Originally posted by clackers This is where the Hold setting becomes important. If your technique is shaky and you let the subject stray repeatedly outside the points, this needs to be cranked up.
Again - perhaps DPR were guilty of this - you need to keep the subject within the focus point area.
That's traditionally easier to do on an APS-C DSLR because that proportion is usually bigger.
PS A problem with Auto 9 or whatever is that the bike's handlebars will be closer to you than the cyclist's face.
When the camera is mounted on a tripod you have no control to keep the points on the subject. At f/2.8 nothing can keep the handlebars and the face simultaneously sharp. They re-ran the test with AF.C (M) 9-S and put the S point on the cyclists head for the first shot - still got bad tracking AF.
I exchanged a number of civil messages and posts with Rishi and Chris. I don't think they're out to 'get' Pentax. I
do believe they have an inflated sense of their position as a source of information and guidance to consumers, and they're convinced they are right, but they both respond politely to civil questions and suggestions. I disagree with their conclusions, as is my right
[above added after some further thought 01/24pm Sunday]
This is what I
think happened in the test.
The test was designed to rate predictive tracking autofocus across the entire image area. The Pentax autofocus sensor was designed for APSc cameras; the focus points are concentrated in the center of the image area (within the crop frame) leaving wide areas of the image area without AF points.
The bicycle weaved in and out of the focus frame (the crop frame). When the bicycle left the focus frame the camera went to infinity. When the bicycle came back in it took the camera time to reacquire focus - and then the bicycle went out the other side . . . .
There are too few (3) f/2.8 cross-type focus points (of 25 total cross-type) so at f/2.8 the only time focus acquisition is quick is when the bicycle is in the exact center of the viewfinder. If you use Hold 3 the bicycle moves itself out of focus when it is out of the focus frame. Ergo, DPR says only 15-20% of the captures were in focus - they actually may be correct, even though Falk Lumo says there
has to be a problem with the gear. Falk says
no modern dSLR achieves a tracking AF.C 85% failure rate.
The problem for K-1 on that test - and it's a real, actual fault of the Pentax AF.C system and physical properties - is there is no possible technique to compensate
when the camera is on a tripod. K-1 was destined to perform poorly on that test. The test actually, correctly demonstrates a shortcoming of the Pentax tracking AF.C system.
My point is, so what? That test is not meaningful as an indicator of real-world competence of the camera. Failing that test does not mean the entire AF system is Poor (their rating). Further, to then conclude the K-1 is not suitable for burst-photographing small, active children is not a supportable conclusion. In the real world,
who takes action photographs of small active children at f/2.8 from a tripod?
What we now know is, the K-1 was not designed to pass this kind of test. And
that's all we know.
IMHO, since they have a Ph.D. writing camera reviews, they should have
explained the cause of the failure and offered some suggestions to correct for it using
technique if a buyer wants to, for instance, burst-photograph small, active children.