I was wondering the same. My thought is something like: the not normalized one doesn't acount for falling response. So if this was a fullrange speaker it would probably look similar as the normalized one. But since level is faling below 1khz. that means that 800Hz @ 45 degrees is lower in level than 45 degrees at 4kHz and there for it appears to narrow. But it narrows the same amount both of and on axis, there for directivity is still held constant, which is clear in the normalized one, since there the reference is the on axis level at the given frequency, not absolute level. Did this make any sense? Absolute level vs level relative to on-axis level at the give frequency is the keyword.
I made a rotating rig from an IKEA TACO serving plate xD. Just screwed it to a larger plate. Then I drew a circle around it and marked of the different angles. Probably not 100% correct, but I think the largest error would be geting the mic at excactly 0 degrees in the first place.
Axis if rotation was horn entry. I didn't have time to do vertical. Will need to modify the TACO rig some more
Listening impressions from this combo will be very short. I think subjective description only has value if you have a good reference which you can A-B test against with minimal time between. I happen to have such a setup playing now, but the reference is a Tang Band5" full range speaker.... The CD-Waveguide outperforms it easliy, leaving the fullrange sounding compressed, hollow, and coloured. The fullranger images better. But mostly because the SEOS is sitting on top of two chairs to the side of the fullranger, and the XO is not at all optimal. No time or phase coherence. It's also difficult to here what is the fullranger and what is the waveguide when I run them together. It's obvious the fullranger can't keep up. A 12/15" mid or dual 8" are probably a better match in terms of dynamics, directivity and distortion below 300hz.
A more interesting A-B scanrio would be with another well know horn/driver combo siting directly above/below it, and some cabable and similar hardware from 800hz and down. This is all possible to do with todays DSP's.
Another thing I would like to mention is how closely the SEOS horn measures outside vs inside. If I use gating or smoothing to remove some wrinkles from room interaction they look almost identical. The fullranger or normal dome tweeters do not.
Here is a measurement from the listening position with same EQ and XO as before 1/12 smoothing:
Truly a product of the narrow and uniform directivity I think.