This is the third time I’m posting about the testing of my D800 autofocus system.  The first time was not long after I purchased it, just to see if it had the problem.  I found that it did have the problem, and I sent it to Nikon for repair.  When the camera came back, I tested it again.  Initially I thought it may have been fixed, but further testing showed there was still a problem.

During this time, I also purchased LensAlign and FocusTune to help me more accurately fine tune my lenses and camera, but also to help me quantify the results.  When I tested my lenses, I noticed that the recommended fine tune values for three of my five lenses was +20 (the highest possible).  This suggested to me that the actual sharpest fine tune value may be even higher than than, but the scale does not allow further adjustment.  Said another way, if the fine tune is at the maximum value, the calibration is not within spec for the camera/lens combination.  I decided to give my D800 and my Nikkor 50mm f/1.8G, Nikkor 80-200 f/2.8D, and Nikkor 24-70 f/2.8 to Nikon along with my results to see if they can fix the left autofocus problem plus calibrate the camera for the three lenses.

I finally received my camera back, and had a chance to test it with two lenses.  I tested with the 24-70 and the Nikkor 105mm f/2.8 Micro VR (this lens didn’t go to Nikon).  I did not fine tune the camera and lenses before testing, this is how Nikon sent it to me.  I used the same test procedures I outlined previously, with a small twist because I now have FocusTune to help me out.  The results here are all RAW conversions (the only adjustment I made in Lightroom was to convert to grayscale).  I shot in RAW+JPG (L/Fine), I shot a manual focus control shot for each of the five focus test points tested (center, top, bottom, left right – all at the farthest AF points).  Between each autofocus test shot, I defocused away from infinity and took five shots for each focus point.  I then loaded all of the JPG files into FocusTune and for each focus point I found the SHARPEST result from the five samples, so the results here are the best shots from the group.

One thing to note, last time I sent the D800 to Nikon the work order showed “ADJ FOCUS SYSTEM” and “ADJ AUTO FOCUS OPERATION”. This time, the work order had:

[box]

B2

Service Repair Rank B2

Repair

RPL O-RING

CLN CCD

FIRMWARE UPGRADE

GENERAL CHECK & CLEAN

RPR ELECTRICAL CIRCUIT

[/box]

It is worthwhile to note that I was not having any problems with my D800 outside of the autofocus system, I’m not sure why they replaced the o-ring and worked on the electrical system. I see no mention of them even looking at the autofocus system.

 

Lenses Tested

AF-S NIKKOR 50mm f/1.8G

AF-S VR Micro-NIKKOR 105mm f/2.8G IF-ED

AF-S NIKKOR 14-24mm f/2.8G ED

AF-S NIKKOR 24-70mm f/2.8G ED

 

Test Results

The grouped results for the 105mm lens, 100% crop under the autofocus sensor tested.

Apologies that some of these are dark, I didn’t want to make any adjustments so they are presented straight from the camera with no editing.  I don’t see any major issues here, the focus system seems to be fairly consistent between all of the points.  Here is a closer look at the 105mm left side results.

 

I would say this is fine, doesn’t quite match the manual focus using live view and 100% zoom but I don’t think phase detect AF is ever going to be as good.

The grouped results for the 24-70mm lens at 24mm.

Here, the left side AF sensor is clearly softer than the others in the group.  Here is a bigger version of just the left side results.

 

The grouped results for the 24-70mm lens at 70mm.

Once again, some softness on the left side.  Bigger version of 70mm left side results.

 

I would say it’s not as bad as the 24mm left side test, however I would not say this is acceptable performance.  The lens works fine at the other focus points, here are details of the performance at the center focus point at 24mm.

 

The 24-70mm center focus point at 70mm.

 

Both of these are completely acceptable and would not give me any reason for concern.  However, the left side AF sensors continue to give me grief.  I’m not sure if it’s the lens or the body though, because the 105mm seems to focus fine at the left side.  I’ll try to test with my 50mm f/1.8G to see how it performs, and I’ll post the results here when I’m finished.

 Update, April 2, 2013

The grouped results for the 14-24mm lens at 24mm, center and lower AF points are good, others soft.

 

The 14-24 @ 24mm L5 sensor detail.

 

 

The grouped results for the 50mm, again center and lower AF points are good, others soft.

 

50mm L5 detail, not terrible but could be better.

 

In general, the whole AF system seems to be a little off.  Fine tune won’t fix this as the various AF points don’t correlate to each other in how far out of focus they are.  If they were all equally off, fine tune could fix that.  Some, such as the center and lower (C, D2) AF point are excellent, others (such as L5) are consistently off.  If I fine tune, I’ll be throwing off the good ones.