Sat 02-Feb-13 01:13 PM | edited Sat 02-Feb-13 01:23 PM by ericbowles
I know the lens has already gone back, but just a few comments for future reference.
The 50 f/1.4 should produce very sharp images, but DOF is very small wide open. That makes it easy to miss focus. But you already know that. The other issue is the 50 f/1.4 is not at it's sharpest wide open. While you should use f/1.4 for fine tuning, it really performs much better at f/2-2.8. So getting sharp images requires a different aperture than fine tuning - and your test images should be a little soft at f/1.4.
In comparison to the 18-200, the zoom has an minimum aperture of f/4 or so at 50mm. That means you are going to have a lot more DOF and many errors will be forgiven. The 18-200 may have issues at the extremes, but at 50mm it is in a focal length where it performs very well compared to any lens including the 24-70.
AF is not perfect. AF accuracy can be viewed as a relative distribution of AF errors over a number of images. If your test equipment is precise, you'll see that every time you focus using AF, the result is a little different. There should be a concentration of AF results at the best the lens can perform, but there will also be a distribution of missed focus that includes some bad misses. There is a thread somewhere here by NatGeo photographer Daniel Cox with a link to his blog. He tested the D4, D800 and D600 AF on a moving dog in great light. With the D600 he only hit focus on 54% of the images - and he thought that was quite good. The D4 hit about 70%. All of the results were better than any prior Nikon camera. http://www.naturalexposures.com/corkboard/testing-the-nikon-d4-d800-d600-in-predictive-auto-focus/
While it did not come up here, the center sensor is the one that performs best on Nikon cameras. The surrounding cross sensors perform slightly less well, and the other sensors less accurately still. This also applies to the amount of light required wide open - the center sensor is the one that focuses at f/5.6 or f/8 and other sensors may only provide focus at f/4.
Phase detect AF is slightly less consistent and less accurate than the contrast detect AF of Live View. Live View will be better and more consistent, but it is slower - and not practical for some images. Still, Live View provides a perspective on what the lens is capable of delivering if properly fine tuned. Think of it as a baseline.
There are a wide range of AF tests for different factors. You can test to fine tune the lens. You can test for performance of phase detect focus on specific AF sensors. LensRentals.com does most of their testing with the center sensor only - other sensors are not going to be as accurate and introduce many other variables.
Stick with the 50 f/1.4 - it really is a very good lens in spite of the copy you received and issue you experienced.