Topic: Oddball microphone technique (OMT) - part 4

[checht]

Wonder if there are isues when a pair of mics is facing 180º away from the others? Do you flip the phase in post?

TIA

[Gutbucket]

^ The short answer is typically no.  But whatever works in the mix. What sounds right is right.

Consider a single pair of L/R stereo pair of cardioids.  Use whatever stereo angle you like between the two.  Pointing them 180º apart would be extreme but you wouldn't invert the polarity of one of them because the angle increased to 180º.  In that case the mics are being angled along the Left/Right axis, but same goes for angling them along any other axis.  If the goal is pickup of content arriving from that direction, you probably don't want to invert polarity.


Long answer exploring why in more depth, feel free to ignore if confusing..

Fundamentally -

An omni is a "pressure sensor" regardless of which direction it's pointed.  Positive pressure at the diaphragm produced by a wavefront arriving from any direction produces a positive signal voltage output. Inverting polarity of an omni will tend to cause cancellation of low frequency content when combined with the other mics of the array which are not polarity inverted.

In contrast, inverting polarity of a figure-8 flips its orientation to the opposite direction.  The fig-8 pattern remains the same, but now a positive output voltage will be produced for wavefront arriving from the opposite direction.  Inverting polarity of a rearward-facing fig-8 produces a forward-facing fig-8 pattern.

A cardioid is an equal combination of omni and fig-8 components from the same point in space mixed together.  Inverting polarity of the microphone output inverts both it's omni and fig-8 components at the same time.  Prior to the polarity inversion those components are combined in the same way as before - the positive output lobe of the fig-8 component mixes additively with the omni component, doubling output on-axis, while the negative output lobe of the fig-8 component mixes subtractively with the omni component forming the rear-facing null.  Since the polarity of both components is inverted, the polar pattern faces the same direction as before, but with inverted polarity.  When mixed with the other microphone channels the inverted polarity output of this microphone will combine subtractively wherever the signal is sufficiently phase-coherent across the summed channels, which with a near-spaced pairs will tend to occur at low frequencies.  Were the signal is not phase coherent across the summed channels it will mix additively. 

So.. inverting polarity of a near-spaced stereo pair pointed in the opposite direction will probably cause the mix to sound leaner in the bass where the content is mostly phase coherent in all mic channels, but probably won't sound much different at high frequencies where the content and phase-relationship between in the various channels is different.

Taking it further-

If instead we were able to invert polarity of only the fig-8 component and not the omni component before the two are combined inside the microphone, the resulting cardioid polar-pattern would then face in the opposite direction, yet the polarity of the microphone's output would remain the same.  That's essentially what is happening when converting Mid/Side to Left/Right.  The fig-8 Side channel's contribution to the Right channel is made with inverted polarity, so the Right virtual polar pattern faces the opposite direction of the Left virtual polar pattern.

If instead we were able to invert polarity of only the omni component and not the fig-8 component, the polar-pattern would also face the opposite direction, AND the combined output would have inverted polarity.

That's essentially what's going on with Ambisonics, where multiple microphone capsules are arranged so as to be as coincident as possible and pointed in all directions.  With all the mic capsules positioned in nearly the same point in space, their outputs remain phase-coherent to a sufficiently high frequency that the virtual patterns which result from combining their outputs in specific ways via the manipulation of polarity and level maintains the polar pattern shape up through a sufficiently high frequency.  The closer the arrangement of capsules in an ambisonic mic, the higher in frequency it is able to maintain a coherent pattern.

[F0CKER]

Gut, I finally got a chance to run the OMT4 setup with the 4018s on the outside about 2' spread, 30* angle with the 2015s in the center as an XY pair at 90*. I had the 4018s about 8 inches in front of the subs in this case. The results are excellent. Big caveat this is a great sounding room so most tapes sound great here. I was keen to try to this out for the times where I will have a less than ideal room. The hypers on the outside sounded quite nice actually but lack any depth and sound thin to my ears. They are, as expected, focused, detailed with good punch and clarity, lacking low end response, etc. The subs XY in the middle by themselves sound quite horrible tbh. The main thing I wanted to test here is if the subs XY in the center added anything of value. When the 4 are mixed down they add quite a nice low end and provide some depth. The sum is definitely greater than the parts. I will post the raw and processed files later today. I ended up with a heavy hyper mix, about 75% with just enough of the subs to give some low end presence and space. Quite nice. Look forward to trying this in a less than ideal room to see how it holds up.

Also because I only have one set of DPA-A Bodies, I have to choose between hypers or card caps with the 2015 being a separate stand alone pair. So it needs to be some combo of subs and then either  hypers or cards which is why I went hypers outside, subs inside.


Edit: Updated with links to the raw and matrixed files
https://drive.google.com/drive/folders/1dIvKOUiE_3DJlB0iU5x5VKlrHiRGyjsm?usp=drive_link

[Gutbucket]

Big thanks for the feedback and happy to hear it worked to satisfaction! Your description of the sum being greater than the parts parallels my experience.  I think to really work optimally in combination its necessary for each of the pairs to sacrifice working ideally by themselves. 

Good to hear the subcardioids worked well in the center X/Y pair position, since despite my speculation that it should provide the right things to work in the mix that's something I've not actually tried myself.  I wouldn't expect the X/Y pair to sound great by itself - the subcards aren't directional enough to provide the necessary channel separation to work really well as an individual coincident stereo pair, but in the mix it's no longer necessary that they provide sufficient channel separation that would be needed on their own.  Their contribution is instead anchoring the center and enhancement of the limited sense of dimension, depth and bass-extension from the more PA-directed supercards. 

What X/Y angle did you use?  Any thoughts about X/Y in the center in comparison to the single forward facing supercard you used previously there? ..recognizing of course that the different patterns used on the recordings will also sound quite different.  One interesting thing to do if you've not already messed around with it is to play with panning the X/Y pair in various ways.  I assume you have the pair hard panned L/R in the mix, which makes sense, especially using subcardioids.  However, if you pan both to center you emulate use of a single mic in the center with a polar pattern somewhat wider than either mic of the X/Y pair.  I find it useful to dial in that panning after getting the levels of both pairs where I want them. Doing so adjusts the stereo blend across the image and the overall sense of width. Also particularly useful is the ability to set the level and panning positions of the individual X/Y channels differently as needed to help correct an off-center recording.

If you decide to use this arrangement regularly I think you'll get good milage out of it.  To best adapt to different situations, I suspect it might be helpful to swap the position of the 4015's and 4018's depending on the room.  Using the 4018 supercards in the outside position should help in more challenging rooms, while using the 4015's on the outside may potentially sound even better in really good rooms.  That's partly because, just like you mention doing, I also tend to treat the wide pair as primary and bring up the center pair at a lower level in support of it.  So it makes sense to use the supercards in the outside position in situations where you might have used a single pair of supercards, and to use the subcards in that position where you would expect a single stereo pair of subcards to work well.  But that said, if you find using the supercards on the outside achieves the sound you want in great sounding rooms as well there may be no need to switch back and forth.

I hope to be able to give a listen to tonight..

[rocksuitcase]

Boy do I have a OMT mixing story for you guys.
One final mix result is here:
https://archive.org/details/eggy2025-03-13.dpa4060akgc426ck61

I "mixed" that one by using the Mid cardioid microphone of a AKG c426 MS pair, using that as center, then added the DPA 4060 omnis which were spread about 5 feet apart, 2.5 feet from center. I then added a bit of the AKG ck61 cardioids spread about 4 feet apart, 2 feet from center in an AB pattern (facing straight at the stage). So I called it an OMT5 mix. I matched each stereo pair to be -6dB from 0, then made the Mid channel at -4dB from 0 for the final "mix". I like the sound of the vocals and ambiance but...
After posting that I thought there was not enough low end in the result, so I went to mix one where I did the Side channel manual decoding which actually made the final mix pretty different sounding.  it is almost OMT7 with the side pair made into Left/Right.
I will upload that one to LMA tonight or tomorrow and let you guys know via editing this post.

OK, here is the mix which adds the Side content and is a bit different in Omni vs inside AB cards content also. (More Omni, less cards vs the first mix)
https://archive.org/details/eggy2025-03-13.dpa4060MSc426ck61

If anyone hears a difference let me know WHAT you hear. I have my opinions I will share after a few others share their opinions.

Anyone get a chance to listen to these? With all our talk of mixing the separate channels, I thought some folks might have time and/or an opinion. If anyone does, let me know either in this thread or PM.
I'm also working on my database in Excel format to indicate how many OMT recordings I have on LMA.

[Gutbucket]

I've not yet had a chance yet, but am hoping to listen tonight along with FOCKER's recording.

[F0CKER]

What X/Y angle did you use?  Any thoughts about X/Y in the center in comparison to the single forward facing supercard you used previously there? ..recognizing of course that the different patterns used on the recordings will also sound quite different.  One interesting thing to do if you've not already messed around with it is to play with panning the X/Y pair in various ways.  I assume you have the pair hard panned L/R in the mix, which makes sense, especially using subcardioids.  However, if you pan both to center you emulate use of a single mic in the center with a polar pattern somewhat wider than either mic of the X/Y pair.  I find it useful to dial in that panning after getting the levels of both pairs where I want them. Doing so adjusts the stereo blend across the image and the overall sense of width. Also particularly useful is the ability to set the level and panning positions of the individual X/Y channels differently as needed to help correct an off-center recording

I ran them XY at 90 in the center. I have dedicated mounts at that angle so playing around with angles will require some new gear. I hadnt thought about doing anything with panning. Will play around with that, but yes it was hard left and right in the matrix above. The other variation of this subcard setup for the center I was thinking about what something akin to the Healy Method, basically back to back pointing diredtly out to the sides. I will give this a shot just to see how it pans out. Playing around with these different setups has brought some additional enjoyment to this sport.

[grawk]

The nice thing about xy is you can adjust width in post just like midside, so you don’t really need to change the actual orientation.

[F0CKER]

The nice thing about xy is you can adjust width in post just like midside, so you don’t really need to change the actual orientation.

Is that just done through the panning? How would you go wider than 90 degrees if you’ve panned full left and right with the mics set at 90? Or is it only to make it more narrow?

[VibrationOfLife]

You can really only go inward with panning/blending, but with a mid side you have all the things.

[Gutbucket]

Its very cool that the physical angle used between a pair of X/Y mics and the virtual angle which results between the two mic patterns in the mix can be different, as it's something we can use to advantage when recording and afterward in the mix, even playing the two off of each other to some extent.

The nice thing about xy is you can adjust width in post just like midside, so you don’t really need to change the actual orientation.

Right.  We can re-adjust the balance afterward somewhat, which can help in fixing it if not right, further optimizing it if already pretty good, or even allowing us to intentionally use an X/Y angle that we know wouldn't work well on its own without a re-adjustment because it can be made even better after the re-adjustment.  More on that below..

Is that just done through the panning? How would you go wider than 90 degrees if you’ve panned full left and right with the mics set at 90? Or is it only to make it more narrow?

You can go both more narrow and more wide than the physical X/Y angle. Panning a stereo pair wider than full Left / full Right equates to mixing in some of the opposite channel with inverse polarity to either side.  But going either way to the full extreme, the result will be a mono signal. More below..

You can really only go inward with panning/blending, but with a mid side you have all the things.

We can do it in a few different ways, including using and enhanced version of stereo panning/blending.. but we don't quite get all things.  There are a few important limitations.  To get all things requires taking it one step further from Mid/Side into ambisonics.  More below..

[grawk]

90deg XY is able to be mathematically converted into Mid and Side, allowing the same techniques on both.

[Gutbucket]

^Sure, 90deg is the safe choice and a good place to start for OMT.  But any X/Y angle can be Mid/Side manipulated and there may be some advantages in using a different angle. 

First how for the others:

Easiest way to make the angle/width re-adjustment is using an enhanced stereo-balance control on the stereo channel of the mixer for the X/Y pair, if that's available.  I use Samplitude and it provides an advanced option on the stereo balance pot control of stereo mixer channels which allows one to alter the stereo width of the pair rather than the L/R channel balance. In that mode, the pan/balance pot becomes a stereo-width adjustment pot. Rather than shifting balance left/right, it simultaneously pans both channels closer to center as the pot is rotated counter-clockwise, or pans them "over-wide" beyond the original full Left / full Right pan pot positions as the pot is rotated clockwise.  To do that, as the pot is adjusted to a "wider than normal stereo position" it begins mixing in some of the opposite channel with inverse polarity.  This is sometimes called "super-stereo" or "over-wide stereo".  Pushing it all the way we end up with an equal amount of each channel mixed into the other, but with opposite polarity.  The result is the identical mono "difference signal" in both channels except with opposite polarity - exactly the same the Side channel from a Mid/Side pair and how that gets used by routing it to both Left and Right channels equally but with opposite polarity.

Aside- If familiar with the "squiggle cloud" of a goniometer display, recall that a mono signal produces a vertical line on the display.  Panning that mono signal Left or right angles the line to the left or right on the display as level sent to both output channels changes.  Fully panned hard right or left with nothing in the opposite output channel places the line at a 45 degree angle. Going "over-wide" angles the line past 45 degrees by starting to mix in some of the opposite channel with inverse polarity.  Going all the way to that "over-wide" extreme we end up with the same content in both channels except fully out of polarity with each other, producing a horizontal line.  A stereo signal instead of a mono one behaves similarly, but instead of a straight line it produces a squiggly cloud that dances around.  The fatter the squiggle cloud, the more stereo stuff that is occurring.  The cloud might tend to spread out more along the vertical axis which is indicative of a narrow panned and mostly mono stereo signal, might tilt toward the left or right indicating that the balance is tilted that way, or might be wider than it is tall which indicates that the content is more out of polarity than in polarity across both channels because it has more Difference (Side) content than Sum (Mid) content.

Another way to do it is using a stereo width adjustment plugin tool.  I commonly use Voxgengo's free MSER plugin which is a Mid/Side tool that has the ability to do some other stuff as well, but can be set to simply convert L/R to Mid/Side, let you to make an adjustment to the level level of Side (Sum) verses Mid (difference), then convert back to L/R again.

[Gutbucket]

The what/why:

We can convert back and forth between X/Y (Left/Right) and Mid/Side (Sum/Difference) without any loss as many times as we like if we make no change in the level of one channel verses the other.  Or we can switch from L/R to Mid/Side, change the level of Mid verses the level of Side (changing the ratio between the two) and switch back to L/R again, incurring only a slight loss which will probably be negligible depending on the extent of the change.  We then end up with an altered virtual X/Y angle AND an altered virtual polar pattern than whatever we started out with.  If we want to take that output and re-adjust back again without reverting to the source material, we can't go too crazy with it without loosing information and the ability to get back to where we started - if we were to adjust the ratio all the way to 100%Mid or 100%Side, we'd end up throwing out all the information from the other channel and would be left with only the mono sum (Mid) in both channels or the mono difference signal which will be out-of-polarity in both channels. 

90 degree X/Y is a good safe starting point because its right in the middle Goldilocks zone between all (180 deg) and nothing (0 deg).

Although we can usefully re-adjust the virtual angle afterward to a useful extent, the physical angle used between an X/Y mic pair when making the recording still matters for a few reasons:
1) If not tweaking/re-adjusting stereo width of the X/Y pair afterward, the physical angle you use when recording is what you end up with in the mix.  90-degree X/Y works pretty well in OMT, although a wider angle often works better for an X/Y pair used on its own in the same situation.
2) Changing the physical X/Y angle alters the virtual Mid pattern.  100% Mid equates to both channels summed together.  If we used no X/Y angle at all (both mics parallel to each other) the virtual Mid would have the same polar pattern as either of the two microphones.  The wider the physical X/Y angle, the wider the virtual Mid pattern becomes when the two are summed.  If we were to use a 180 degree X/Y angle, the virtual mid polar-pattern that results would be omnidirectional regardless of the physical polar pattern of the microphones (Exception is fig-8's, which would cancel out entirely producing no virtual Mid).   This is a big one if the intent is isolating pickup of the center pair to direct sound arriving from the stage and PA as much as possible.  A pair of supercardioids at a 90 deg X/Y angle do not produce a supercardioid Mid, but something more like a cardioid pattern.
3) Response of most microphones is best directly on-axis.  It makes sense to orient them on-axis with the source of interest when possible, particularly when we are attempting to attenuate off-axis pickup by using a highly directional polar pattern.

We might leverage 2 and 3 a bit by pointing the X/Y pair directly at the PA.  The resulting X/Y angle will probably then be less than 90 degrees.  That's good in that it keeps the mics on-axis with the source, and because it makes the virtual Mid more forwardly directional. We can sum the X/Y pair to mono and get a tighter virtual single center mic pattern than we otherwise would have achieved.  But maybe its not so good if the stereo width that results from that narrow PAS angle is insufficient.   In that case we can re-adjust the virtual angle afterward to regain the stereo width we want from that pair and still have the physical mics pointed advantageously ..within limits.

[Gutbucket]

[snip..] with a mid side you have all the things.

You gain some useful degree of adjustment. But the useful range is limited by a few things. One is that virtual X/Y angle and virtual X/Y polar patterns are inextricably linked.  If we change one we change the other along with it.  The wider the virtual X/Y angle, the narrower the virtual polar patterns.  At one extreme (100% Mid) the virtual polar pattern is the combined pattern of both mics summed together with their physical patters overlapped by whatever X/Y angle was used, and for any practical X/Y angle greater than 0 degrees, that pattern will always be wider than the physical polar pattern of the mics being used, as previously mentioned.  At the other extreme (100% side) the resulting virtual polar pattern will be a side-ways facing fig-8.

Unfortunately that trend is the opposite of what tens to be desirable for an X/Y stereo pair.  As the X/Y angle is made narrower, we'd like the pattern to be narrower too, and vice-versa.  Assuming a Blumlien pair of fig-8's works optimally at 90degrees, we'll probably want a wider X/Y angle than that if using supercards in place of the fig-8's, and wider angle still if using cardioids.

Ambisonics goes further in that it allows us to alter virtual pattern and virtual X/Y angle independently of each other.  It gives you all things.. at least from one-point in space.