Title picture: Guess who’s face is in focus? That’s right, not the player’s …

As the more frequent readers of this blog (if there are any) may have noticed I am a happy and loyal user of Pentax. When I bought my first DSLR in 2011 I chose it over the more obvious Canon and Nikon, and the emerging mirrorless systems by Sony, Olympus, Fuji and Panasonic for a number of reasons, among the most important were ergonomics, value for money, image quality, general workmanship and weather-sealing, in-body stabilisation and all the other sensor-related features that enables. One reason that certainly did not drive me to Pentax is the brand’s AF system. While perfectly adequate for single shots and most slow-paced aspects of photography, Pentax DSLRs have always played second fiddle to the big players when it comes to fast and reliable AF.C and subject tracking in sports and action photography. I am not a sports photographer and thus AF performance does not overly concern me, but the odd times I do shoot (or rather ‘try to shoot’) friends mountain biking, or playing football, or sliding down a snowy mountain side, the limitations of my camera’s AF system are immediately obvious.

With the introduction of its first full-frame DSLR, the K1, this spring Ricoh Imaging have launched the Pentax brand into a new era. Finally Pentax users, who so far were limited to either APS-C and smaller formats, or the massive but functionally limited medium format, can play with the big boys and wave around a proper, professional tool. The K1 has many redeeming qualities, among them possibly the best image quality of any full-frame camera on the market, a rock solid construction, a rather adventurous new screen construction, built-in pixelshift, shake-reduction and star tracking and many more. All of these features were recently pointed out in a review by Digital Photography Review (DPR), the web’s most visited photography related website. Still, as soon as the full review and verdict had been delivered, a storm of protest arose that saw the comment section swell to 2,412 comments (the largest number I have ever encountered on DPR), many of them highly critical of the results, some of them dismissing them outright. What had happened?

One of the most important aspects of modern cameras is their AF performance, and some time ago DPR introduced a test to evaluate AF.C by making the camera track a cyclist moving towards it at a moderate speed. They did the same with the K1 and concluded:

“Unfortunately, the improvements really don’t seem to make that great of a difference in terms of performance as the K-1’s autofocus system behaves in much the same manner as the K3 II. Even in the most basic, single point AF shooting modes, the results are far from what we would expect from a modern DSLR focusing system. The autofocus tends to hesitate, even in AF-S mode with the center point – meaning it’s not as consistently fast as most Canon and Nikon DSLRs. This hesitant behavior is more noticeable in AF-C mode, with focus falling behind the subject then having to jump to catch up. Subject tracking – where the camera shifts the AF point automatically to follow your in initial subject if it moves away from the initial AF point – has a very poor hit rate and seems to default back to infinity once focus is lost.”

They further elaborated:

“In addition to those issues the subject tracking mode seemed to have a great deal of trouble with even the slightest amount of movement. In our test designed to simulate a subject moving unpredictably at a moderate speed (such as a small child running towards the camera) the subject tracking failed nearly 85% of the time with limited attempts to reacquire the subject after losing focus.”

Immediately Pentax users all around the world started frothing at the mouth, calling the methodology flawed, downloading the original files and pointing out that the EXIF readout indicated a different AF point was used from the one reported, and so on, and so forth … I, personally, did not froth, or fume, or rage, rather I took DPR’s results seriously, for the simple reason that they exactly mirror my experience with the K3.

Now, to get that out of the way – I love my K3. It is a brilliant camera that I have used with glee for the past 1.5 years and do not plan to replace for at least another 3 or so years. It can do so many things well, except its AF.C is complete rubbish. Some months ago I took it to a university football (or ‘soccer’ for out American friends) tournament, accompanied by a friend with a Sony A6000. As I was getting low keeper rates, just for fun he one-handedly pointed his camera in the rough direction of the action without actually looking at it and kept the shutter pressed for two seconds. Going through the images, out of over a dozen all but one were in focus and acceptably sharp. Meanwhile I couldn’t even dream of such a rate with all the preparation and tweaking of settings in the world.

But let me elaborate. Tracking a subject coming towards you requires the camera to a) acquire focus on whatever you are aiming at (most likely the face or chest) and then to maintain that focus by continuously driving the AF motor in the right direction at a speed that matches the subject’s speed. The faster and more erratic the movement, the more difficult the task, the slower and steadier the movement, the easier. The best AF systems use a mix of information consisting of the actual position of the subject and its past movement to analyse the speed and drive the AF based on its prediction of where the subject will soon be. In the easiest scenario the subject is coming right towards you and you can choose one AF point and keep it on the subject’s face, in a more difficult scenario the subject may always move laterally and you either need a surgeon’s hands to keep that one AF point on target, or you enable a wider range of AF points and let the camera track the subject.

The way my Pentax K3 behaves is that even with a subject coming straight towards me (meaning I can simply use a single AF point and the camera does not need to 3D track), it will acquire focus once, but then stop and let a certain time pass before it re-evaluates and focuses again. Within that time even a subject simply moving at a slow walking speed will have left the focal plane before the camera reacquires focus, and shots turn out blurry. With subjects moving faster (but by no means super fast), such as a slow jog, as many as 80% of frames are out-of-focus. Yes, this is a K3, not a K-1. It is older and has a less advanced AF system, however it is not that much older, and from the looks of it the AF system has not advanced that much. Most importantly, my observations have been mirrored by Pentax users around the world and showcase the exact same behaviour that DPR reported in its test of the K1’s AF.C. For a reasonably modern DSLR, that performance is atrocious, and it is really annoying. Sure, no one buys a Pentax specifically for action sports, but almost everyone will want to shoot their playing kids, or their friends playing football, or a flying bird at some point, and what Pentax DSLRs deliver in that department is still behind what your average mid-range Canon or Nikon DSLR could already do 5 years ago. But what is even more annoying is that for years every new iteration of Pentax’ SAFOX AF system has promised major improvements, and each and every time they have disappointed. The system has been tweaked, a couple more percent of keepers edged out, but the big step that Ricoh needs to make, namely move from the focus – hesitate – refocus – hesitate approach to a continuous, ideally predictive, drive is simply not being made.

Funny enough, every time the issue is brought up in Pentaxland (such as Pentaxforums or the Pentax section of the German DSLR-Forum) there are users who corroborate these observations, but there is always a hard-core of fanboys who simply won’t have it and are ready to blame anything and anyone else (preferably the user) but Pentax. Sample shots have been delivered to prove that Pentax cameras can reliably track subjects, but these have often simply been single shots and everyone can pick that one sharp shot out of a sequence of a dozen blurry once, and even when a sequence was presented either the subjects moved more laterally than on the z-axis, or shots were missing from the sequence, or the subject was shot with a wide-angle lens set to f/8, providing a depth-of-field great enough to mask any focus errors.

I have been lectured on how I am doing all of this wrong and how my poor results can only stem from user error, however even after painstaking descriptions of my methodology, including lenses and AF modes used, AF hold options and focus point selected etc., none of the great masters were able to suggest what I am apparently doing wrong. Now, after 1.5 years, many tutorials and about half a dozen sport events shot, I can confidently conclude that no, it’s not my fault, it really is the camera’s. I will no longer accept responsibility for something that I have no control over. The only limiting factor I acknowledge is that none of my lenses is of the super-fast category designed for action, rather most use Pentax older screw-drive technology. The mostly slow lenses of the Pentax system have often been blamed and used as an excuse for the camera’s slow AF performance, but it seems this excuse is not a valid one, because a) it is not responsible for the behaviour described above, which is that the camera simply stops focusing for some time (even a slow AF motor could work continuously rather than just stopping) and b) I have experienced the same lens, a Sigma 17-50 f/2.8 HSM, with a reasonably quick AF motor performing much better in AF.C on a Nikon DSLR compared to my Pentax. Also note that DPR used the modern and ostensibly fast D FA* 70-200 f/2.8 DC in their test, so again the lens is certainly not to blame.

And for that exact reason I was not a bit surprised to read DPR’s conclusions about the K1’s AF.C performance. Some of these days when I find the time I will grab my K3 and Sigma 17-50 f/2.8 HSM, shoot several sequences of images showcasing the problem and add them to this post with all the shooting parameters. But for now, all I can deliver is a description and let the fanboys rage.