I'd have to agree. This is (oddly enough) why I decided to go with a center, in that my HT is not a studio, or to put it differently, there may be 8-12 folks in here trying to enjoy the show. After experimenting around with a lot of different configurations I decided the only simple way to get reasonable dialog_in_center for a variety of locations was a center speaker.
Obviously my requirements are different than yours since I'd assume it's no problem for you to sit near the sweet spot.
I've done some active crossovers and other simple op-amp circuits in the past. I'm thinking about building a pair of centers and locating one on each side of the screen, then doing some tiny bit of mixing left into left center and right into right center. It would be cool if studios would do this (8.1?) but if I get time it sounds like a fun project.
Actually rather than x.1 I'd like to see the 'next' generation of bitstream audio data contain waveforms and location info, then the processor could map the waveforms to transducers as it deems best. All we need to do is tell the processor where the speakers are and it could figure out the rest.
Then we could set our systems up to meet the requirements of our room/audience without fiddling with downmixing etc.
If you do the mixer, any chance you might release the design into the PD?