A few other take ways from the study.
Under the performance graph
Quote: We see different behavior of devices that can be classified into these categories:
Constant
DevicesandtheirAFsystemthatfallintothiscategoryman- age to keep the AF performance in a relatively small win- dow, steady above 80%. That means that the devices can actually keep the object in focus when it is moving. An example for this is the Panasonic GH4 in all presented con- ditions.
on/off
Weobserveddevicesthatmanagetokeeptheobjectinfocus while moving, but fail after they performed well for some images. So the system lost track of the moving object, some- times getting it back after some images. Example: Sony alpha6000 in both modes at 300lux.
swinging
While the devices manage to keep the focus somehow onto the object, they show increased fluctuation from frame to frame. This means that the focus is basically tracking the general movement of the object, but fails to put the focus exactly onto the object. Example: Nikon D7200 at all con- ditions, most obvious at burst mode with 1m/s movement.
notrack
Somedevicesfailedtotracktheobjectatall.Weseethatthe AF-Performance is good on the first image, but it constantly lowers with every following image. An example is the Leica T in burst mode. Under 300 lux, it managed to re-focus once, but otherwise did not follow the object at all
and further
Quote: Conclusion
• Whentestingthelimitingresolutionofacamera/lenscombi- nation, we have to assume that the AF system is not capable to achieve the best possible performance, even when select- ing ”best of 10”. A MF sequence is required. We performed this test for the limiting resolution only, but it extremely likely that this result is also true for all other metrics that are direct or indirect related to the SFR of the system under test.
• EventhoughthemaincomponentsofanAFsystemarepart of the camera, the lens also has a significant influence. So when comparing cameras with each other, it is important to make wise decisions which lens is used for the evaluation. if possible, best solution is to use the same lens for different cameras.
• To track a moving object is still a difficult task for todays cameras and there is still a lot of room for improvement.
• WeseethatD-SLRcamerashaveatendencytohavealower
226
IS&T International Symposium on Electronic Imaging 2017 Image Quality and System Performance XIV
accuracy and repeatability in comparison to system cameras.
It is such treat to read some science instead of anecdotal mental wanderings.
Two comments of note.
Quote: • To track a moving object is still a difficult task for todays cameras and there is still a lot of room for improvement.
You can pay a lot of money for cameras that don't actually solve your problem.
Quote: • Even though the main components of an AFsystem are part of the camera, the lens also has a significant influence. So when comparing cameras with each other, it is important to make wise decisions which lens is used for the evaluation. if possible, best solution is to use the same lens for different cameras.
There's the science behind why you should by a DA 55-300 PLM or DA* 70-2000 if you want to make Pentax AF work. Pentax probably has fewer lenses optimized for sports than any other system. And the lens used does make a significant difference, from a scientific perspective, that means, yes you do have to buy a fast focussing lens to test Pentax AF performance. The difference is measurable and repeatable scientifically, regardless of what beliefs various forum posters might hold.
Pentax overall doesn't do as well, but there are some spreadsheet lines where it leads the pack... something not noted by the hundreds of anti-Pentax posters posting the "Pentax AF sucks" theme. Believe it or not, the science doesn't support you in assuming Pentax AF is poor in every situation. Sometimes it's the best.
Not that I expect most of the anti-Pentax posters to be able to read the science and alter their view points. The ability to understand the science is also a talent not everyone has.
The science says exactly what I believed, Pentax is not quite as good as most (but better than Lieca), but not as bad as everyone says, and capable of being the best in some circumstances. Sometimes you get shot with your Pentax everyone else misses. And nobody is perfect all the time.
The best take away from this is AF is not some one element constant than can be described in one fashion. The authors list four different ways to evaluate AF. And no system is "the bet' at all of them. They all have strengths the other don't and they all have weaknesses the others don't.
Which system is right for you? That's anybody's guess. But one dude telling you how bad pentax AF for what he does, means nothing, unless you shoot exactly what he does (a statistical improbability.) The fact that it isn't right for everyone in no way implies it isn't right for you. And regardless of what brand you shoot, if you want something you camera doesn't do well, accept the trade offs, and buy something else. After all, we all want you to be happy.
But please don't come here and imply that the Pentax trade offs in AF design aren't the best for what I do. That's not your call.