Yeah I really dig the rear-facing channel. I've said it here many times, but almost every time I'm tempted to leave it out, I find that things are just not quite as good without it. I keep meaning to experiment more with an additional rear-facing channel, but I value what the addition of other microphone channels do more, so tricks such as we are discussing which "de-monoize" a single rear-facing channel will remain attractive unless I move to recording more than 6 channels.
"De-monoizing" the rear ambient material, either by manipulating the single recorded channel or by recording more than one rear channel seems to help with a couple things- It keeps that material from imaging strongly in the middle and thus directly competing with the forward-facing channel intended to monopolize the monophonic center-portion of the image; and it conveys that material in a more more diffuse, wide and enveloping way, which is appropriate as it ideally consists of all reverb, reflections, and audience reaction, with as little direct sound leakage from the front in it as possible..
I am curious how would compare the Mid/Rear and a pair of rear mics in XY.
Stick with me here and I'll get to that..
TLDR- Other than one test mentioned below, I've yet to move beyond a single rear-facing channel mainly because any additional recording channels have proven more valuable to me when pointed elsewhere, at least when recording up to 6 channels total.
IME, the primary requirement of the rear-facing channel is minimizing pickup of direct sound arriving from the front. Next is appropriate time-alignment, such that the direct sound from the front which "leaks in" to that channel will not be smear the imaging provided by the the primary microphones when everything is mixed together. That happens automatically if all the microphones are on the same stand and the rear-facing microphones aren't distant from the others. It doesn't have to be perfectly aligned though, as it does for a typical matrix of multiple forward-facing mics- and I suspect a few milliseconds (corresponding to a foot or two between the front/rear pair) is helpful for differentiating the front from back, "keeping the front sound in front" via that slight time of arrival difference.
^That's most definitely the case for discrete surround playback where the rear channel is routed to speakers behind the listening position. In that case the front-bleed into the rear channel, combined with insufficient front/back spacing perceptually pulls too much of the front direct sound into the rear surround speakers. A little spacing provides time-of-arrival cues which helps keep the direct sound from the front sounding like its coming from the front, such that more rear facing channel ambience can be used before the front imaging shifts rearwards. And for quieter sections where the front sound isn't considerably louder than the sound from the rear, it also helps keep the "rear sound in back", and better differentiated from the front.
However, I'm not as certain about the front/back microphone spacing when mixing to 2-channel L/R stereo. I know that the spacing I am using works nicely in my 2-channel mix, so I stick with it because it works well for both surround and stereo, but I'm not certain if less spacing might be better or not for 2-channel stereo alone. It could be that having only level differences between the front/rear channels and no time of arrival differences will make for cleaner, less smeared front imaging. It's probably a trade off depending on how much front/back level isolation can actually be achieved, because in a practical sense, the greater the front/back level difference there is, the less time-of-arrival difference should be required. Varying the front/rear microphone spacing is something I encourage those of you trying these OMT setups to experiment with, including trying a coincident front/rear arrangement. If you do so, please let me know. From a practical point of view, reduced front/rear spacing will make for a more compact and potentially less visually intrusive setup.After minimizing pickup of direct sound arriving from the front and achieving appropriate time-alignment, the third most important thing is the "de-mono-ization" of the rear-ambient pickup we've been talking about. I'd probably not choose a rear-facing X/Y pair for "de-mono-izing" the rear ambience because X/Y with cardioids is center mono-heavy unless you use a very wide inclusive angle, and then you are going to get lots of front leakage into the sides of the cardioids. X/Y supercards would be better, angled such that their shared null-axis points directly forward, but if using a pair of cardioids or supercards, I'd rather just space them out on the same bar which is supporting the omnis and point them directly rearward or maybe with a slight outward angle, keeping their minimum sensitivity axis facing towards the front.
I make a practical exception for a Mid/Side rear-facing pair (by the addition of a coincident Side fig-8 to the rear-facing cardioid/supercard), partly because it means the simple addition of the figure-8 to the already existing rear-facing microphone without otherwise changing it. If the rear-microphone is a cardioid, the pattern nulls of both the cardioid and the 8 face directly forward and regardless of the Mid/Side output ratio all virtual output pair combinations will likewise feature a forward facing null, helping to limit pickup of direct sound from the front. This is essentially the same as rear-facing X/Y supercards/hypercards angled such that their shared null angle faces forward.
^ I made some test recordings at Wanee year before last to experiment with front and rear facing Mid/Side pairs, but have yet to do anything with the rear facing pair. In that case I placed Naiant X-8S fig-8's coincident with my standard front-facing supercardioid and rear-facing supercardioid, in order to turn them into front and rear facing Mid/Side pairs. I recorded the rear-facing pair to the stereo channel of the DR-680 through a V3 feeding the digital input of the DR-680 so as to be able to record all 8 channels time-aligned on the same machine, and only messed around with comparison between the split omnis + a coincident Mid/Side stereo center, versus split omnis + a near-spaced 3-channel L/C/R center setup, as well as various combinations of the two. I did so via a 2-channel mix directly out of the recorder using its Mid/Side playback function option. The conclusion was that either center setup works well in 2-channel mix, they sound different from each other, both represent a significant improvement over a single center channel, and both together in a 2-channel mix is not necessary. Also that the 3-channel near-spaced L/C/R center setup is superior for playback over 3 front speakers, which is not surprising.
In doing that comparison, I was struck by the absence of the rear-facing ambience channel, since in the direct-off-the-DR-680 monitor mix the stereo channel cannot be played back simultaneously along with the other 6 channels. I missed it! And I've yet to get around to playing around with all 8 channels in the DAW including the rear-facing Mid/Side pair.
I update the link for phase plugin. I don't know if the plugin is high quality. But I tested it on the sum of mono. It is working very well. When decreasing the phase, the mono signal increases. And all is canceled for 180°.
Good test, which seems to confirm it is working correctly.
wfowumbo reminds me that this kind of
equal phase rotation by frequency manipulation is called a Hilbert filter.