Originally posted by BigMackCam Just to show that the K-5's AF is reliable with a fast lens, I started from scratch with adjustments for my FA50/1.4...
I set up my SpyderLENSCAL and camera as below:
This is a pretty poor setup, really... Typically, I would mount the camera on a proper, sturdy tripod, but just wanted to do this quick and dirty test for the OP's benefit.
I reset the AF fine adjustment number to zero, set the camera in Live View, and adjusted the position of it until it was pointing at the dead centre of the black and white checkerboard. I then cycled AF, looked at the distance readout of the lens, and kept adjusting the position of the camera and recycling AF until the distance showed 1 metre every time:
Next, I switched off Live View, set the camera to use a single centre AF point, put my hand in front of the lens and cycled AF so it would be out of focus (this lens doesn't have Quick Shift
), took my hand away and cycled AF once more, all without moving the camera. Initially, the distance readout on the lens was slightly to one side of the 1 metre mark, so I made AF fine adjustments and repeated the process until I got a 1 metre readout every time (sorry for the shaky handheld phone shot
):
With that completed, PDAF and Live View (CDAF) should match.
[Actually, this is a nice, quick and easy way to carry out AF fine adjustment for any camera with Live View, as it negates the need to actually take photos and therefore doesn't increase shutter count ]
Finally, I went outside to take some test shots. It's very dull out today, and I was shooting handheld, braced against the wall of my home. The first shot was taken in Live View, and this is a 100% crop:
For the second shot, I switched to PDAF, focused on a distant object (to ensure the camera and lens would need to refocus), then cycled AF and took the shot. Again, this is a 100% crop:
These shots are
both in acceptable focus, considering the screw-drive AF mechanism. Actually, the PDAF shot is ever-so-slightly more accurately focused here, but a tiny difference like this can be attributed to the screw-drive mechanism or even the smallest back or forth movement of the photographer. This was quite a challenging test, since the padlock was less than a meter from the camera, and the 50mm lens was set to f/1.4, so depth of field was very shallow indeed.
Through numerous iterations of PDAF and Live View shots, focus accuracy was within the variance shown above, with Live View sometimes being that bit more precise and on other occasions, PDAF being the winner (just). This is in line with reasonable expectations; after all, AF is good... but
not perfect
My calibrating process is a bit different than this, beginning by the target, where i usually use AF adjusting targets, but i make sure that PDAF and CDAF go to the same place, which they do.
Considering what you asked me, i'm kinda in the process of switching, already sold some lenses and my K-70, but still;
Pentax K-5 and Pentax K-70 couldn't realiable focus a face or even a whole static car while on a tripod with either a Tamron 90mm Macro or a Tamron 17-50mm, both 2.8.
Pentax K-5 and Pentax K-m couldn't reliably focus ANYTHING with either a Pentax 35mm or the Tamron 17-50mm
CDAF works wonders on those cases, but when shooting moving subjects like cars going +200Km/h, CDAF isn't an option in the K-5, specially for lenses with a long throw like the 90mm.
As far as i'm concerned, the K-m doesn't have AF adjustment, so i can't do nothing there
On the K-5, focus adjustments will work in front of the thingy used for calibration, out of there, randomness applies.
For moving subjects i use AF-C, for static subjects i use AF-S. Both modes can and will lock focus onto something that isn't in focus and won't snap into focus if i don't defocus the image first,.
Always use center point. Might be bad, but i got used to shoot like this because 99% of the time my subject is in the center so it wouldn't make much of a difference, except if the center point is somehow damaged (?)