How does this affect "slightly wet" sample sets? e.g. Haverhill, which is not particularly reverberant, but is wet.
Since the Haverhill is a wet-recorded sample set, in the sense that it's recorded with early reflections in the samples, the signal mixing effect (unnatural phase addition/cancellation) shouldn't really apply.
Likewise, because early reflections are already contained in the samples, the pipes should already sound distinct (perceived clarity of the plenum), at least when heard in a dry/near-field listening environment.
So in some respects there would be less need for additional audio channels.
However, of course, playing a wet sample set in a reverberant space has its own set of compromises, mainly in that:
1. The organ will sound wetter overall than the brain would expect for the listening environment, which will detract from the overall perceived realism. Because of this, you would probably want to go for a sample set that's as dry as possible overall (even if the sample set is wet-recorded - with early-reflections, like the IA PAB). Testing the Haverhill in situ will probably give a reasonable idea of whether it's too wet or not to sound convincing in the building.
2. The real acoustic will add another set of early reflections to the sound coming from each of the speakers, which will tend to make the brain localise the sound to one one point (the loudspeaker) again, to some extent undoing the benefit that the wet samples brought (unique patterns of early reflections for each pipe). For this reason, you would probably still want as many audio channels as possible.
3. The early reflections contained in the samples will be heard combined with the early reflections of the real acoustic, which will make it more difficult or impossible for the brain to construct an intelligible model of the shape of the listening environment, which again will make it sound less realistic overall. I.e. adding early reflections to sound that already contains early reflections usually won't lead to an overall set of early reflections that could describe a real building, with the result that it will sound less natural.
A true dry-recorded sample set with one speaker per pipe is probably the ideal for reproducing a sampled organ in a wet listening environment, but that would probably need a similar amount of money and space to a real pipe organ. So you have to choose a reasonable balance of compromises.
Generally, for use in a wet listening environment, you probably want to look at either a true dry-recorded sample set, or a nearly/fairly/semi-dry but wet-recorded sample set. I'd suggest as many speakers as possible in either case, but you might perhaps get away with less in the semi-dry case.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.