Seb Rogers: mountain bike photography: Confessions of an autofocus cynic (part 1)

Regular readers may have picked up on my casual disdain for most things automatic, and particularly autofocus. Sure, there have been occasions when it's saved my bacon. But for the most part, I've found it rarely lives up to the hype and almost always gets in the way.

I might - just might - have changed my mind, at least a little bit. Of which, more in part 2, to follow at a later date. But first, I thought it might be helpful to go into a bit more detail about why I've been so ready to switch off the result of decades of Japanese optical and electronic experience and resort to my left wrist to get things sharp.

There are, in my experience, two major problems with most autofocus systems. First, they tend not to be sensitive enough to focus on smallish, fairly distant subjects accurately, preferring on the whole to deliver crisp renditions of the background. That's a fairly major practical problem when you're shooting, say, distant riders with a longish lens at a wideish aperture. There's not enough depth of field to cover major focus errors, but it's also pretty hard to see on the screen when the riders are perfectly sharp (or not). The only solution has been to shoot plenty of manually focussed images, and check that at least a few of them are sharp.

(There's a related problem, which is that this lack of sensitivity to foreground subjects often means autofocus will 'see' through the holes in even a closer, larger bike and rider.)

Second, most autofocus systems are pretty dumb when it comes to following a moving subject across the frame. Whilst it's simple enough to get around the idiosyncracies of autofocus for straightforward shots by prefocussing manually, the type of situation where you might hope that all those clever algorithms would actually help - capturing a sequence of a rider negotiating an 'S' bend, for example - also tends to cause problems of the 'anything but the subject in focus' variety.

It's enough to drive a poor photographer to distraction, frankly.

Now, it may well be that my experience has been tainted by nearly 20 years of shooting Nikon. Canonistas, before they were shouting about full frame and low high ISO noise, tended to be a little smug about their brand's focus prowess. I don't know, frankly, but I suspect that the reality doesn't entirely match up to the hype. I know of pro Canon shooters who moan about their cameras' autofocus, just as I know Nikon shooters who are perfectly happy with theirs.

Still, I've often fallen for the autofocus hype of a 'new, improved' model... and almost always been disappointed. Here's a brief, affectionate history of the good, the bad and the ugly Nikon autofocus systems that I've tolerated over the years:

- c. 1989: Nikon F801
My first Nikon camera, and it still has a place in my 'spare' camera bag despite an autofocus system that's all but useless. Boasting a single, central sensor and originally launched with lenses featuring tiny, rattly manual focus rings, it would (eventually) rumble into focus... provided the subject in question sat still for long enough. Literally... because it couldn't cope with moving subjects.

- c. 1996: Nikon F90X
I dropped close to £1000 - a fortune in 1996 terms - on one of these, lured by the promise of its wide-area autofocus and compatibility with the newly-launched (and astronomically expensive) 'silent wave' lenses. Breathless reviewers compared its performance to the benchmark Canon EOS-1, but even after upgrading my trusty 80-200mm f/2.8 to the 'new' model (for a further £1100), I wasn't much closer to a system that worked. It'd follow a moving subject, just. But not in a way that was much use to me.

- c. 1997: Nikon F5
Taking a deep breath and shaking every last penny out of the piggy bank, I stumped up over £2000 for Nikon's EOS-1-beating F5. And discovered, at last, a system that worked. Kinda. It needed the new silent wave lenses to reap the maximum benefit, and its ability to track a subject moving across the frame was patchy, but at last I had something that almost delivered on its promises... just so long as the subject played nice and I put it near the middle of the frame.

- c. 2005: Nikon D2X
Nikon's D2 series promised a lot, with 9 cross-type sensors spread widely across the frame and some fancy new algorithms to keep track of errant subjects. Trouble is, I'd already experienced this system in the film-based F6, and the results weren't entirely encouraging. Although it was undeniably fast - and the central sensor was pretty accurate - it had a nasty tendency to 'see' beyond a subject and focus behind it. Small, distant subjects? Forget it.

- c.2006: Nikon D200
The D200's autofocus was, on paper, less impressive than the range-topping D2X. But, although it only had a single cross-type sensor and the outlying sensors didn't cover such a wide area, I soon found that it didn't have the D2's tendency to focus behind the main subject. At least, not to nearly the same extent. It isn't as sensitive as the D2X, but in the real world it's capable of equal or better accuracy.

Which brings us to the present day, and a re-evaluation of autofocus's usefulness in my day-to-day work. Why? Ah, well, that'll have to wait until part 2...