Random thoughts floating in my head that will hopefully come out intelligibly. Also note that I've never had a chance to play with Dyn EQ/Vol, and don't have the cool selection of target curves that many others do. I know that various mftrs have their own curves, I think Marantz had an exclusive one or two, NAD definitely did, and so forth. The House curve aka Audyssey curve has an HF rolloff if I'm not mistaken.
When Audyssey is selecting a xover point for you, it is a reference of sorts for you, and is trying to give you an idea of what is "usable". It is finding the F3 of your speaker, and then applying the filters down to that point. (This tech applies hundreds of filters, btw, I don't know how much more it gets with XT32 and so forth.) IOW, if it finds a speaker at 40hz, it will correct down to that point. So, if you xover at 80hz, you're golden. If you go full range, you're poopie for everything below 40hz, because Audyssey thinks it can't even reproduce down there, so it didn't even bother trying to correct for that range. The point is, it's always up to you to decide where xover should be, and it is correcting for whatever range is even "usable" by the speaker. Make sense?
Distance; if there is anything that everyone could possibly agree on with this tech, it is distance. Last I heard from the horse's mouth, it is more accurate than 0.1' tolerance, but the real figure was confidential many years ago. If you think your distance is off, think hard about how different your mic position is to your seated ears. Also, the first position will be doing the levels and distances, I'm guessing, gee it's been a real long time since I've talked about any of this.
Subwoofers trip up many people, "OMG, it was off by 20ft! What a POS!". A tape measure can only do physical delay/distance, Audyssey is also accounting for electrical delay, and the reason why subs often so much more is because of all the EQ/DSP in them.
The reasoning behind doing some of the level cal and what not before running Audyssey is that hopefully whatever processing power your unit has is more free to do some other and more fine work. How real this is, I have no idea, but obviously at least a few found better results having done so- I still don't think it guarantees better results for any given person.
Lastly, I've always wanted to play with those "Dynamic" features, as I had a different impression from my readings. As so very, very, very few people listen at reference level, dialogue intelligibility could be increased in the best case scenario. I mean, when I read rmk stating how he thinks it helps out, and he has those freaking industrial grade things going on full blast in his room, it piqued my interest. But anyway, my $600 refurbed Onkyo is still going strong as ever after almost half a decade is it already(?), so no Dynamic nuthin' for me, for now.