The D800 Fix: some pragmatic thoughts

Started Oct 4, 2012 | Discussions thread
Flat view
tashley Veteran Member • Posts: 4,186
The D800 Fix: some pragmatic thoughts

There are any number of people who are either waiting to get their left points fixed, wondering if their local service centres are yet capable, or delaying purchase until the situation is resolved.

I have posted before about my experiences and about how my D800 eventually got refunded because Nikon Richmond UK were unable to repair it (even the replacement they gave me at one time was afflicted too). I have also written of how I waited for months before sending in my D800E, partly so I had something to use while the D800 was shuttling back and forth, and partly because I wanted to avoid the frustration!

The E went back a few weeks ago and I've had it back now for long enough to test the butt off it, and to do some serious thinking and research about what constitutes a fix, and what are realistic expectations.

On its return I initially reported that I thought it was about right, having done initial tests. Then I did a longer series and the results looked worse. From this I realised that you shouldn't test for these things under tungsten lights! I was using my studio modelling lamps, pretty bright but still tungsten, and they were giving worse results than the first round of tests which I had run outdoors. Tungsten is very red-heavy and can focus in a slightly different plane: it is recommended to perform AFFT operation with the light you will mainly use, which for me is daylight. Or even know your two sets of values and use them as appropriate.

Now I have done a further and far more intensive round, using scrimmed bright daylight, on my most commonly used lenses. The results are pretty interesting and I will go into some detail here:

Firstly, the most demanding lens AF I own. This is a 28mm f1.8g and I have blogged extensively about its cone (or 'nose') shaped field of focus. It also is prone to focus shift and to various colour aberrations. So much so that Thom Hogan told me it was possibly too tricky to make judgements from. Nonethelss, it is very sharp when correctly deployed, and it is on the Nikon 'approved' list for the D800. And I use it a lot.

Before the repair it was great on centre and useless with the left point, a bit soft with the right point.

I just did a repeated series of tests (tripod, correct distance, target etc) and ascertained with high consistency that:

At F1.8: it needs a +6 on the centre point though a +12 gives very very slightly sharper results. On the left point it needs a -6

At f5.6 (I know I know, PDAF is performed at maximum aperture, but this lens has a shift and odd field of focus and I use it most at f5.6) the results are the same, though the lens is sharper on centre and slightly weaker at the left point than wide open.

So in both cases, there's a 12 point range, left to centre if I am being a tad generous, an 18 point range if I am being 100% demanding.

Conclusion: there is no AFFT that gives good performance at both points, so I opt to use the +6 setting and give up hope of accurate outer point focus.

Now: I could also just throw up my hands and send it back BUT I have read a zillion articles on this, and it's just a fact of PDAF that there's point to point variability in ALL manufacturer's versions.

So I read some more and got in contact with the excellent Falk at Lumolabs. He has determined through extensive testing of many brands of cameras, both with and without Reikan, that with an F2.8 lens, the point to point variation should be within 10 for Nikon, to be broadly in line with the average performance of other systems on the market. For an F1.4 lens, that expands to 20 points.

Not great, but realistic. And my F1.8 experience of between 12 and 18 AFFT points difference equates to an F2.8 lens at between 8 and 11.5 points. If 10 is the normal limit for reasonable expectation, I am about there. So what this means is that on this lens, most modern cameras wouldn't reasonably be called 'faulty' at this level of performance.

Moving on:

My 24-120 lens at F4 performs best at both 28mm and 120mm at or around 0 AFFT on left and centre. No complaints. But then it is an F4 lens...

My 70-200 VR II at 200mm wide open is best for the left point at -5, for the centre at 0 (but it's still pretty good on centre at -5) and I tested the right point here too, since in the studio I often use the right point in landscape orientation to focus on the eye. The Right point seems OK at -5 but really sharpens up perfectly at -10. However, -5 is acceptable, and with a little extra sharpening in post, no-one will ever know. So my most important F2.8 lens has a 5 point range if I am being a tad generous and a 10 point range if I am being picky. Still within the Falk parameters of what is normal, if not desirable...

My 85mm F1.8G is best on centre at +5, and scores slightly better at 0 on both left and right, though the performance is only slightly better than at +5 there too. So the range here is 5 or less compared to a normal outer limit of acceptability of somewhere in the region of 15 or 16 for an F1.8 lens.

So if Falk is correct (and my experience of other current DSLR systems isn't wide enough or systematic enough, but what I do know leads me to intuit that he *is*) then all my lenses apart from the 28mm f1.8G are within normal limits after this repair. The 28mm is either comfortably within or only slightly outside those limits, depending on how picky I am. Given that it is a weird (if also wonderful) lens with many quirks, I reluctantly deem this performance within limits. It's just that even within those limits, it's not usable with the outer points. Not what I wanted but if that is what 'within spec' PDAF technology is delivering with difficult fast wides, then there's no point complaining, and no point sending it back again in the crapshoot hope that I'll get one that's tighter.

Outdoor shooting in the real world seems to confirm these results for the 28mm and the 24-120. The others, I have not tried yet. Nor have I tried my MF lenses to see how the green dot focus performs. And there are other lenses I might get, e.g. the 14mm F1.4, that might show an 'outside spec' behaviour and trigger a later desire to return the camera. But for now, for me, though this ain't great, it sounds like it is par for the course and so... I'm off to take some photos!

-- hide signature --

Gallery & Blog :

Nikon D800 Nikon D800E Panasonic Lumix DMC-F5
If you believe there are incorrect tags, please send us this post using our feedback form.
Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow