NOTE: Click on any image to see it at 100% crop view with caption describing shooting settings.
These images (shot using a 28mm F1.8G lens) show a series of images from a once-repaired D800E with targets shot at 1.4 metres in lightly scrimmed daylight, using a very sturdy tripod and head. Auto Focus Fine Tune is set for +6 in order to get the best performance from the centre point. They clearly show that the left point Phase Detect AF varies from the right. This is the problem for which it was originally sent in. The left point produces the most accurate Phase Detect AF at a fine tune setting of -12 but at -9 it is 'as good as' Live View Contrast Detect AF. So if we take the benchmark for Phase Detect focus as being not optimal focus fine tune, but a fine tune setting that matches Live View Contrast Detect, we see that the number is -9 for the left point, whereas the centre point matches Live View at +6. This is a range of 15 between the 'right' fine tune settings for Centre and Left points. There is no compromise value that suits both at all well, and if I set it at +6 so that the centre performs well, the left is unusable. The right point, at this setting, is not great but is just about acceptable to a very generous eye.
The outdoor shots show the exact same frame cropped far left, centre and far right. They are from a twice-fixed 24-120VRII at f5.6, again on a very sturdy rig. The frame was focussed manually in Live View zoomed in, and the results are the same with viewfinder AF. They clearly show that the RHS of the frame is asymmetrically much weaker than the left: this is the problem for which this lens has already been 'repaired' twice.