I'm a bit amazed at all the fuss this comparison from Michael has created. I appreciate both sides of the argument. I can see that Michael's test in itself would not be particularly useful to those who are undecided whether or not to buy an M9 or NEX-7.
From my perspective, the situation of having just one lens, on which Michael has based his comparison, would rarely apply because I've long since been converted to the convenience of zooms.
As I see it, if one is concerned, perhaps even obsessed with subtle differences in image resolution, then utilising the entire area of the sensor, in whatever camera one is using, is something one would always strive to do.
However, if one is travelling light with an M9 and a Summilux 50mm prime, perhaps attempting to get shots in the manner of Henri Cartier-Bresson, the additional weight of an NEX-7 body is trivial. I'm amazed how light that camera body really is. It's the lenses which are the weight problem.
In such circumstances, one would definitely want to know to what degree one might be disadvantaged, resolution-wise, when in any particular situation. If the lens when attached to the NEX-7 doesn't provide a suffient FOV for the composition, then that's reason enough to switch the lens to the M9.
If the lens is already attached to the M9 and one realises that a particular composition will need significant cropping, it would be good to know for purely paractical reasons just how significant an improvement in image resolution would be achieved by taking the trouble to change the lens over to the NEX-7, at the risk of perhaps losing the shot. Missing the shot completely has
to be worse than getting the shot with slightly compromised resolution.
If I were to criticise Michael's test, which I'm reluctant to do because he's already received such a blasting, it would be for not going far enough with the test. The 8mp of the M9 after cropping is really only good for an 8"x12"print without interpolation, perhaps a bit larger depending on chosen ppi. For purely practical reasons, it would be good to know how both images compare at A2 print size, which of course involves interpolation of the M9 image and appropriate sharpening of both images.
I know this raises other matters of contention, upsampling algorithms and sharpening procedures, but that can't be avoided. The NEX-7 image at its native resolution is good for an A2 print. For purely practical reasons one would surely like to know how the cropped M9 image stacks up when both files are prepared for printing at A2 size.
A theoretical flow-on from Michael's test is the issue of the practical benefits of increased pixel count, in relation to lens quality. Full-frame sensor technology has a habit of catching up with the pixel densities that first appeared on cropped formats. The 21mp Canon 5D2 has almost exactly the same pixel pitch as the earlier 8mp 20D. If one crops the 5D2 sensor to the same FoV as one gets with the 20D, using the same lens from the same position, the resulting images are essentially equal in quality in all respects, with only very minor differences, according to DXOMark.
I wonder how long it will take for Sony, or Nikon or Canon to give us a 50-60mp full frame, 35mm format sensor.