I wouldn't be obsessed too much with tracking, unless you photograph dogs running towards you, or cars or whatever else is approaching at a constant, easily predictable speed. All cameras are comparably pretty good at this easy task.
However, what makes Pentax AF often useless for any serious people photography is not related to tracking. But instead to two unrelated flaws:
1- huge AF areas, each covering a big portion of the motiv, sometimes (outside the center spot) in awkward big T or L shaped areas.
Mind that the tiny squares in the view finder are only about 1 to 10% of the true area size of the AF regions.
For the German test, where they used a two dimentional flat "wall" to approach the camera, this is very favourable for Pentax.
However in real life situation, e.g. heads of persons (or animals) being a bit further away, thus appearing small to the camera, huge AF areas tend to fail.
A Nikon D7200 would fare well, because its AF points are really only just AF points, rather than huge oddly shaped AF regions as in Pentax cameras.
2. Pentax is bad in detecting (and tracking) small, erratic, random movements, as typical for e.g portrait photography with shallow DOF.
It is a pity that the German test had only a very simplistic test setup, only simulating a flat wall coming perpendicularly towards you in absolute constant speed.
In a better setup simulating real life problems, let's expect that the K-3 would have failed a lot, confirming the bad reputation of Pentax AF implementations out in the market.
Last edited by Frater; 03-25-2019 at 12:23 PM.