Arcane ramble-
I have a pet theory about the problematic sound of more directional patterns - A more natural sounding recording results in part from microphones with better balanced frequency pickup across their off-axis region. I think few would argue with that. But I think what is really important on a more basic level isn’t the behavior of each mic in isolation, but rather the collective response of the entire mic array.
I think the contribution of other mics in the array can mask problems, if arranged so that the well-behaved on-axis region (higher in level) covers the less well behaved off-axis (and lower level) region of other mics. By that I mean that a well balanced frequency response in all directions is what I think is really the most important thing. In other words, the combined polar patterns can make for desirable level differences which reduce the audibility of undesirable frequency differences. That can be achieved by using great quality mics with exemplary off-axis behavior like the MK41, or by the strategic arrangement of multiple mics to achieve a similar end result as a group, taken as a whole. Of course it’s a lot easier to use mics with great off-axis behavior to start with, and that’s the only way to do it in 2-channel, since two microphones can not fully cover the misbehaving off-axis polar regions of each other.
However, since I’m recording 4 sometimes 5 or a couple times 6 directional channels and recently decided after recording and listening that masking effects can go a long way in helping make up for substituting a less 'good' center cardioid for a much better one, I’ve convinced myself that masking had a larger effect than I though it would. In that example I substituted AT933 cardioids for Gefell M94s as center and back mics, without changing L/R mics, and it was suprisingly not as noticable as I thought it would be played back out of the R44 both hard LR mixed to 2-ch and multi-channel. Not nearly as much of a downgrade as I imagine it would have been if substituting the ATs for the Geffs in straight 2-channel, which I guess is somewhat obvious when mixed to 2-channel, but suprisingly not when played through individual speakers either. It was ear opening, at least that test.. any way I'm thinking I might be in search of 'good enough' light-weight low powered cards and hypers to play around with as a center plus rear facing pair for the 6' spaced LR 4060s in a compact 5 channel surround rig into the DR680.
Although earlier I mentioned using the TLs in hypercardioid mode more often recently, I should clarify that I’ve never used them that way as a 2-channel stereo pair alone*. I pulled out the TLs again and started using them more frequently because I wanted increased channel separation in multi-channel surround configurations. When adding a third (center) channel to what was an optimized 2-channel stereo pair, with the idea of playing back the recording across the same stereo soundstage width, no matter if that playback is done way of two or three front speakers, I want to narrow the stereo recording angle between each mic-pair segment to something like half of what it was previously when using only two mics.
To achieve that, I can angle the mics apart more (but Left and Right then point sideways or even backwards,) and/or space them apart more, and/or increase the directivity of the pickup pattern.
I picked up a 3rd ADK TL so I could have three identical mics to make L/C/R groupings- in particular, hypercardioids at reasonable spacings which didn’t require the mics to be pointed in absurd directions to get enough channel separation. The other ways I’ve found of doing it which also work optimally require me to space Left/Right omnis 6’ apart with a directional mic in the Center. Using hypercardioids I can space Left/Right a couple feet apart and point them more forward. Even then, the theoretically optimal configuration would have Left/Right pointing directly to the sides, which I’ve tried with Gunther Thiele’s OCT setup [explanatory links-
DPA,
Schoeps,
Thiele PDF] (who suggests using MK41s specifically because of their good off-axis response since the Left/Right mics are pointed 90 degress to the sides). Even with the TLs, which as twin diaphragm LDs I’m sure exhibit off-axis irregularities which Joe pointed out earlier, I can hear better L/C/R imaging and cross-channel seperation (less L in R and R in L) provided by OCT in a couple recordings I’ve made that way, but it doesn’t address practical issues of the situations I record in. So I put the mics in a somewhat similar triangle, increased the directivity of the center mic from a cardioid to a hypercardioid and angled the Left/Right mics forwards ~45 degrees as a more practical solution. Maybe that's more like a small-dimension Decca tree using hypercardioids. The point is that I wouldn't want to put those three mics that close together, or that minimally angled apart, without them being hypercardioids.
----------------------------------------
* Note to 2-channel X/Y aficionados- If I wanted to record a 2-channel X/Y stereo pair (and not specifically X/Y Blumlien, or X/Y as mid/side) I’d almost always choose hypercardioids over cardioids. An exception would be close mic’ing something I wished to be stereo but completely phase coherent, but then I’d probably still prefer M/S unless decoding it posed a practical problem.
[edited to fix it]