I don't have the answer, but I've asked different folks about this in the past, and I've gotten a wide range of answers that just leaves me back at square one.
A few years ago, I ran a pair of omnis on stage lip for a Meat Puppets show, and it was pure distortion. I was primarily trying to capture a solid stage sound, because another taper was patched into the SBD, and I thought we could synch a post-matrix, with vocals from his source, and guitars from mine. I ran a second pair of omnis from the side of the stage, and it was clean - even got a good deal of vocals from the stage monitors. Later, when I talked with someone about the distorted recording from the lip, I was told that I should have used omnis for in front of the amps.
Here in Portland, I recorded a guitar + bass duo (definitely NOT pretty quiet lite music) at a gallery, and close-miked each amp, using cards. The bass amp overloaded slightly, but not much; the guitar (which was equally loud and noisy) was clear as a bell. Split-channel separation of course, but that recording could easily be their next album. I did not use omnis because I knew the bass would be extremely loud, and I wanted to mic the amp, not the room. The room itself was a concrete echo chamber, and I figured that using the omnis would pick up a lot of the echo, which I wanted to keep out of the recording.
I just recorded two bands at a house show this past weekend, using a pair of cards - a noisy industrial group, and a whisperingly quiet semi-acoustic duo. The industrial group was recorded front + center, mics panned about 90 degrees. For the duo, I ran one card at a weird cross-angle in front of the vocal speaker and the guitar amp, and the other mic at an opposing angle in front of the bass amp and the guitar amp. Forgiving a little channel separation, near-album quality, keeping the audience sounds to a minimum. The reason I did not use omnis is because I wanted to keep the audience out of the "mix", since there was no real division between "stage" and audience, and my goal was pretty well accomplished.
Okay, I know these are all different stories, and I certainly have more, but the point is that in every instance, I get drastically different results. Nothing consistent from each set-up to the next, and my overall consensus is that it's all a certain amount of trial and error. I'm not crazy about that idea, but it seems to be the case.
What I'm wondering these days, is when I have any use for omnis at all, since I'm not really interested in hearing the audience, or even the acoustics of most rooms, which generally sound pretty bad to my ears.