The manual states that the choice of sound for the "granulator" will affect the character of the sound that the engine produces, but I'm not understanding how. Maybe I didn't read carefully enough, but could someone explain in a little more detail what exactly the granulator is doing to affect the sound in this system? How does the sampled sound itself and its parameters (i.e. grain size, density, speed, etc.) contribute to the physically modeled sound it drives? Apologies for any misconstrued terminology.
The granulator is the input to the physical models that follow. The input to models is also called the exciter sometimes. The exciter determines what comes out of the models, because for the most part only frequencies that are in the exciter can be in the output.
I say "for the most part" because a nonlinear model can produce frequencies not in its input. However nonlinear effects here are a small portion of the sound.
It may help to think of the models as collections of delays and reverbs. fundamentally there is no difference.