A friend came around the other day so I could assist him in calibrating his D7000 body to his 3 lenses; Nikon 18-105 VR, Nikon 70-200 f2.8 VR, Nikon 50mm f1.8D.
I used my usual test set up with the camera mounted firmly on a tripod and Spyder Lenscal target set on the conservatory table. A combination of natural light and flash was used for the tests.
The first lens up for test was the 18-105 kit zoom. Its a better constructed lens than the older 18-135 but at the long end we really struggled to get satisfactorily sharp images, even using remote release. I tried mirror up and this helped a bit, then finally I used flash and a shutter speed of 1/250th second. That gave the best results, so some induced vibration must have been at work. Its a fairly light combo, so maybe mirror slap was an issue. Even after these tweaks the sharpness results were somewhat disappointing. A minor adjustment to AF fine tuning made a difference but it was clear that the long end was showing signs of optical weakness on the demanding D7000 sensor.
Next up was the 70-200 and this produced instantly stunning results, both in terms of its sharpness and contrast and its calibration with the body. The difference between this lens and the kit zoom at the tele end was staggering.
Last up was the great little 50 f1.8. It lacks a bit of sharpness and contrast wide open but AF accuracy was spot on...I found my own copy my most accurate lens in terms of calibration adjustment.
I then did a run of tests with my Carl Zeiss 25mm f2, which is a stunning piece of glass and has been tested as one of the sharpest DSLR lenses there is. I do get some inconsistency on the D7000, whereby a fair proportion of images are back focused. I do tend to at least partly rely on the AF confirmation when using DX (crop) bodies due to the smaller, darker image in the viewfinder, whereas on FX (full frame) I can usually see the contrast pop. I wanted to see if I could tune the AF to help me get better results with the D7000.
What I found was very interesting. If I rotated the focus ring from a more distant setting to a closer one (in) to focus, I got very significant back focus, plenty sufficient for me to bin images due to blur. If however, I focused out from a shorter distance to a greater one (out), the focus confirmation was as accurate as it could be as soon as the green dot illuminated. I repeated this several times to confirm the issue by focusing from far and closer settings. Obviously at f2 depth of field is limited, even on a 25mm lens, but focus confirmation was fully useful as long as I focused from near toward far distances and stopped focusing at the start of green dot illumination or shortly thereafter.
The best way to show this is the images taken of the test target:
Firstly, this was a typical result with the lens focused from a more distant to a closer setting. Focus was stopped as soon as the focus confirmation dot illuminated. The back focus is very evident and renders the planar focus target soft. Contrast and sharpness is optimal at number 3 behind the target. This can also be seen by noting the presence of red longitudinal chromatic aberrations (LOCA) forward of "3" and green LOCA rearward of it.
Finally, this is a typical result with the lens focused from closer to further distances and again, focus was stopped as soon as the focus confirmation dot illuminated. In this case focus accuracy is spot on the zero numeric value and the focus target is sharp and rich in contrast. I was able to get consistent results using this method. There was no point in dialling in AF fine tune to correct the back focus when focusing "in" as this threw out the accuracy when focusing "out".
This may be a significant finding that may help manual focus accuracy with all my Zeiss primes, but it has certainly been a worthwhile exercise in identifying the issue with the 25 f2 and finding out how to use AF confirmation effectively with this lens. It would appear that the tolerance of the AF system covers a range of 0 to +3 on the ruler scale when focusing in from both directions. The range was far less when focusing in one direction. Yet again, it shows that AF (or AF confirmation) cannot be relied upon when shooting at the widest apertures, at least without some understanding of its limitations.