[tduell]

Back in 2016, testing front/back focus of my DFA 150-450 on both my K-3 II and K-1 bodies showed that both were back focussing, and an AF fine adjustment of +7 seemed to put things right.
Since then there appears to have been a slow drift towards further back focussing on both bodies, and they are now both set at +10, but my guess is that the K-3 II really needs about +12 and the K-1 about +11.
I had assumed that front/back focussing was all due to the camera body, as the lens just responds to the demands of the camera AF system, but given that both my bodies have been exhibiting drift of a similar magnitude in the same direction, I now wonder if the lens is the cause of the problem.
My front/back focussing testing method is as follows...
(1) A vertical target with black on white pattern, with a 45 degree graduated 'ruler' attached to the side, placed between 10-15m from the camera/lens, testing at 450mm.
(2) The vertical target carefully aligned to be normal to the lens axis.
(3) Using spot focus, pick a focus point on the vertical target, and clearly not on the sloping ruler.
(4) Camera/lens mounted on tripod, SR off, 2 sec delay using remote shutter release.
(5) Defocus lens prior to each test shot.
(6) Repeat test shots a number of times and check for consistent results, for each AF adjustment setting tried.
I also take a few live view shots to prove that correct focussing can be obtained with the setup.
Does anyone have possible explanations for what might be going on here...apart from the obvious possibility that my testing process resulted in errors?

Cheers,
Terry

[stevebrot]

Back in 2016, testing front/back focus of my DFA 150-450 on both my K-3 II and K-1 bodies showed that both were back focussing, and an AF fine adjustment of +7 seemed to put things right.
Since then there appears to have been a slow drift towards further back focussing on both bodies, and they are now both set at +10, but my guess is that the K-3 II really needs about +12 and the K-1 about +11.
I had assumed that front/back focussing was all due to the camera body, as the lens just responds to the demands of the camera AF system, but given that both my bodies have been exhibiting drift of a similar magnitude in the same direction, I now wonder if the lens is the cause of the problem.
My front/back focussing testing method is as follows...
(1) A vertical target with black on white pattern, with a 45 degree graduated 'ruler' attached to the side, placed between 10-15m from the camera/lens, testing at 450mm.
(2) The vertical target carefully aligned to be normal to the lens axis.
(3) Using spot focus, pick a focus point on the vertical target, and clearly not on the sloping ruler.
(4) Camera/lens mounted on tripod, SR off, 2 sec delay using remote shutter release.
(5) Defocus lens prior to each test shot.
(6) Repeat test shots a number of times and check for consistent results, for each AF adjustment setting tried.
I also take a few live view shots to prove that correct focussing can be obtained with the setup.
Does anyone have possible explanations for what might be going on here...apart from the obvious possibility that my testing process resulted in errors?

Cheers,
Terry

Perhaps the lens has drifted out of alignment (either lateral or axial) for one or more groups?


Steve

[biz-engineer]

Were the ambient temperature, atmospheric pressure, humidity and light source illuminating of the focus target the same in 2016 than the were when you did the calibration again recently? Did you refocus the lens in the same direction by the same amount in 2016? When you defocus the lens did you do it manually or by having the camera focus itself on a different target? What is the reproducibility of your calibration procedure, if you do it three time at one week interval do you get the same results?

---------- Post added 26-10-19 at 06:11 ----------

P. S. I've assessed the reproducibility of AF fine tuning and concluded that the best value for all weather was the factory default. In fact, when the dispersion of AF from shot to shot is as large as equivalent to +-5 tuning, the AF fine tuning is not that effective to get every shot best focused.

AF fine adjustment drifting

Maybe there is a long term drift. In that case, it's a good idea to perform AF fine tuning calibration on a regular basis, every three months for example.

[tduell]

Perhaps the lens has drifted out of alignment (either lateral or axial) for one or more groups?


Steve

Maybe. I would guess axial alignment change may give rise to what I'm seeing.
Not sure if I could detect lateral alignment.
If it's either of these, I guess the lens would require factory or service centre attention.

Cheers,
Terry.

---------- Post added 10-25-19 at 10:35 PM ----------

Were the ambient temperature, atmospheric pressure, humidity and light source illuminating of the focus target the same in 2016 than the were when you did the calibration again recently? Did you refocus the lens in the same direction by the same amount in 2016? When you defocus the lens did you do it manually or by having the camera focus itself on a different target? What is the reproducibility of your calibration procedure, if you do it three time at one week interval do you get the same results?

---------- Post added 26-10-19 at 06:11 ----------

P. S. I've assessed the reproducibility of AF fine tuning and concluded that the best value for all weather was the factory default. In fact, when the dispersion of AF from shot to shot is as large as equivalent to +-5 tuning, the AF fine tuning is not that effective to get every shot best focused.


Maybe there is a long term drift. In that case, it's a good idea to perform AF fine tuning calibration on a regular basis, every three months for example.

Not able to answer all those questions, they suggest one would be doing the tests in a laboratory.
The questions re refocussing are interesting. I'd need to see some good data that the refocussing really made any noticeable difference under the test conditions I use, given that I reproduce results before accepting that they are valid.
If there is a long term drift, it has been in the one direction thus far, and as it appears that I've reached the adjustment limit I'm in a situation that I'm unlikely to be able to correct.
I've been researching this subject a bit more, and now pondering if the dot-tune method may be worth trying, at least to see if that produces the same results, and if so then it's probably time to talk to the service centre.

Cheers,
Terry

[biz-engineer]

The questions re refocussing are interesting.I'd need to see some good data that the refocussing really made any noticeable difference under the test conditions

If that's of interest to you, I can describe my Af fine tuning experiement.

I've done an experiment with my K1 and D-FA 28-105 @ 105, 60 AF shoots alternating each shot with pre-focused manually (turning the focus ring) at nearest focus distance and prefocus at infinity, that's 30 shoots prefocused at nearest distance and 30 shoots prefocused at infinity. I've alternated near prefocus and infinity prefocus to eleminate the effect of gradually increasing temperature inside the camera for the duration of the experiment. Then I plotted the MTF of test chart, and I could see two populations, one population of the shoots pre-focused at infinity separate from the population of shoots prefocused at nearest distance.

Then, I've repeated the same procedure but without touching the focus ring for the pre-focus operation, using the camera AF itself and a movable target to force the camera to focus near or at infinity (the camera will stop focusing at infinity is the targe is too close, I set the movable target closer to nearest distance to have the camera AF fail and default to focus the lens at infinity). Again captured 60 shoots alternated between nearest pre-focus and infinity pre-focus (30 shoots pre-focused near, 30 shoots prefocused at infinty, all using IR remote, 3 s. delay, and pixel shift for max sharpness from the sensor). Plotted the MTF distribution again. Still observed two separe populations, each centered around the mean value of near-prefocused samples and the mean value of the infinity pre-focused samples.

Interestingly, the gap between the two populations of near-prefocus and infinity pre-focus was smaller when using camera AF to pre-focus. I've use camera AF to pre-focus to compute the best AF fine tune value. In order to find the best AF fine tune value, I've shot again 10 shoots alternatively prefocused at nearest distance and infinity, for each of AF fine tune setting from -10 to +10 (that 200 shots, 100 shots prefocused at nearest distance and 100 shots pre-focused at infinity by camera AF itself). Plotted the mean value of MTF as a function of AF fine tune value, it looked like bell curve flattened around AF fine tune value = 0. Then I plotted the MTF value dispersion vs AF fine tune value, and I found the the dispersion was greater at the outer regions of the AF tune values. So , for example, the best possible sharpness was obtained with AF fine tine value = -8 but at that setting some of the shoots were softer than the worse shoots at AF fine tune = 0.

My conclusion was that AF fine tune = -8 would give the best possible sharpess, at the expense of having less than 50% of the shoots less sharp than when using Pentax factory default. I prefered to have a higher percentage of my shoot in good enough focus as opposed to having some of the shoots a little sharpner and some of the shots worse than default. So I decided to reset the AF fine tune to value 0.

[kiwi_jono]

I used to have this issue with the K-5. After a year I would start to notice small focus errors creeping in and the AF fine tune would need tweaking a bit.
I should point out I was using a few f2 and f1.4 lenses where focus errors are much more noticeable and the probably was not major.

So far I have not seen any shift with the K-1 though and accuracy seems better with the faster lenses too.

[Mikesul]

If that's of interest to you, I can describe my Af fine tuning experiement.

I've done an experiment with my K1 and D-FA 28-105 @ 105, 60 AF shoots alternating each shot with pre-focused manually (turning the focus ring) at nearest focus distance and prefocus at infinity, that's 30 shoots prefocused at nearest distance and 30 shoots prefocused at infinity. I've alternated near prefocus and infinity prefocus to eleminate the effect of gradually increasing temperature inside the camera for the duration of the experiment. Then I plotted the MTF of test chart, and I could see two populations, one population of the shoots pre-focused at infinity separate from the population of shoots prefocused at nearest distance.

Then, I've repeated the same procedure but without touching the focus ring for the pre-focus operation, using the camera AF itself and a movable target to force the camera to focus near or at infinity (the camera will stop focusing at infinity is the targe is too close, I set the movable target closer to nearest distance to have the camera AF fail and default to focus the lens at infinity). Again captured 60 shoots alternated between nearest pre-focus and infinity pre-focus (30 shoots pre-focused near, 30 shoots prefocused at infinty, all using IR remote, 3 s. delay, and pixel shift for max sharpness from the sensor). Plotted the MTF distribution again. Still observed two separe populations, each centered around the mean value of near-prefocused samples and the mean value of the infinity pre-focused samples.

Interestingly, the gap between the two populations of near-prefocus and infinity pre-focus was smaller when using camera AF to pre-focus. I've use camera AF to pre-focus to compute the best AF fine tune value. In order to find the best AF fine tune value, I've shot again 10 shoots alternatively prefocused at nearest distance and infinity, for each of AF fine tune setting from -10 to +10 (that 200 shots, 100 shots prefocused at nearest distance and 100 shots pre-focused at infinity by camera AF itself). Plotted the mean value of MTF as a function of AF fine tune value, it looked like bell curve flattened around AF fine tune value = 0. Then I plotted the MTF value dispersion vs AF fine tune value, and I found the the dispersion was greater at the outer regions of the AF tune values. So , for example, the best possible sharpness was obtained with AF fine tine value = -8 but at that setting some of the shoots were softer than the worse shoots at AF fine tune = 0.

My conclusion was that AF fine tune = -8 would give the best possible sharpess, at the expense of having less than 50% of the shoots less sharp than when using Pentax factory default. I prefered to have a higher percentage of my shoot in good enough focus as opposed to having some of the shoots a little sharpner and some of the shots worse than default. So I decided to reset the AF fine tune to value 0.

This is really interesting. Are you saying in these posts that if the range goes beyond +/- 5 it may be better to leave it at the factory default?

[biz-engineer]

Are you saying in these posts that if the range goes beyond +/- 5 it may be better to leave it at the factory default?

That's not what I'm saying. What I'm saying is: if the correction of AF accuracy comes at the cost of a drop in precision, then the AF fine tune is not necessarily better than leaving it at the default factory setting. In my case (camera + lens), the best sharpness was best with -8 setting, but the worst sharpness was also with -8. Since I'd rather shoot for taking pictures than shoot for making statistics, I reverted to factory default, since this was the setting that gave me the best AF precision. I tried with the D-FA24-70 (f2.8), stopped down to f5.6, and the precision was much better than with the D-FA 28-105 @ 105mm f5.6. I attribute that to the f2.8 that allow the camera to focus better at f2.8 before it is stopped to f5.6 to take the picture (that how I realize how much better it is to use an f2.8 lens vs an f5.6 lens, even if both lens apertures at set to f5.6 for the DoF).

[zapp]

On my K1 the screw of the quick release plate changed the position of the AF sensor back and forth. That may not cause your problem, but it also shows that the camera is also prone to change.
Before I found that AF fine adjust was stable over 1-2 years after it was done properly in good light conditions and many shots.
Results also depend on zoom position, Focus distance, and aperture, so you need to find one correction value working for your work with a particular lens.

AF precision is important. Make sure to use good light and several AF actions to reach real top focus of the important images - as biz engineer pointed out. A setting close to 0 will be easier to correct by the system than an extreme value. Not sure how it works exactly, but zeroing camera and lens in factory is said to be the best.

[stevebrot]

On my K1 the screw of the quick release plate changed the position of the AF sensor back and forth. That may not cause your problem, but it also shows that the camera is also prone to change.

This fascinating and potentially important. Are you willing to share in detail on a new thread?


Steve

[c.a.m]

On my K1 the screw of the quick release plate changed the position of the AF sensor back and forth.

I agree with @stevebrot. It would be interesting to hear more about this phenomenon.

I assume it refers to the tripod socket on the camera's base. The AF module is certainly in the vicinity.

- Craig