Len, it is very nice recording. Please, what configuration was the stereo track created from? Right channel hypercardiod -65 and left channel a mix of two hypercardiods 0 and +65?
The left second-order ambisonic microphone was decoded to a first-order hypercardioid and rotated left 65 degrees. That same mic was also decoded to a first-order hypercardioid facing forward at 0 degrees.
The right second-order ambisonic microphone was decoded to a first-order hypercardioid and rotated right 65 degrees.
And the entire recordings was first rotated left five degrees to correct for the slightly right-off-center location of the mic stand.
Mixing coincident hypercardiods 0deg and 65deg will result in a less directional pattern, e.g. "supercardiod". So is this stereo recording hypercardiod and "supercardiod" spaced 14cm? Have you tried comparing it with a stereo pair of two hypercardiods spaced 14cm panned to the same acoustic center and with an equally wide stereo image? That might be interesting to hear.
Kuba- Your thinking that the left stereo channel consists of a virtual pattern shaped something like a cardioid or supercardioid is along the correct lines. However, depending on how it was mixed, the right stereo channel is likely something altogether different and far more complex.
Lets walk through it..
In this situation we have two physical, near-spaced ambisonic microphones (lets call them: mL, mR), from which three virtual microphone channels are being derived (vLl, vLc, vRr), which are being mixed down to a two-channel stereo output (Lt, Rt). I'll ignore the 5 degree rotational correction that was made.
The Left channel of the stereo output (Lt) consists of the sum of the two first-order virtual microphones (vLl + vLc) derived from the left ambisonic mic (Lm). Because both of those virtual channels are being derived from the same physical microphone, they are fully coincident with no phase difference between them, and the virtual polar pattern which results from their sum will be a non-complex standard first-order pattern as well. The shape of that combined pattern is going to be determined by the shape of the two individual patterns, the angle between them, their relative levels, and polarities. We can take it one step further and break each pattern down into its fundamental omni and bidirectional components.
In this case, the two individual virtual patterns (vLl, vLc) are identical (hypercardioid) and their phase and polarity are also identical. Only the angle between them is different.. and perhaps their relative levels.
If both virtual microphones were to share the same orientation angle, along with the same pattern, polarity, and level, the sum of the two would produce a combined virtual pattern with the same shape as the original virtual patterns, oriented in the same direction, yet with a 6dB increase in level. If the two virtual hypercardioid microphones were instead oriented 180-degrees apart, their resulting sum would be an omni, with a
decrease in level, the exact amount of which depends on how "hypercardioid" the pattern is. The omni components of the two virtual hypercardioids get summed (added together, since they have the same polarity) and the bidirectional vector components of each also get summed (subtracted and nulling out, as they have the same polarity yet are facing in opposing directions).
^
From that we know that given a 65 degree angle between virtual microphones rather than 0 or 180 degrees, the pattern that results from the sum of Ll + Lc will be a shape that lies partway between the hypercardioid limit associated with 0 degrees and the omni limit associated with 180 degrees, we can't tell exactly because we don't know the relative levels at which they were mixed. But going further, since a cardioid is produced from the sum of omni and bidirectional components in equal amounts, we can assume the resulting pattern will be a shape that lies somewhere between hypercardioid and cardioid.. something cardioid or supercardioid-ish depending on levels.
The Right channel of the stereo output (Rt) is likely not nearly as simple! I'm going to assume that during mixing the 0-degree forward facing virtual mic channel (vLc) was not routed hard left, but rather panned center in the stereo mix.. or somewhere close to center. In other words, it was routed more or less equally to Lt and Rt. Len, if I'm mistaken about this please correct me.
If that is the case, like before both vRr and vLc are virtual hypercardoids with a 65 degree angle between them, except this time the two are no longer coincident but are instead spaced apart by 14cm, and that will produce a complex phase relationship above some threshold frequency which I won't bother to calculate. Below that frequency the pattern shape of the sum of the two virtual microphone channels will behave similarly to the Left stereo channel, except angled 65 degrees right rather than 65 degrees left. Above that threshold frequency the pattern will be come complex, with peaks and nulls that form, shift and multiply with increasing frequency.. as does the pattern which results from the sum of any spaced pair.
But such is the nature of summing non-coincident mic channels. Not easy to analyze, but that doesn't mean it can't sound great if well managed. It's the way my arrays work whenever I mix them down to 2-channel stereo rather than routing the individual channels discretely to multiple speakers for playback in surround. Many roads to Rome, and for music at least, the ear is the only compass that really matters.