See yesterday’s discussion of focus problems with the Nikon 35mm f/1.4G on the D3x. It seems that while the Leica S2 has focusing issues, the Nikon D3x and D3s can be worse.
The example below is with the 12MP Nikon D3s. Obviously, the problem is not body-specific, since both the D3x and the D3s exhibit severe mis-focus. This suggests a general Nikon AF system flaw, or a bad 35/1.4G and a bad 24/1.4G (multiple samples). I suspect the AF system in the camera. I’ll be exploring further.
This is not a mechanical precision issue; the lens zeroes in perfectly using contrast-detect autofocus in Live View mode; that’s how the image at right was taken. In other words, the focusing mechanism has pinpoint precision, if operated appropriately by the camera software. Which implicates the camera AF software, or some kind of problem between two Nikon pro bodies and multiple Nikon lenses.
I shot these tests in AF-S mode (not AF-C) using the single center autofocus sensor, which should be the most accurate. Both cameras were set to Focus Priority. One reader suggested using AF-C mode (continuous), and that is something I will explore, but it’s nuts to have to think that AF-S should be less precise, for if anything it ought to be more precise. But AF-C is extra work, because I often focus lock, then recompose, which means I’d have to always use the AF-on button, which is awkward in many situations. I don’t want a nose instead of the eye, just because I’ve recomposed slightly.
Two professional photographers with the 35/1.4G have emailed to say that their 35/1.4G is working fine, but neither of them made any defensible test (e.g., Live View as a reference and real 3D images of several kinds). I cannot accept evidence of a lens working in one situation as proof of anything, certainly not without a reference frame, and not one example by itself.
If you want to test your own lens (see also Making Sharp Images):
- Use a tripod and shoot at f/1.4.
- Use contrast-detect autofocus in Live View mode as a reference frame.
- Use autofocus for the test shot, compare it to the Live View version by layering in Photoshop; click layers on/off for easy comparison.
In addition to D3s example below, I made a number of other images. Some were focused accurately (or nearly so), but I also found numerous failures similar to this example. I also repeated the same tests on the D3x, with similar results.
More examples with the D3s were added to my review.