As clear from several internet forums, multiple people encounter autofocus problems with their Nikon D800 DSLR. The success rate of the repair seems to vary a lot, depending on the status of the camera, or the quality of the repair (centre) itself. In this post I will share my latest findings on the Nikon D800 calibration procedure applied at Nikon Service Point in the Netherlands (NSPtN), which manages to solve the autofocus problem completely!
Although an initial calibration of my D800 provided substantial better results, there was still something to be wished for (e.g. my 14-24mm lens required additional calibration). After some additional tests and contact with Evert van Stijn from the NSPtN, my camera got 2 more calibrations, with eventually very good results. The reason for posting these calibrations results are 2 new main observations:
- Autofocus Distance Matters: Although the camera may autofocus correctly at close distance (say < 1m), it may not focus correctly at larger distances. This was not a standard check in the calibration procedure, explaining why I still did find some focus issues.
- Autofocus is not perfect: Although there are still some visible autofocus issues, these are not specific to the D800 and also visible on for instance the D700.
3. Measurement Procedure
In a home situation it is difficult to build up a test setup that is perfectly calibrated. Setting up a camera on a tripod straight to a plain surface is not easy to accomplish. Also, some lenses slightly bend the surface because of field curvature effects, which become slightly visible at extreme focal points using very wide open apertures of some specific lenses. In such a case it makes sense to make relative comparisons. For a selected focus point, a first picture is made using live view, which provides a direct optical path to the sensor. Based on the same focus point, a second picture is made using mirror view. In an optimal case both these pictures should look the same for a lens that has been correctly AF fine-tuned (there are some differences in exposure).
The following example shows my own test sets. The first test is a book shelf filled, providing a lot of contrasty details. If you are an autofocus neurotics, you can also glue special targets to the bookshelf (in practice I see that this doesn’t make a difference at all…). I’ve also positioned a LensCal element, to determine whether the camera front- or backfocuses. This situation is photographed at about 1 meter and 2 meters.
The second test is outside, photographing some houses.
Subsequently, I’ve made overview pictures consisting of 6 crops at 100% magnification, where the comparisons should be made from left to right for relative comparison:
The result for a test at a distance of 1m for a 50mm/f1.4G lens at f2.8 looks as follows (click on the picture for a larger version):
At the position of the lenscal (using image processing to emphasize the sharpness), where one can see everything is in perfect focus for all focus points:
The result of the calibration of a camera and lens combination is expressed in tables. These tables show 3 tests, one where the distance between the object and the camera is about 1m, one where the distance is 2m, and one where the distance is “far away” (> 100m). In case of 1.4 lenses, I also perform the test at f2.8. The values in the cells of the table denote the amount of front or back focus. Severe front or back focus with visible sharpness effects is denoted with ++ or — and a red colour. Front or back focus with no visible effects is denoted with one + or -.
The 14-24/2.8G, shows overall good results:
The 24-70/2.8G still has the tendency to strongly back focus at the left side when using it at its wide angle range (24-35mm). It also back focuses at the right side with large distances. As will be shown later, this behaviour is also visible with the D700.
To get an idea of the effect, the following two crops show the results for 24mm@2m and 24mm@100m (click on the picture for a larger version):
The following crop shows live view versus mirror view underneath the selected focus points, where the difference in sharpness is less prominent, but still visible.
The following crops shows how autofocus results changes from left-to-right:
The good news is that the sharpness underneath the focus point itself is OK-ish for short distances (but still very different from the live view situation in the overall picture). Luckily, my practical use-cases hardly will give me the above situations:
- Most of the times I use the centre focus point, and re-focus
- I use the left-right focus points only at short distances, and the sharpness underneath the focus point is quite OK in those cases
- For far distances I hardly use the left-right focus points, and mostly apertures > f5.6
So in practice I assume this setup will work fine for me.
The 50mm/1.4G shows overall good results:
I’ve also test my 60mm/2.8G macro lens, which is extremely sharp all over:
The 70-200mm/2.8G also gives good results:
Finally, my (favourite!) 85mm/1.4G lens has a strange front focus effect at the right-most point. This is restricted to two auto focus points, the right-most, and the one above that, all the other points are razor sharp. The D700 shows exactly the same effect, which is kind of weird.
Using the 85mm at such a wide open aperture for long distances at the right focus point is a very strange case, so I’m fine with the above results, and consider the lens – camera combination to be fine. For an impression of the effect, see the picture below (click for a larger version).
To get an idea of the differences between the various calibrations, I’ve constructed the following table (click for a larger version). The first tests I’ve only photographed at one distance (between 1 to 1.5 meter), therefore the table only shows the common case. In general you can assume the results would have been worse for larger distances, so I would expect the table to be filled with more red boxes if those cases would have been included as well . As is clear from the table, the last repair shows significant better results.
5. Comparison with Nikon D700
As is obvious from the tables, the results are not perfect in all cases. For instance the 24-70mm lens has issues at 24mm. The “good news” for the D800 is that my D700 shows even worse results for these cases. This means that the focus errors are not specific to the D800, which implies that the D800 can be considered to be fixed in my opinion.
The picture below shows the mirror view results of the D700 on the left side, and the mirror view results of the D800 on the right side. I’ve scaled the D800 pictures to roughly the D700 resolution.
24-70/f2.8 @ 24mm – f2.8 -1m
24-70/f2.8 @ 24mm-f2.8 -100m
85/f1.4G @ f1.4 – 100m
The D700 results are as follows, and as you can see it is even worse than the D800:
Although mirror-based autofocus is still not perfect, I consider my D800 to be fully fixed. All ‘n’ all I’m very satisfied about the current status of my D800 now, and if I had to grade it, it would get a 9.9 out of 10!
Special thanks goes to Evert van Stijn from Nikon Service Point in the Netherlands. His intrinsic interest in finding the root cause of the problem, the time he spend with me on the phone and in the lab shows that he wants to improve the calibration procedure to be able to provide a better service to all of us. It is people like him that eventually create satisfied customers.
There is a critical word to be said about Nikon Japan, who seem to think Silence is Golden. If the problem is small, it is a small gesture to take back the problematic cameras, and replace them instantly. If the problem is big, I expect a better roll-out to the service centres, and more clear guidelines to the user community. No information at all creates a lot of fuzz and emotions in the forums and blogs.
Finally, I wish anybody who is affected by the same issues good luck with resolving it with their local service centres. I hope the info on my blog helps to get your case forward. Happy shooting!