It seems that in a live performance one is very rarely (if ever) as acoustically aware of space between musicians or their three dimensionality. My suspicion is that hifi compensates for the visual stimuli that is provided in a live performance with the over emphasis of these audible features. Small ensemble acoustic or vocal performances might be the occasional exceptions to this.
Its been my impression that there are several things that can contribute to this- how a recording is made (many are un-natural) and how well its details are reproduced. For this subject though, let's throw out the un-natural recordings, at least for now.
What I have have found is that as a system looses the ability to reproduce low level detail (and consequently ambient information) the result is that the musicians in the ensemble tend to have a quality if a cardboard facsimile rather than the real person. So spaces between the instruments are larger, making the individual instruments seem to be in greater relief.
Now as the system attains greater ability to reproduce ambient information, that information tends to flesh out the performers and fill the spaces in between. This causes the soundstage to *seem* less distinct, until one realizes how much more reflection information is being reproduced!
Of course in a real music situation, the ambient information is seamless with the instruments themselves, but I think a lot has to do with the space in which the music occurs. Having done a lot of recording and listening in such spaces locally, I don't think I can agree with the comments in quotes, at least with the recordings I have heard- I hear the imaging information quite easily in those rooms and halls, whether live or recorded!