I'd like expand a bit more on what I posted above, to address a basic issue of the 'adjustable' aspect of the M/S stereo microphone technique, which I find is not generally well understood.
One of the frequently cited advantages of M/S is that the pickup pattern and angle between the virtual microphone pair are adjustable after the recording has been made. That's true, but as explained above the relationship between pickup-pattern and angle is always locked, and you cannot change one of those things without also changing the other at the same time. The real problem with that is not so much that the relationship is locked, even though it would be nice to be able to adjust angle and pattern separately. The bigger problem is that the locked relationship between the two is the inverse of what we'd really like it to be. At least if stereo is the primary intention of the recording rather than mono.
[self quote from my previous post above]
With a mix ratio of 100% Mid, 0% Side, the virtual microphone output is two forward facing cardioids with no angle between them. With a mix ratio of 0% Mid, 100% Side, the virtual mic output is two sideways pointing figure-8s. So if your mix ratio is mostly Mid, you get a virtual microphone pattern closer to cardioid shaped without much angle between them. If mostly side, the virtual mic patterns are closer to 8-shaped, with a large angle between them. At mix ratios somewhere between those, you get virtual crossed hypercardioids at one particular angle and crossed supercardioids at a narrower angle.
In terms of stereo, it would be much better if it worked the opposite way, so that the pickup patterns became more figure-8 like as the angle between virtual microphones became smaller, and more cardioid-like as the angle between microphones became greater. That's the basic relationship between angle and pattern which we play off of each other in selecting appropriate stereo pair configurations, ignoring the additional aspect of spacing between the microphones of course since M/S is always coincident.
Because that relationship is locked in the inverse of what we'd like, M/S usually only really works well at one setting within in a rather narrow range of it's mix-ratio 'range of adjustment' for a stereo output. The often touted advantage of being able to flexibility adjust angle and pattern after the recording has been made isn't really as great as we'd like it to be when the application is a main stereo pair. If it works well near that one optimum mix-ratio great, but it's not because of any 'M/S adjustablility', it's because that one specific angle/pattern combination is appropriate for the recording situation.
That's not a problem when using M/S as a way of adding a bit of optional stereo width to a mono channel, where the goal isn't achieving a great stereo image on it's own, but rather smoothly and predictably adding stereo width and dimension to a mono channel without corrupting the mono channel or introducing phase complications.