Topic: Oddball microphone techniques - part 1

[mfrench]

Busy times here, getting ready for the fire season to erupt (brush clearance, fire fuel abatement touchups).
Hopefully your head is above water, and you're out of the wind.

Apologies for not replying sooner. I'm still trying to wrap my head around the complex way off doing this mix.  I've tried the simple method of just flipping polarity in the right channel, and then making micro-adjustments to all tracks, and have found a nice place.

So, I looked at the concert program for this particular show, and, followed that up with a search for one of the bass soloists, Andres Martin, and I found a video (two parts) on Youtube that shows the same show that I'm working on in this mix. And, in fact, you see my DPA4060 pair, and 8" baffle, and my partners Royer ribbon. The ribbon is at the stage lip, and my baffle is back at first row, about 4' or 5' apart.
You cannot see the DPA4028 up in front of the baffle, as it is out of view in that perspective.  But, it gets you an idea of what I' looking at in part of the mix. In the second half, I have two old school jazz players, sax and drums, on stage. So, I have a very strongly populated center image to deal with and sort out.
This is from the first set, and is a Bottesini bass duet/ w surrounding Hutchins players. I do not know the audo source; its better than most vid cam sound tracks. Nope, its brickwalled.
https://www.youtube.com/watch?v=aNSHWxe72yA
This is the baffled mic array in the video: http://taperssection.com/index.php?topic=96009.msg2239276#msg2239276

Challenge:
In the first set, getting mid-range players into the mix, while maintaining articulation of the bass duet soloists.
In the second set, having a tenor sax and drum player in the center field, both as acoustic players, articulation between those two solists, and surrounded by the frequency range sweep of the consort.
Its been a fun experiment.

my view of the same concert in the vid, below:

[Gutbucket]

I haven't posted much in this thread recently.  I've drawn conclusions and logical extensions from each of these oddball techniques, all of which has informed the next incarnation to try out.  I still have variants I want to play around with, but for the most part I've settled down to what works best for me, and feel like I have a pretty good grasp of "why" this stuff works.  Yet I keep realizing more about the "why" part- both the technical aspects and the more philosophical angles, and I've had something of an "ah-hah" technical realization recently which is exciting to me because it fits a lot of the puzzle pieces together for me logically.

So much of recording is finding the right balance.  Here's a list of a few things which need to be in balance:

Tonal spectrum
Recording levels (neither to high, nor too low)
Loud/Quiet dynamics
Left/Right channel balance
Front/Back ambience
Near/Far psychoacoustic balance
Flat/Deep sonic perspective
Direct/Reverberant-diffuse sound balance

Also balancing things like:
Cost/Value
Effort/ROI
Enjoyment/Hassle
Simplicity/Complexity

I see finding an optimal balance for each of these things a key to success. I suppose its a basic philosophical tool that applies broadly in life-  Finding the good middle way.  The trick is realizing what needs to be balanced.

Recently I've realized a fundamental technical balance aspect which is pivotal to what I feel makes for a really excellent recording.  An aspect I've talked around, listened around, and designed microphone techniques around, but never really conceptualized directly in terms of "finding the right balance".  It has to do with the listening experience, and finding a balance between sonic immersion and sense of space on one hand and solid/sharp/clear directional imaging on the other.  On one hand we have the "I am there" experience, a sort of a right-brain experience gestalt thing, and a big part of the actual live music experience.   On the other we have the solid and clear stereo image aspect, presenting a seamless soundstage sonically in front of us with sources we can point to in specific places, a more left-brain experience, and one where most studio recorded material excels.

The "ah-hah" was realizing that the key to achieving both of those aspects is finding an optimal balance between the correlated and de-correlated components in a recording.

I've talked a lot about the importance of achieving low diffuse field correlation in a recording.  It's something I feel is critical and is a big part of what makes a recording sound ("feel") open, big, spacious, enveloping and "real", much like a live experience.  What I haven't realized as directly is that I've been finding an optimal balance between what should be corellated and what should be decorrelated to achieve an optimal balance of "there-ness" and "sharp-imaging", achieving lush depth with good closeness at the same time.

[Edit- Here is an excellent paper on the topic I recommend to anyone interested in digging deeper into the effects of decorrelation on stereo signals - The Decorrelation of Audio Signals and Its Impact on Spatial Imagery - Gary Kendall (1995)[/i] - http://www.garykendall.net/papers/Decorrelation1995.pdf]


More on this tomorrow and it's connection if I find time..

[Gutbucket]

Correlation is mysterious.  I'm talking here of audio signal correlation as a relationship between audio channels.

It's a measure of the phase relationship between signals.  Two signals might be identical in every way except for their phase relationship, in which case those two signals will sound identical when listened to individually.  Individually they have no particular correlation and they cannot have any correlation, the term is meaningless in isolation.  Phase correlation and all the attributes influenced by it are emergent properties having to do with the relationship between two or more signals.  It takes two to tango and this aspect simply does not exist at all until there is another signal present with which to compare the first.

This is different from most other audio signal aspect we think about - level, frequency balance, dynamics, direct/diffuse balance of the content - all attributes which exist in a single channel signal.  Correlation only comes into being when another signal interacts with or is compared to another.

Okay, how does that apply to microphone configurations and mix choices? We effect the correlation relationship between channels by choosing a microphone configuration.  We can also manipulate correlation during mixing.

What does correlation or decorrelation between channels sound like? Plug in your headphones and head over to this sample player on the hauptmikrofon site- http://www.hauptmikrofon.de/audio/diffusefield.html

First listen to each of the samples using only one ear.  Just put the headphones on in such a way that you can only hear through one ear or the other, doesn't matter which.  Switch through the various samples, all of which were made using Scheops microphones. Astute listeners might notice a slight difference in low-frequency pickup between samples which relates to the pickup pattern of the microphones used.  The MK4 cardioid X/Y sample has a bit more very low bass content then the MK41 X/Y sample (and the M/S sample) which have a touch more low bass than the Blumlien sample (MK8).  But other than that each sample otherwise sounds essentially identical, does it not? (This step is to prove to yourself that the only significant difference is the low bass extension)

Now listen again with the headphones on normally through both ears.  The samples will sound very different from each other.  The difference is not subtle.  Its pretty amazing how similar the samples sound through one ear verses how different they sound through ears.  What you are hearing are differences in phase correlation between channels.

What's best? What's the take away from this?  The samples linked above are all of totally diffuse ambient content.  Its pretty clear that microphone configurations which produce low correlation between channels convey a more realistic sounding 3-dimensional sense of enveloping space.  That's a rather important attribute in a live music recording.  But is optimizing for that quality alone the best choice?  What's the trade-off, and what's the appropriate balance to try and find? 

In short- a correlated stereo signal, otherwise identical in each channel is heard as a narrow, sharply defined phantom image in the center of the stereo soundstage between the two speakers.  An entirely decorrelated signal is heard as a broad, diffuse, and open "cloud" of sound filling the entire space, possibly extending out beyond either speaker.  In both cases the content is essentially the same other than those imaging and spatial aspects. 

When we choose a microphone configuration we are manipulating this relationship, in addition to other relationships.  By using clever microphone arrangements with more than two channels we not only gain control over signal correlation so as to be able to optimize that aspect separately from other stereo aspects, we can also gain control over which portions of the sound are recorded with high correlation and which are recorded with lower correlation.

More on that next.

[Gutbucket]

There is a generally accepted convention that coincident microphone configurations produce "sharp imaging" yet can sometimes be sort of "flat and boring sounding".  And conversely that spaced microphone configurations produce "lush open sounding ambience" yet can sometimes be "loose and phasy with diffuse imaging".  There are elements of truth in that which I'm taking advantage of, but its important to note that its more complex than that over simplification.  I have spaced omni recordings which exhibit clear and distinct imaging, and the examples linked in the previous post provide a good example of some coincident configurations sounding very open and ambient. 

Ideally we'd like to have control over which sounds produce correlated signals and produce decorrelated signals in the recording.  It would be great to record the direct sounds which arrive through an imaginary front "window" which encompasses all the direct sound of interest (like the Stereo Recording Angle or SRA in William Stereo Zoom terms) in such a way that it is correlated and produces sharp imaging, while the sounds arriving from everywhere outside that window are recorded in such a way so as to be as decorrelated as possible.

[Gutbucket]

Obviously we can choose stereo microphone configurations and microphone pickup patterns which alter these things, but it's all tied with other things which may be more important to the recording.  Is is possible to find a way to tease these things apart in order to optimize and better-balance these elements with a bit more control?

Each of the samples at the hauptmikrofon link above were made with coincident microphone configurations, yet produce vastly different ambient impressions.  For those who may be thinking, "but there is no direct sound to judge the imaging aspects of the configurations used to record those samples", that's the entire point of the demonstration.  One of the most important things to balance is the ratio between direct and reverberant sound pickup of course, primarily by choice of recording position but also by to a lesser degree by choice of microphone pickup pattern and stereo microphone configuration.  While it is true that when used in a normal recording situation these microphone configurations will be picking up plenty of direct sound (hopefully in a good balance with the indirect diffuse sound) these examples are illustrative precisely because they specifically avoid direct sound, allowing us to clearly hear the differences in diffuse pickup without distraction. 

Like different coincident techniques producing various degrees of correlation in their pickup of the diffuse field, spaced techniques produce various degrees of correlation in their pickup of direct sound.  Direct sound arriving at both spaced microphones simultaneously will produce a correlated signal in the two channels which images sharply in the center on playback.  But that only holds true for central sounds arriving from along the center median plane.  The reproduced image of those center sources will be sharp and clear, even though the sources located off to one side or another may be are reproduced more diffusely.  For this reason, the well-known "hole in the middle" problem from overly-wide spaced omnis is something of a misnomer, the problem can be more appropriately described as "small island in the middle" between two other islands at the speakers locations.  This off-center decorrelation is frequency dependent as well as being dependent on the angle of incidence.  Low frequency content arriving from not overly far off the center line will be more correlated than higher frequencies arriving from the same location.  The specific shape of that frequency/correlation curve is determined by the microphone spacing and the angle of incidence.

[Gutbucket]

When we use a center microphone and mix it's signal to both left and right channels,  that signal has identical phase and is fully correlated across the left and right channels.  In combination with an otherwise overly-wide spaced or angled left/right pair, that addition can help "fill the hole in the middle" or rather "extend the boundaries of the center island" so that we get a more solid, clear and upfront soundstage.  Call it a sand-bar I guess.  We get the nice solid middle across the front and a wide, diffuse ambience at the same time.   Works very well and sounds much like the live experience.  This is why I like and commonly suggest 3-microphone techniques around here.  Its still pretty simple, but solves many problems and works really well.   A spaced omni pair plus a directional center mic goes a long way.  Or a less-widely spaced pair of directional microphones angled widely apart plus a directional forward facing center mic.

Nakamichi understood this way back when.  I know think of their 3-point stereo recording system essentially leveraging a simple way to manipulate the balance between the correlated and decorrelated elements of the sound.


Yet it's not perfect and won't satisfy rabid fans of pin-point stereo imaging. 

If we turn that center mic into a stereo pair, and choose a configuration for that which provides high correlation for all the direct sound arriving through the breadth of the SRA window, we can fill that window not just with clear upfront sound which also has with sharp directional imaging.  To do that we ideally need some uncluttered space to fill (a slight hole in the middle) and clear sharp-imaging correlated material to fill it with.  A wide spaced omni pair fits the first requirement and a narrowly directed coincident X/Y or M/S pair fits the second.   Four channels total.

[voltronic]

Thank you so much for explaining this concept in a down-to-earth way.  I feel that I understand DFC more clearly after reading your mini-treatise.   

A few points in response:

1. I never understood the pinpoint imaging devotees.  In my experience, live music does not ever sound like that unless you are conducting or performing with the ensemble (sometimes not even then either; I've performed in really awful rooms where I can't hear the people 5 feet away).  There is always some degree of blur from even a close audience perspective, and DFC partly explains why certain arrays sound more "real" than others.

2. It should be emphasized that careful placement is equally as important as what array to choose, if not more so.  Manipulation of the direct / diffuse ratio in this way (preferably with knowledge of the Critical Distance of the room) is job 1.  I used to accept being forced outside of that zone; no longer.  It's just too, um, critical...
 

3. It's good you mention that these common arrays don't necessarily work the same in all situations.  My recordings with the best balance of reasonably sharp imaging combined with very good ambiance have been from a simple pair of spaced omnis, set close, but narrowly spaced (i.e. the "European" model).  There still seems to be a lot of reluctance to doing that on this side of the pond.  Since being converted, I find that I can cover even very large ensembles this way, with no holes / blobs in the image.  Your 3-mic suggestion or dual-array setup with wide omnis is great (a lot of the classical orchestra guys do exactly that) but I would encourage those who can go close with AB omnis to try narrow spacings (40-60cm) as well.  I'm not sure how much correlation does or does not have to do with that.

Thanks again for sharing the knowledge and furthering intelligent discussion on this board.

[Gutbucket]

Good observations, I agree.

1) Yeah, the most important thing for me in a good live recording is a convincing spatiality and "sense of I am there".  Actual live music experience is far more about that than imaging.  If I have to make a choice I want realistic spatial capture over pinpoint imaging.  But what I really want is both!  And that's what a coincident pair in the center can provide.  Sort of the best of both worlds as long as you can get the blend right so the alternate aspects get along well and don't fight with each other.

2) Amen.

3) Nothing wrong with narrow-spaced omnis as a two channel technique in the right place.  To my mind that represents a sort of compromise, as do near-spaced configurations like ORTF, NOS, DIN, or what have you.  What do I mean by that, lumping omnis with near-spaced configurations?  Each of those is a somewhat different solution for balancing a whole lot of different things.  They are either coincident, nor wide-spaced and get some of both aspects going.  But that also means they don't achieve either in a fully exemplary way.  They can't provide fully decorrelated diffuse pickup and the direct pickup isn't fully correlated either, but represent an optimized compromise using only two channels.  Their strong point is simplicity and purity.  Straight two channel.  Two mics through to two speakers.  An argument for similarity to HTRF head-spacing can also be made for near-spaced configurations.  What I'm trying to do is move beyond the imposition of those compromises by combining techniques and using more channels.  More complex, yes.  Worth it to my ear.

But for most TS music tapers the problem is not being able to get a pair of omnis spaced far enough.  Its just a practicality thing.   I've seen plenty of near-spaced omnis in the section used on standard near-spaced mic-bars.  Too far away for that to work well IMHO.  On-stage, maybe.   Back in the audience even big mic bars can only go so wide with typical mics flown from a single on one stand.  That's wide enough for a stereo pair of omnis, but not really optimally wide with a center mic or pair.  Miniature omnis make sufficiently wide spacing a bit easier as you know.

I feel what I'm describing here is different than the general wide-space classical thing with a center pair and wide flanking mics to cover an orchestra.  This is more of a single array using multiple channels.  Than an addition of wide flankers and typically used at greater distances.

[Gutbucket]

Here's another way to do it using just 4 channels, which gives up some of the pin-point imaging but substitutes even more realistic "you are there" spatial qualities-

Instead of making the center mic an X/Y pair or M/S pair, use a center rear-facing directional microphone in addition to the front facing one.  Yes, I'm going through previously discussed oddball variants here, but I have a twist this time, which is what motivated all this discussion about correlation and decorrelation.  Mixing in a touch of that rear facing microphone sent to both left and right channels at the appropriate level does a lot to further improve the "you are there" spatial quality.  Others have tried it and reported the same positive results.  We've talked about this previously, both in this and in other threads here at TS.

I always wanted to somehow "stereoize" that rear-facing contribution so it was spread out a bit and didn't clutter up the middle.  I've thought about and experimented a few times with the most obvious solution- using a pair of rear facing microphones instead of a single one, but find it hard to justify the investment in yet another microphone, recording channel for it and all that.  Plus I have higher priority uses for additional channels.  This lead to thinking of various ways of decorrelating that center rear-facing channel signal in the mix-down stage, diffusely spreading out it's contribution across the soundstage.  There are various ways to decorrelate duplicated copies of the signal which are routed left and right.  Running it through a stereo reverb is the most common way, or some other pseudo-stereo effect, but that goes against the ethos of not modifying the sound in an obvious way.  Maybe it could be done subtly enough, but philosophically it rubs me wrong. 

I started looking into the particulars of methods of decorrelation, wondering how the manufacturers of matrix surround decoders and up-mixers do it.  I found lots of pseudo-stereo techniques.  Meh.  Older analog channel matrix surround decoders use decorrelation to help spread a single monophonic or two-channel surround signal across multiple surround speakers, and modern digital surround tech like Dolby Atmos, DTS-X, and Auro use decorrelation to up-mix to numerous surround speakers without localization to any one speaker.   This seemed the most promising.  The basics of the old analog matrix techniques are published at least.  I couldn't find much on their proprietary techniques but generally it involves symmetric left/right channel phase rotations using all-pass filters to altering the phase of one of the duplicate channels by 90 degrees and the other by -90 degrees (270 degrees?) making the two copies 180 degrees out of phase and providing full decorrelation.  Sometimes by some other amount but remaining symmetric (+/- 50 degrees say, retaining partial correlation).

How about just flipping polarity on one copy.  Old Halfler-surround difference-signal stuff.  Inverts the phase which reducing the correlation coefficient to zero.  Mixed back in with L/R the low bass would be attenuated on the inverted side and emphasized on the non-inverted side.  I've long thought the ideal way of doing this will involve duplicating the single channel and finding a way to alter the phase of one copy by 90 degrees and the other by -90 degrees (or +/- 50 degrees or whatever works best).  I have a phase manipulation Voxengo plugin to do experiment with this, but haven't gotten around to messing around with it enough.

Then I went the easy route..

[Gutbucket]

I just ran the front and rear facing center directional channels thorough a Mid/Side decoder.  Forward facing channel as Mid, rear facing as Side.  Dialed in whatever width works best by ear.  Works great! 

This (or a narrow x/y center mic configuration) is what I suggest to most folks do who want to try this stuff-
Most 4 channel recorders have built-in Mid/Side decoders.  Record the front and rear facing mics to an adjacent channel pair on the recorder such that the front facing center mic will serve as the Mid channel and the rear facing one the Side channel.  Listen right off the recorder, balance playback levels of the wide-spaced pair alone first with the center channels muted.  Then unmute the center channels and switch in the Mid/Side decoder with it adjusted to 100% Mid / 0% Side and bring up the Mid/Side level until it balances appropriately with the wide mics.  What you have is the basic configuration of a wide spaced pair plus a center facing directional mic. The basic 3-microphone configuration providing the aforementioned advantages with lots of after the fact adjustability.

Next mute the wide-spaced pair and play with the Mid/Side width control.  As your bring up the rear-facing Side channel you 'pseudo-stereoize' the monophonic forward facing center with rear-facing audience reaction and hall ambience.  Find a nice sounding ratio setting.  Notice that this is not stereo in a Left/Right directional imaging sense.  There is no pinpoint imaging, there isn't even fuzzy left/right imaging, but an ability to make the sound either more direct and center-focused or more broadly diffuse and ambient.  Cool.

Okay, now note your preferred Mid/Side ratio setting, revert back to 100% Mid, and unmute the wide pair.  Bring up the center again to balance well with the wide mics like we did originally.  Notice that you now have Left/Right stereo again.  Mentally compare that to the mid/side pseudo-stereo.  Now the kicker- slowly adjust the Mid/Side ratio away from 100% Mid.  What you have essentially is a blend control between center and sides which does not reduce image width as it is increased like a further increase in Mid level will do.  It affects the smoothness of the blend between across the front between center and sides.  It simultaneously acts as a sort of depth-control going from up front and flatter (all Mid) to very deep and overly ambient (all Side).  Find the best compromise setting.

Note that the preferred setting is likely to be less wide than when listening to the Mid/Side pair alone.  You only need a little Side as you are getting sufficient stereo interest (and difference information) from the wide mics, but that little bit goes a long way.

I love this.  I've been going back through my four-channel LRCB recordings (Left/Right/Center/Back) and listening this way.  Its relatively simple, easy to do and fantastic sounding.

[voltronic]

(You had two more posts as I was writing this, so pardon me if some of these points have been addressed.)

3) Nothing wrong with narrow-spaced omnis as a two channel technique in the right place.  To my mind that represents a sort of compromise, as do near-spaced configurations like ORTF, NOS, DIN, or what have you.  What do I mean by that, lumping omnis with near-spaced configurations?  Each of those is a somewhat different solution for balancing a whole lot of different things.  They are either coincident, nor wide-spaced and get some of both aspects going.  But that also means they don't achieve either in a fully exemplary way.  They can't provide fully decorrelated diffuse pickup and the direct pickup isn't fully correlated either, but represent an optimized compromise using only two channels.  Their strong point is simplicity and purity.  Straight two channel.  Two mics through to two speakers.  An argument for similarity to HTRF head-spacing can also be made for near-spaced configurations.  What I'm trying to do is move beyond the imposition of those compromises by combining techniques and using more channels.  More complex, yes.  Worth it to my ear.

You're absolutely correct that the single pair of narrow-spaced omnis is a compromise, but I'm just lucky that it seems to work for what I record.

I applaud you for pushing the envelope and innovating.  You should have some arrays named after you like a few of those people on GS.

But for most TS music tapers the problem is not being able to get a pair of omnis spaced far enough.  Its just a practicality thing.   I've seen plenty of near-spaced omnis in the section used on standard near-spaced mic-bars.  Too far away for that to work well IMHO.  On-stage, maybe.   Back in the audience even big mic bars can only go so wide with typical mics flown from a single on one stand.  That's wide enough for a stereo pair of omnis, but not really optimally wide with a center mic or pair.  Miniature omnis make sufficiently wide spacing a bit easier as you know.

I totally acknowledge that what I'm recording is in the extreme minority on this board, and narrow spaced omnis in the section would be not much better than just a single omni mic.

Early in my teaching career when I new next to nothing about recording, I used a pair of omnis maybe 20-30 ft from the stage, but on tall stands spaced 20-25 ft apart.  I recently pulled up some of those old recordings of student jazz bands and they sounded surprisingly good.  That's definitely not practical for typical TS recording!

I feel what I'm describing here is different than the general wide-space classical thing with a center pair and wide flanking mics to cover an orchestra.  This is more of a single array using multiple channels.  Than an addition of wide flankers and typically used at greater distances.

For sure.  I'm just drawing comparisons with the classical / acoustic side of recording.  What you're describing reminds me a bit of the Faulkner 4-mic array in that one of the goals is to compensate for farther distance with multi-channel forward pickup.


Since very wide set omnis are logistically impractical in taper situations, I can now see how your spidery multi-DPA setups with APE spheres make sense within the current discussion.  Correct me if I'm wrong, but it seems you're getting greater decorrolation to work with by using the rearward pointing mics, combined with the fact that they are all spaced some distance apart, with the benefit of adjusting the amount of it in your mix.  Drawing another parallel with classical recording, I once saw a video of a professional engineer setting up to record a pipe organ album in a large Gothic cathedral.  He used an array of 4 DPA 4006s all on the same bar with the diffuse grids.  One pair pointing straight ahead about 50 cm wide, the other pair a bit wider but facing to the rear and angled 45deg outward off center.  That setup obviously had other goals, such as fully capturing the acoustic.

The ultimate, I suppose, would be getting everything you're describing into a physically compact setup.  Perhaps the multichannel systems developed for surround ambiance could be adapted to music recording.  The 8-channel Schoeps ORTF-3D setup certainly looks like a good candidate, but I imagine the typical taper doesn't roll up with an $18k+ mic setup!

[Gutbucket]

Early in my teaching career when I new next to nothing about recording, I used a pair of omnis maybe 20-30 ft from the stage, but on tall stands spaced 20-25 ft apart.  I recently pulled up some of those old recordings of student jazz bands and they sounded surprisingly good.  That's definitely not practical for typical TS recording!

Not practical but its been done plenty of times.  Numeous examples of those kids of wide splits for Grateful Dead taping, and a taping aquaintence used to do the same at the Spirit of the Suwannee Music Park ampitheater in Live Oak, FL (where I've developed a lot of these techniques) with stands at either corner of the SBD area.  IMO it works without an egregeous hole in the middle because the PA content is predominantly mono (yes even the 'good stereoised' Dan Healy Dead mixes) and as long as the time of arrival distance is about the same from the PA stack on each side to the mic on each side, that mostly monophonic content will translate as correlated material filling the center of the playback image.  I speculate it may enhance any stereo PA effects content such as stereo verbs, choruses, double-mic'd leslie cabinets etc, and it should completely decorrelate all the venue ambience and audience reaction down to the very lowest frequencies.

What you're describing reminds me a bit of the Faulkner 4-mic array in that one of the goals is to compensate for farther distance with multi-channel forward pickup.

Since very wide set omnis are logistically impractical in taper situations, I can now see how your spidery multi-DPA setups with APE spheres make sense within the current discussion.  Correct me if I'm wrong, but it seems you're getting greater decorrolation to work with by using the rearward pointing mics, combined with the fact that they are all spaced some distance apart, with the benefit of adjusting the amount of it in your mix.  Drawing another parallel with classical recording, I once saw a video of a professional engineer setting up to record a pipe organ album in a large Gothic cathedral.  He used an array of 4 DPA 4006s all on the same bar with the diffuse grids.  One pair pointing straight ahead about 50 cm wide, the other pair a bit wider but facing to the rear and angled 45deg outward off center.  That setup obviously had other goals, such as fully capturing the acoustic.

Yes, there are definite parallels with the phased-array stuff.  And with David Greisinger's 5 mics in a line across the front of the stage technique.  At some point I'll post thoughts about all that.  There is phase-array antenna theory and is Hass-delay presence and clarity enhancement stuff going on, especially with the bigger 6 channel arrays I've posted about, but I think its also a factor with just 3 points/mics in a line.  Besides being far more practical, I've found going wide with additional mics is better than going deep, that is to say adding mics out to the sides rather than a pair further behind the main pair, perhaps because it preserves arrival transients from frontal direct sound instead of smearing them out.  One can always slightly delay the ambience/surround/rear-facing mics if necessary, which also servers to further decorrelate them from the main channels.

The ultimate, I suppose, would be getting everything you're describing into a physically compact setup.  Perhaps the multichannel systems developed for surround ambiance could be adapted to music recording.  The 8-channel Schoeps ORTF-3D setup certainly looks like a good candidate, but I imagine the typical taper doesn't roll up with an $18k+ mic setup!

I've pulled a lot of ideas from various surround ambience recording arrays.  One way to think of the 4 directional mics at the center of of my 6-channel setup is as a bastardized IRT-cross turned 45 degrees so it faces front/left/back/right instead of left-front/left-back/right-back/right-front.  I've gotten this thing very physically compact using low voltage DPA miniature omnis and supercards powered by PFAs mounted on the folding the telescopic antennas. Since its all wired up and ready to go I can typically set up and break down considerably faster than most 2-channel tapers who break everything.

[Gutbucket]

Getting back to a few final thoughts on the 4-channel setup with center front and rear facing mics run through the Mid/Side decoder-

So what is this bastard Mid/Side decode thing actually doing?  It's decorrelating the rear facing channel by mixing a phase-inverted copy into the left and right channels.  The Mid/Side ratio control simply makes for an easy way to balance it against the forward facing mic.

So I have the primary wide microphone pair providing open decorrelated ambiance along with some correlated information- enough correlated information to provide left/right directional imaging, but preferably set a bit wider than I'd want from that pair if used alone (more of a dent than a hole in the middle).  I have the forward facing center directional microphone channel which maximally focuses on the direct sound arriving though the front 'window' of primary interest and excludes the audience and hall ambience as much as possible providing only correlated information (its signal is routed to left and right identically).  And I have the rear facing directional mic channel information fully decorrelated, clearing it out of the center of the playback image and acting as a blend and depth control between the wide ambient omnis and the up-front present center mic.

[voltronic]

Getting back to a few final thoughts on the 4-channel setup with center front and rear facing mics run through the Mid/Side decoder-

So what is this bastard Mid/Side decode thing actually doing?  It's decorrelating the rear facing channel by mixing a phase-inverted copy into the left and right channels.  The Mid/Side ratio control simply makes for an easy way to balance it against the forward facing mic.

So I have the primary wide microphone pair providing open decorrelated ambiance along with some correlated information- enough correlated information to provide left/right directional imaging, but preferably set a bit wider than I'd want from that pair if used alone (more of a dent than a hole in the middle).  I have the forward facing center directional microphone channel which maximally focuses on the direct sound arriving though the front 'window' of primary interest and excludes the audience and hall ambience as much as possible providing only correlated information (its signal is routed to left and right identically).  And I have the rear facing directional mic channel information fully decorrelated, clearing it out of the center of the playback image and acting as a blend and depth control between the wide ambient omnis and the up-front present center mic.

Could you accomplish something similar by replacing the front/rear center mics with a single forward-facing fig8?  It would be like a double-MS but rotated 90 degrees.

[thatjackelliott]

Instead of making the center mic an X/Y pair or M/S pair, use a center rear-facing directional microphone in addition to the front facing one. 

Instead of using two mics in the center -- one directional forward-facing mic and one directional rear-facing mic -- would a figure-8 work?

(EDIT: Looks like voltronic beat me to the punch on this question.)