I would be interested in comparisons at equal f-stop between cameras with sensors of the same size but different resolutions, like 16MP, 24MP vs 36MP for Nikon, and 18MP, 22MP and 45MP or whatever for Canon. Say comparing those several resolution options all at f/5.6, and all at f/8, and all at f/11. That would show which DOF choices can benefit from greater resolution, and by how much, and also at what stage there is little resolution gain (even though there could be other gains, like reduced aliasing and maybe less jaggies). It seems common opinion that there will be little resolution benefit in going beyond 22/24MP when you need f/16 to get your desired DOF, but f/11 and f/8 are controversial.
I've already done such comparisons, BJL. A few years ago I made the mistake of jumping in too soon and getting a Canon 40D, not realising of course that Canon would release a significantly upgraded 50D a short time later, which I also couldn't resist buying. The 40D at 10mp is equivalent to a 26mp full-frame, and the 50D at 15mp is equivalent to a 39mp full-frame.
The target I used was an Australian $50 bill taped to a wall, and the lens used was the Canon 50/F1.4. With such tests at relatively close distances, focussing is absolutely critical. That was the main challenge. Both cameras have Live View, but the 40D's LCD screen is lower resolution than the 50D's screen. This fact alone created some doubt as to whether or not focussing was identical with both cameras, which were placed alternately on the same tripod in the same position after attaching the same lens.
To overcome such focussing problems, I simply relied upon moire and aliasing artifacts as my guide for best focus. The $50 bill has similar characteristics to any test target with fine lines. When focussing is 'spot on', there's always a blaze of colorful artifacts that appear in certain parts of the image, when the camera is within a particular range of close distances from the target.
From memory, the results of my tests were:
(1) At F5.6 the 40D was very marginally sharper than the 50D at F8, at extreme pixel-peeping levels of 200% or more on screen.
(2) At F11, the 50D was fully equal in resolution to the 40D at F8. In other words, the 50% increase in pixel-count of the 50D had fully compensated for the increased diffraction at F11.
(3) At F16, the 50D had about the same resolution as the 40D at F11. I say 'about' because the results were not quite as identical as the 50D at F11 compared with the 40D at F8. In other words, at 400% magnification on screen one got the impression that maybe the 40D shot was very, very slightly more detailed.
(4) At F22, the 50D image was noticeably softer than the 40D at F11. At F22 I believe diffraction has fully kicked in and no increase in pixel count could compensate for that, without also an increase in sensor size.
(5) Comparing both cameras at equal F stops, from F5.6 to F16, the 50D image was more detailed to varying degrees. The same should apply to comparisons between 26mp and 39mp full-frames, except at the edges and corners of course.