I had a thought on this when David made his comment on sharpness, but let it pass without posting.
But now that you have brought it up again, I felt my thoughts might bring an alternate idea to the discussion.
The original posting was about accuracy of autofocus not of sharpness. And to me Jacques’ woodpecker image is a good example of an image with acceptable focus, but low in acutance. And a quick trip into PS with a modicum of Smart Sharpening cleans it up nicely. We often think that sharpness is mutually inclusive with focus, but they aren’t.
There are a number of factors that can affect acuity, focus being one of them. When I look at Jacques’ example, again it appears that the plane of focus is contained within the subject. So we then must ask what caused the low acuity. One thing already mentioned in this thread is the AA filter. Another is interpolation. Jacques mentioned it’s about a 60% crop, but I not sure if he meant that 60% was cropped away leaving 40%, or 40% was cropped away leaving 60%. Then again did he mean that he cropped out a section that would equate to 60%view in an editor such as PS much like the term “100% crop” is used? In either case, it would seem that he would have needed to down-sample the image to meet posting requirements. So perhaps the lower acuity (or sharpness if you will) is due to improper file handling (lack of resharpening after resampling) and not poor AF accuracy.
Also, you often see complaints about the D7000 with people comparing the results with their other cameras. Quite often there a mix of DX and FF 12 MP bodies. An inherent danger when comparing different resolution bodies is a failure to remain cognizant of what you’re looking at with 100% view. Not knowing what Jackques meant by his example, I’ll use an example that I used about a year ago when someone asked what I thought about my AF accuracy after the repair compared to prior to repair (it was repaired for lubricant spray issue not AF, but the teardown necessitated a recalibration of the AF).
For reference this is the full DX frame capture.
Hand held at 300mm with VR on, f/5.6, 1/200th and ISO160.
And this is a 100% crop of the bird.
Now those two images were processed from the NEF.
It appears to me that the AF was fairly accurate, and even the 100% crop looks rather good.
Now, this is a 100% crop of the NEF without processing (no levels or curves adjustments and no sharpening other than capture sharpening in LR3).
The focus was the same, the difference is the acutance. The unprocessed shot does not restore the loss of acutance from the AA filter. Also, at 100% view on the average monitor, you are seeing the effects of diffraction as well. And perhaps that is where a number of people go astray, especially when comparing their new higher resolution bodies to their older lower resolution bodies.
For example my monitor displays about 93 PPI. A D7000 full resolution image displayed at 100% would be about a 57x enlargement. A D90 image would be about 49x and a D700 would be about 32x enlargement. In my example the diffraction spot size at f/5.6 is about 7 microns. My viewing distance is about 30 inches or about 1.6x the viewing distance of an 8x10 in which the tolerable CoC is often calculated. With the D7000 at 57x enlargement the diffraction spot size in my example would become about .4mm; so it would appear relatively soft. With the D90 at 49x enlargement, the diffraction spot size becomes around .34mm and at the viewing distance would be at the edge of appearing relatively sharp. With the D700 and its 32x enlargement, the spot size becomes around .224mm and is well within the acceptably sharp range for the CoC.
So when we analyze real world images, a lack of sharpness doesn’t necessarily mean a lack of focus accuracy. But a lack of accurate focus will cause a lack of sharpness.