With the Nikon D850 loaned to me by a reader, I found a substantial error in the Nikon D850 flange focal distance which required all my Zeiss Otus lenses and others to be racked to the mechanical limit in order to achieve infinity focus—a very bad sign.
Further testing showed an unacceptable sensor/mount misalignment resulting in a badly blurred right side (if focused at left or center). I returned that D850 to its owner, who is returning it for exchange for another one, a course of action I recommended and agree with—it was just too far off to make me happy about adjusting it.
Early this morning I mounted the excellent Really Right Stuff L-bracket for the Nikon D850 (all my cameras use RRS brackets), and proceeded to check for left/right symmetry as per Testing for a Misaligned Sensor.
In general I am pleased—this Nikon D850 is one that perhaps is not absolutely perfect (too close to say), so it is good to go for work I had to abort with the previous one.
I found that within a very small margin of error (my ability to focus, small lens deviations, small sensor/mount deviation, etc), I would call it “in spec”. I would need more testing with suitable subjects and dual cameras (D850 and D810) to say if the camera is off: I think the sensor/mount parallelism might be off by a very small amount, but it is as good as I might hope for out of the box. I might send it in at some point and ask for perfection, that is, perfect parallel alignment to the lens mount as good as Nikon can achieve. This was done once for me with the Nikon D800E, and my Nikon D810 is perfect as far as I could say for 3 years. For now, I feel the D850 is good to go.
- The Sigma 85/1.4 DG HSM Art was very slightly asymmetric. The asymmetry is within a tiny margin of focusing error. Indeed it is less than the margin of error I saw with that D850 AF system with the Sigma. I recommend to not rely on conventional AF for pin-sharp results (at least at f/1.4 at far distance), and from what I saw in Live View, I would not rely on contrast detect AF in Live View for peak results without verifying best focus; I found I could improve sharpness by manually focusing at 100% or 200%.
- The Otus 55/1.4 was symmetric, within my ability to check on the subject at hand.
- The Otus 85 was very slightly asymmetric, as a bit concerning since I have special MTF symmetry data for it from Zeiss, and it is as perfect as one could ever build, with 2% variation to the corners, I wouild call it cine-lens grade symmetry. That suggests that the D850 sensor is not perfectly parallel to the lens mount.
- The Sigma 135/1.8 DG HSM Art looks symmetric, but longer focal lengths are less sensitive to the issue; I’ll see what further field tests show.
To emphasize, the very slight asymmetry I am seeing is at a level which will be seen only by careful scrutiny at 100% or 200% at f/1.4 or f/2 in Live View (or taken image). I can’t hold Nikon to a tighter standard than that in a mass market camera, though I would like perfection or as close to it as technicallly feasible. Tolerances are extremely demanding with more megapixels, for cameras and lenses alike.
Flange focal distance
There is one thing I am not happy about with this new D850, and that is that my Otus lenses use about half their focusing range beyond the infinity mark, as shown below whereas the Nikon D810 focuses at or very close to the mark with the same lenses. So that flange focal distance tolerance seems to be off, just as with the prior problem D850. It is concerning in that very cold temperature might make infinity focus unreachable (or at least I think it is cold that would be the issue, versus, say, 110°F heat).