As I see it, tracking AF is the mechanism that attempts to select the appropriate focus point based on information provided by the light metering sensor. Except for the increase in AF points, the rest of the autofocus mechanism, at least to me, feels largely unchanged between the K-5 and the K-3 (except that the K-3 doesn't FF in tungsten light anymore). Lenses don't necessarily focus faster. In fact, where SDM is concerned, I don't think that has changed since the K10D, because the motor is in the lens. At least when we are talking about pure rotational speed, e.g. how fast does a given SDM lens go from closest focus to infinity.
What has improved is that the AF doesn't stop as much on the way. I think that older Pentax bodies used to stop a few times on the way to correct focus, presumably so that the camera could get a bearing on how much further the focus needed to be. This was actually independent of the AF motor itself, as it occurred both with both screw-driven and SDM-driven lenses. I remember, when I had a K100D Super and got to play in a shop with the K-7 for a minute, how much snappier it was. Newer cameras are probably a bit better still. Also, the in-body motor has been upgraded over time to be faster. The one in the K-3 is much faster than that in my old K100D Super.
However, most of the time, my K-3 still has to adjust focus at the last moment. So the lens does not stop where it should. It either over- or undershoots, then gets to the correct position, and focus is confirmed.
I have recently played with e.g. the 16-85 DC lens in a shop for a few minutes, and I did notice much less over/undershooting. I think the DC motor has greater torque, so it has greater control over the focusing mechanism, and stops where it should.
As for this new APS-C camera, what I hope for in terms of autofocus is a) that the AF points cover a greater horizontal area and b) that the active AF point is always illuminated while using the tracking mechanism. If they could implement SR on the AF array, that could also be awesome, but I'm not counting on it. That's about it. I think AF on the K-3 is already very accurate, if it gets the time to focus.
Any further improvements would come in the form of new lenses. I feel that the 50-135 and 60-250 especially would benefit from a better AF motor, both in terms of torque (minimizing over/undershooting) and in terms of rotational speed, so that once the tracking mechanism has selected the correct AF point, the AF gets there more quickly, it doesn't have to adjust as much at the end, and focus is still accurate when the shutter is released. The 16-50 isn't really slow to start with, and it's not a tele lens, so I feel there's less of a need for blazingly fast autofocus on that.
Just my perception and my thoughts, for what they're worth.
---------- Post added 08-16-2015 at 04:02 PM ----------
Originally posted by falconeye At the same time, in that forum, I see tons of photos which are jaw dropping and rather difficult to get (and of a kind I don't see here at PF). The group of 3 bikes in a motor bike race curve with just the middle bike in focus (ok, technically feasible w/o AF too), the flying eagle with all head feathers, the dog running towards you with everything but the dog out of focus, you name it. OTOH, I met a few of those guys in real life and some are full time pros loading gear into their cars worth more than the car ... So, results may more be a sign of passion, experience, value of gear and time invested than a brand difference in AF performance.
I also sometimes see shots like that. And it does make me wonder, "in choosing Pentax, am I missing something"?
Originally posted by falconeye I wonder why we discuss it in this thread?
Good question. I actually joined the discussion without knowing how we got here.