It might just be me, and my ears, but there's something about playing dense chords on sampled pianos vs. live on a real piano (and definitely not singling out VSL, here) that makes me hear things as somewhat distorted. It just seems to me that all the individually sampled notes, played together, creates what I perceive to be distortion, that I don't hear on a real instrument when everything is combining on one, single soundboard at that precise time. I don't have a scientific explanation, but I would imagine that in the combining samples notes, we might be getting an extra quantity of resonances that we don't get when it's all simultaneously on one soundboard. It's more noticeable to me with brighter instruments. It would be an interesting to research (set up A/B comparisons of combined notes vs. notes played together on the same piano, with the same mic setup).
I also seem to think that mixing multiple mics makes things worse.
Hi, this is not my field at all but as a physicist I could share with you some reasonings.
In general the instruments are highly complex physical bodies with a lot of complex interactions between their different parts. So, of course, just sampling the different notes and then playing them together does not keep count of these interactions.
One thing that comes to my mind is that our instruments are tuned with the 12 intervals evenly divised. That means that, in general, the frequency of each note is calculated from the semitone lower note by multiplying the frequency for (2)^(1/12). So, the SOL of the piano has not 3 times the frequency of a lower DO.
However, it exists a physical phenomen called synchronization that you can see in this video:
I do not know exactly how it enters in the physics of a piano with many keys pressed but my intuition says that it is connected.
As someone said next, there was a group that preferred to sing together. I think the explanation is the same. By singing together they are more likely singing notes that are harmonically connected.
There are also the resonances that take place, especially when the sustain pedal is down. It's not the same phenomenon even if it is connected to synchronization.
I know that some sample libraries try to take account of resonances, but when there are many notes the resonances can become quite difficult to calculate.
Do not take this as the true explanation, these are just thoughts that I had by reading the posts in this thread.