The DxOMark folks are reporting that the Canon 5D Mark III per pixel sharpness is 15% higher than the 5D Mark II and moreover that it is in practice about as sharp as the D800.
Of course my reaction was one of skepticism: there is a ~28% linear resolving power advantage with the D800, so how could this possibly be? But I’ve learned to guard against assumptions over the years, because a result out of a camera system can be quite different from what one would expect from specifications.
I favor the Nikon D800E (vs D800) with its funky glass sandwich over the sensor, which is claimed by Nikon to be equivalent to not having any anti-aliasing (optical low pass) filter— no artificial blur. (From what I can tell, the smaller 24MP DX Nikon D7100 does not use this approach, but eliminates the extra glass layers).
But might Nikon’s thick glass sandwich interact with a lens in other ways that degrade its sharpness nonetheless? A lens design does have to compute in a sensor glass thickness for optimal results, particularly off-center. And I’ve suspected some per-pixel sharpness under-performance from the D800E sensor for a while now, but could never quite nail it down, never having quite found the right way to do so.
I have now done that exercise—several times in 1mm increments, using the Zeiss 135mm f/2 APO-Sonnar, my new reference lens.
Each time the same curiosities manifested themselves: I can say unequivocally that the D800E is delivering far less per-pixel sharpness than its 28% linear resolution advantage over the Canon 5D Mark III ought to allow. And there are other peculiarities too, particularly at wider apertures. It is an exciting and disappointing finding, and I can only hope that Nikon gets its act together and delivers a pro body that extracts full resolution properly at all apertures.
None of this sharpness discussion should be confused with total image quality, where one must complement the Nikon D800E, and there I found no surprises.
Report to follow. The report will go into DAP, as this is a camera capability issue, not a lens comparison, and I don’t believe it has anything to do with any particular lens. However, it is relevant to what to expect with the 135/2 APO-Sonnar, so it will also be cross-posted in Guide to Zeiss.