To look at it in an extreme way, if you plugged in one of those little $5 mikes you get with a Sound Blaster and ran the Antares Mic Modeler on the signal, would it sound like a Neumann? The software has to have a decent signal to start with. It cannot guess what was there in the space that a Neumann would have picked up and add it in... It's a bit like scanning a really grainy photo. You might be able to scan it at high resolution but how can you expect to know which light levels should go in between all the grains? The software can make some educated guesses but that's all.
You need a reasonably accurate mic to start with, with as little coloration as possible, to really get something out of a mic modeler. Then you can do neat stuff. But as they say in the computer world, "garbage in, garbage out."
-AlChuck