pixels there are to work with.  If the 500 image map was used to make an image of 2000 pixels then each pixel would have to used 16 times (4 times each direction).  However, if the 2000 image map was used then it would be a 1-1 rendering and look the sharpest.

Poser 5, 6, 7, 8, Poser Pro 9 (2012), 10 (2014), 11, 12, 13, 14