The theory behind it-
(you may already know this!)
Fundamentally, all first order directional polar patterns are created by the combination of pressure (omnidirectional) and velocity (bi-directional or fig-8) components.
A cardoid pattern results from a 50/50 combination of pressure and velocity (which is to say, 50% omni + 50% fig-8). A subcard results from more omni, less fig-8, a super/hyper from less omni, more fig-8. In most of the microphones we use, this combination is determined by the construction of the microphone, and is fixed. But in microphones that feature mechanically or electrically switchable patterns, this is what is actually being changed to affect a change of pickup pattern.
If you mount an omni and a figure-8 so as to be as coincident as possible and record their outputs as two to separate channels, when you mix those channels together later, you get a polar pattern that can be varied depending on the level difference between the two. Cool, huh?
Practicalities-
- If possible, use two microphones of the same model with different polar pattern capsules (as they are more likely to have close phase and frequency responses)
- Make sure the polarity of all channels is the same. If polarity of one channel of the packet is reversed, whatever virtual pattern is produced will face backwards (and in most cases will not be the same forward-facing pattern you would've gotten with the two in the same polarity)
- To do it right, tape or otherwise mount a microphone of each pair together so that both are parallel and the diaphragms as close together as possible, one directly above the other in the vertical plane. Do not allow either microphone of the "packet" to shift at all in relation to the other (you also don't want the two "packets" to shift relative to each other, in the same way that you don't want two microphones of a stereo pair to shift relative to each other, but it is more critical in this case)
- If it gets weird or sounds odd, you can always revert to using just one pair or the other..
- ..or you can try implementing a cross-over between the two. This will keep the two from interacting outside of the cross-over range. So if using say an omni and supercard together, you can have omni response below the cross-over point, supercardioid response above that point, and a morph through cardioid in the crossover region
- A simple version of that is low-pass filtering the omni wherever the supercardioid begins to naturally roll-off, and leaving the supercardioid unfiltered. You then get extended omni bass at the very bottom, cardioid through the rolloff region and supercardioid response through most of the supercard's frequency range.
- Shaping frequency response of the two channels differently using complementary EQ (one cutting where the other is boosting) is a more subtle way of morphing polar pattern than a hard cross-over, but requires good interaction between the two.
- You'll get the greatest range of adjustability by combining patterns that differ most. If you don't have a fig-8, that means omni + hypercardioid (hypercardioid is mostly bi-directional with a little omni thrown in)..
- ..but unless you've already determined that it's working well for you, it will be safer to use the two patterns you think will be most appropriate to the situation on their own. This will reduce the total range of adjustability, but either on its own is more likely to make a good recording.