Hi Tanuj
In theory, I believe the 'correct' approach would be to route individual VE tracks/instruments through VE busses, and apply Altiverb (in stereo->stereo format) at the bus point. Obviously you would need to pan the individual tracks/instruments correctly, and leave the busses panned 'open'. This way the instrument is panned to it's correct stage position prior to being treated with reverb. Obviously, you should then NOT pan your instruments in Altiverb, and NOT pan the signal which leaves your bus. Also, assuming you want the true sound of the hall (or other sampled space), you should set Altiverb up so that you hear only the wet signal.
That said, I have found this approach often results in signals which are not panned extremely enough. For example, even if I pan my 1st Violins fairly hard to the left, and reduce the width of field to be fairly narrow, the signal which comes out of Altiverb after reverb is applied, masks a lot of the panning information - ie. the violins don't sound clearly like they're sitting to the left of the stage, the reverberated signal almost makes them sound in the middle. Perhaps this is a shortcoming of the Altiverb impulses, or perhaps this is the way the sampled spaces really behave, but it doesn't sound quite right to me. I'm still working on a way around this, as panning the reverberated signal is technically 'wrong', if one cares about being accurate.
Personally I don't like using the stage position setting in Altiverb to move instruments further forward or back - it just doesn't sound natural to my ear. I tend to use different samples from the same space for different sections, in order to achieve the sense of depth - ie. use a position of say 8m for the 1st violins and celli, 12m for the 2nd violins and violas, 16m for the brass and percussion etc. Not strictly accurate either, and certainly less flexible, but it sounds more natural to me.
I'd be interested to hear the techniques of other Altiverb users too.
Jules