Cool. Did you determine the crossover frequency beforehand or decide upon it by shifting it up/down and listening for best effect? If determined by listening, how broad was the region in which there was little significant difference?
There are a few things in play here. On the listening side- the Duplex theory of localization, on the recording side the AB omni spacing and its relation with wavelength, stereo recording angle, and the frequency above which pickup of diffuse sound becomes decorrelated.
Would be interesting to compare this to the same setup with a 2-3 times wider AB spacing if you get the chance. The "texbook" Stereo Recording Angle of significantly wider AB omni spacings would no longer be as close to that of the Blumlein pair as they are in this case (where both are within 10° as indicated by the sengpielaudio visualizer; 90° Blumlein ~= 75° SRA; 30" AB omni ~= 85° SRA) but the SRA of omnis is pretty loose to begin with, especially at wide angles, and would only become more so only across a partial frequency range, presumably.
Increased AB spacing will lower the diffuse pickup decorrelation frequency between the two omnis, but also between the Blumlein pair and each omni (half the distance - twice the frequency) which reduces the phase interaction between pairs that is likely to be a leading contributor to a "cloudier & thicker" sound.
I suspect this kind of carefully tuned crossover technique may be most advantageous for blending pairs that have similar SRAs and are less widely separated from each other. It represents sort of a "dual-element main pair incorporating a careful hand-off that avoids conflict", rather than a mix of two pairs with different strengths that accentuate each other with somewhat less inherent conflict.
I tried and dismissed the "similar SRA's" approach in building up OMT arrays, but never got around to implementing a crossover to manage the hand-off, and that is what I suspect is required for such an approach to work correctly. Thanks for posting about it here.
