The best figures, like the best rigging, do not follow the physical models found in real life.  Meshes are hollow, human bodies are relatively solid.  (A little squishy in places, but not hollow.)

 

Rigging in the places where human joints exist seldom create the best joints.  To mimic a solid body at the skin level with just a skin, the joints have to be closer to the outer bend, even with weight mapping.  If you place the joint where it exists in a human, you get less than optimal joints.  This has long been a problem in 3D, people thinking that the joints need to match the location in the human body.  The same principle applies to morphs.

 

So a realistic skin on a base figure is not going to be optimal for rigging or morphing.  It is easier to expand the vertices than it it is to shrink them and maintain the best mechanics of the figure.  So a base figure should be on the skinny side, a little anorexic, a little athletic, for the sake of mechanics.  

 

And like I said, Poser and DS do not have rendering engines powerful enough to touch realism.  They can get moderately close, but they aren't going to touch the top dollar rendering engines that can truly achieve realism that suspends visual disbelief.  Their strong point is representational art.  (If you're looking for realism, go spend some money on software instead of content.)

 

(And yes, Jan, modo is the best thing that happened to 3D, IMO.  It takes a while to learn, because it is different, but different for a very good reason.  It was no small coincidence that it was used to model Dawn.)