I have the usual issue of not understanding what the fuss is about, and why DNG always creeps into the discussion.
This is a case of spatially offset multiple level sensing, the Fujis did it too. No theoretical innovation here. For me the problem is there is no "canonical" image. But maybe that was the case with Bayer also? To me, reconstituted imagery is hamburger rather than steak. Adobe serve a mediocre meal which please most comers, while Illiah is a master of cooking the hamburger for gourmets, but -no disrespect- that still doesn't unchop the meat. And you still have different sampling frequencies for RB and G, with all the afferent effects.
Edmund