[gilbie2000]

I followed Fixing Front and Back Focus - The Remedy - In-Depth Articles
I tried both 28-105 at f/5.6 as well as Sigma 105mm macro f/2.8 at f/2.8. Results are attached - BOTH weren't focused right - on my test target it was at ~2.5 closer for both lenses. (I used RAW+JPG; jpg shown.)
I went to C4, 25 AF fine adjustment, tried either apply all or apply one, but changing them didn't change the focus at all. I expect a -ve value, but I tried various -ve and +ve values, yet the focus did not change.
I suspect it must be a simple mistake I made to not able to adjust. Can you help? Could it be my settings for phase or contrast focus or other AF controls that made the fine AF adjust not work?

Attached Images

View Picture EXIF

File Date

 

View Picture EXIF

File Date

 

[lithedreamer]

I don't know enough to actually fix your problem, but I did want to point out, it looks like the lenses are front focusing, so I'd use a 'apply one' negative AF adjustment to compensate, I believe. Apply All is for if all of your lenses front focus. Good luck!

[gilbie2000]

Oh, I did "save" - and I went back to double check that the new adjust value shows. I used a tripod. And I changed the focal length manually first so that it will have to AF...

Well, I switched back to "apply one" for the Sigma, I can "sort of" AF adjust correctly using "-1".

By that I mean - if I half-press first to focus, it now focus correctly. However, if I half-press AGAIN, it focus back to the wrong focus. And if I keep trying, whether with the half-press or with back AF button, it goes to the wrong focus. In other words, I have only ONE chance where it focuses right. That probably explains why I initially didn't get it right.

So I still have a problem. Although I know now that -1 will get me to the right focus - if only something else in the camera keep focusing to the wrong place... Is it the phase detect AF that I have set up that messes it up?

Please help. Thanks!

[lithedreamer]

Oh, I did "save" - and I went back to double check that the new adjust value shows. I used a tripod. And I changed the focal length manually first so that it will have to AF...

Well, I switched back to "apply one" for the Sigma, I can "sort of" AF adjust correctly using "-1".

By that I mean - if I half-press first to focus, it now focus correctly. However, if I half-press AGAIN, it focus back to the wrong focus. And if I keep trying, whether with the half-press or with back AF button, it goes to the wrong focus. In other words, I have only ONE chance where it focuses right. That probably explains why I initially didn't get it right.

So I still have a problem. Although I know now that -1 will get me to the right focus - if only something else in the camera keep focusing to the wrong place... Is it the phase detect AF that I have set up that messes it up?

Please help. Thanks!

Good work! Try autofocusing in Live View and see if it's still inaccurate.

[caliscouser]

You can only do lens-specific adjustments for Pentax lenses I think, not-third party like Sigma.

**edit: I could be wrong on this now that I think about it. Please feel free to correct me on that.

[gilbie2000]

thanks lithedreamer! -1 seems to be the right adjust - your guess of a -ve adjust is good.

I also tried LV: same thing MOST of the time. It seems like contrast detection is what messes it up - I can see that it was actually focusing on the front "3" line.
Occasionally it would focus on the back 3, and still sometimes it'd focus correctly under LV. Maybe I am not lighting the target well... (However, this kind of light is common - that's what I normally use for working.)

Can someone answer a related question? Can K-1 remember the Sigma as one of the 20 AF fine adjust settings?

[lithedreamer]

thanks lithedreamer! -1 seems to be the right adjust - your guess of a -ve adjust is good.

I also tried LV: same thing MOST of the time. It seems like contrast detection is what messes it up - I can see that it was actually focusing on the front "3" line.
Occasionally it would focus on the back 3, and still sometimes it'd focus correctly under LV. Maybe I am not lighting the target well... (However, this kind of light is common - that's what I normally use for working.)

Can someone answer a related question? Can K-1 remember the Sigma as one of the 20 AF fine adjust settings?

Happy to help! I don't think third-party lenses are supported. You could do a camera wide (Apply All) adjustment for your Sigma, then just do Apply One adjustments for the rest of your lenses, if you think it's worth the effort.

I've been fortunate enough that it seems my K-1 just autofocuses spot on with all my lenses, so I haven't messed with this too much. According to the theory, Live View (contrast detection) should be more accurate (yet slower) than Viewfinder AF (phase detect). But...YMMV.

The next question to ask is, are you selecting the appropriate AF point for the target? If you're not using one of the select AF point modes (or spot AF), then you're liable to end up with whatever the camera thinks the subject should be, which...isn't helpful here. Apologies if you've already checked all that.

[UncleVanya]

Live view isn't adjustable the af adjustments only apply to PDAF.

Do you have your focus point fixed to the single center point?

I'm not certain that the target you are using is sufficient. The best targets are a bit more complex and less ambiguous.

[gilbie2000]

I tried "apply all" - it didn't do anything, as before. So I tried "apply one" again, for the Sigma. Again same as before, focused wrong.

This time I look at the focus area indicated on the viewfinder. It was mostly on the front 3, sometimes on the back 3, and sometimes on the right 0, just like LV. But regardless of what it shows, the resulting focus was nearly always wrong.

So I wasn't able to replicate my prior "first half-press works" scenario anymore...

[UncleVanya]

Did you your focus point manually to center point? That's the first requirement to ensure where it is trying to focus.

[gilbie2000]

UncleVanya, good call, thanks! The random focus seems to indicate it wasn't able to focus point - I changed to single focus instead of the default AF33.

I changed the target - THAT was the problem as you said,. I used a different target, and then I used the same target with a modified contrast (that particular target requires user to adjust). BOTH new targets allow the K-1 to focus fairly much on target without AF adjustment. I then try using +7, with the results attached.

---------- Post added 12-10-16 at 12:14 AM ----------

Just for comparison - I understand LV doesn't use AF fine adjustment because it was contrast detection - I tried again LV focus with the two targets which seems to work with phase detect...
The bad news: just like before, it focuses to different points, and sometimes didn't focus at all - whether I use point focus or AF33.

This is a good exercise for my new K-1. I think I can realize why before I bought the camera, the most common complaint I read was about AF. It seems very easy to get confused and thus give a wrong focus. I'll need to be careful to select the right focus pattern and double check if it gets confused or not. And maybe use manual focus more often...

Unfortunately, I can already see, the 28-105 doesn't resolve as well as the Sigma 105 at 105mm, which makes it difficult to manually focus as well.

[UncleVanya]

So was the PDAF adjustment successful with the single point af and better target?

In my experience the LV focus points seem larger and less precise at pinpoint focusing than I would like at times. I use a k-3 not a K-1 so there could be differences there also. Generally I get good results with focusing with either method but my impression is that the PDAF center point is smaller than the CDAF center point. This can be a minor annoyance but when faced with this I use magnified live view or phase detection to get proper focus.

(And I could be mistaken, it may just all be phantom differences in my head. )

[gilbie2000]

UncleVanya, thanks for asking. Since this K-1 is new for me, it took me a night to ponder and think through... Thanks to lithedreamer too.

1) Prior to center-point AF and better target, the K-1's default AF33 was simply "randomly" picking up a target to focus on. By that I mean it appears it tried all 33 focus points and then figure out which one(s) to use - and most of the time it picked wrong. (lithedreamer was right in calling to such attention also.) That was why I got all the strange focusing errors at the beginning. It looks like the AF33 algorithm was the same for either viewfinder or LV, thus both acted strange.
2) After center-point AF and better target, my K-1 seems to be fairly well-calibrated at factory. This is a September 2016 manufacture after all, and they probably learned from their prior experience. The PDAF seems to be close enough to be acceptable for simple applications.
3) I tried to AF fine adjust PDAF to "best" and I ran into all sorts of trouble - it turns out I have to shift and tilt and move exactly right on the tripod, or else the calibrations on the target on both sides would not be the same - I'd get one side exactly focused and another side front-focused for half to one scale. Once I get that, the AF fine adjust seems to be adjusting very little - using +1 or +10 makes slight difference. I ended up using +7, but am not absolutely sure if it is the best, since it was already close to the 0 scale and around 1/2 or 1.
4) Since I was simply testing my camera to see if I should return it (it sounds fine so far; and btw it came directly from the factory with zero shutter clicks), I did the test without upgrading firmware. I think I'll upgrade firmware next and then do more accurate tests then. And maybe looks for better targets - the targets weren't giving me enough resolution for small adjustments that I hope to do...
5) The Sigma is the only lens I can do these slight adjustments with f/2.8 - the 28-105 with f/5.6 just don't have give any noticeable difference for the targets, plus its resolving power seems far less than the Sigma - that is noticeable and I am surprised I didn't see any mention of that in comments about the lens. (Of course it ain't fair to compare a prime with a low-cost zoom, but the resolving power difference is significant on test targets like these.) I have another 80-320 that can be used to test after firmware upgrade to 1.4, but I doubt it'd be any better since it's probably also f/5.6.
6) I got the feel that the PDAF point is smaller than CDAF too. However, I've got pretty good results with CD manual focus inside a darkroom - that was my first test - contrast detect would work maybe 50% of the time, which is not bad considering it is a photographic darkroom. I did not enable AF because AF assist light would mess up real printing in the darkroom.

The moral of this exercise - I learned to be careful watching the "proposed" AF points that the camera give me - if it tells me the wrong focus point, I shouldn't be pressing the shutter! I think a lot of the bad comments about AF may actually be due to lack of watching for focus point. Of course it is painful for anyone accustomed to, or who wants, all automatic "AF." I suspect Pentax has a loooong way to go in putting more intelligence into AF33 and face detection and other detection algorithms. The good news is maybe we can expect improvements in future firmware updates.

For anyone watching this thread - caliscouser - it doesn't seemed to be mentioned clearly in other threads or documentation - but since I am an engineer and a firmware developer, I think there are 20 memories that we can use for AF fine adjust - the camera detects the lens type and put it in the memory - automatically associated with the lens type. So there is ONE memory slot for ALL non-Pentax lenses. Since I plan to use only the Sigma 105mm (which I borrowed), it would work for me I think.

---------- Post added 12-10-16 at 12:13 PM ----------

On 2nd thought, saying the 20 AF fine adjust slots are for Pentax lenses is misleading. It is for any lenses used that have electrical contacts that the camera can read. So if it is old Pentax lens without the contacts, or new/old Pentax lenses connected via adapters, macro rings, bellows, etc. - they still would be classified in the final "non-readable" slot just like non-Pentax lenses.

---------- Post added 12-10-16 at 12:15 PM ----------

And if Ricoh/Pentax reads this, I suggest to update firmware to provide auto lens detect for AF fine adjust just like currently, plus a manual mode where user can specify the memory.

[lithedreamer]

Thanks for keeping us posted, gilbie. The feedback for Ricoh is good, too. They actually have an email address for suggestions now, it's TSSCRicohimaging@ricoh-usa.com and they've advised using the Subject "Suggestions".

[clackers]

The moral of this exercise - I learned to be careful watching the "proposed" AF points that the camera give me - if it tells me the wrong focus point, I shouldn't be pressing the shutter!

User error is very common, Gilbie, you can take solace in that!

Don't automate unless you understand what the consequences will be.

If you go to 33 point autofocus (similar to Canon or Nikon or whatever) the camera will usually end up choosing whatever point contains something closest to the camera.

That's the nose rather than the eye, the microphone instead of the singer, the bars instead of the zoo animal.

You bought a DSLR to take charge of your images, so put it into single point, watch the point acquire focus and listen for the beep before snapping.