[tibbitts]

I'm not sure if this is a lens or body issue, so...

When I bought my 60-250 new, it performed pretty poorly at all focal lengths on my K5 and K5II, until I made a 10 (plus, minus, don't remember) focus adjustment. It needed every bit of that 10 setting - anything less was worse. Eventually I sent it for warranty service and miraculously/surprisingly it came back and appeared to be as close to 0 as I could measure. Pictures after it came back, but before I removed the 10 adjustment, were obviously off.

Then yesterday I was experimenting and it seemed like the lens was focusing just very slightly off, at least at 250mm, like maybe off by a couple of points, nowhere near what I was seeing when it was new. So I got out my 8.5x11" test target and tripod, filled the frame at 250mm, and experimented at f4 with my K5 and K5II. Nothing I did mattered with either body: +10, -10, 0. Results were indistinguishable at every setting. I can't say they weren't as sharp as possible, since they were all the same - I would say they were reasonably sharp at least, not like the lens when it was new and unadjusted. I can't explain it. Just for a sanity check, I attached my 17-50 Tamron at f2.8, and sure enough the focus adjustments were obvious. Unfortunately it's always needed different adjustments at different focal lengths, even after servicing twice, but that's another story. The point is that focus adjustments mattered with the Tamron, a lot, just like they did before on the 60-250.

I'm using the "one only" adjustment mode, like I did for the lens when it needed the 10 adjustment. All the tests were center-point only, not that that should matter on a flat test target.

I'll repeat my tests today, but in the meantime, any ideas?

[pschlute]

I can't say they weren't as sharp as possible, since they were all the same - I would say they were reasonably sharp at least, not like the lens when it was new and unadjusted.

First thing to say is that the AF/FA adjustment is not about "sharpness", but is about front or back focus. If a lens cannot give "sharp" images then AF/FA wont help that.

I'm using the "one only" adjustment mode,

If you are using the Adjust One menu to adjust the lens, make sure your AF/FA menu is in Adjust One when you exit the menu. Otherwise it may not be applied.

The protocol is:

AF/FA Menu OFF................. No adjustment will be made
AF/FA Menu Apply ALL..........Apply all setting will be used for ALL lenses (including ones that have been individually set in "Apply One" menu)
AF/FA Menu Apply One.........Apply One setting will be used for lenses that have been individually calibrated. Mounting a lens with no calibration and the camera will uses the Apply All setting

All the tests were center-point only, not that that should matter on a flat test target.

The target must be at an angle, otherwise how can you see if there is front or back focus ?

[tibbitts]

First thing to say is that the AF/FA adjustment is not about "sharpness", but is about front or back focus. If a lens cannot give "sharp" images then AF/FA wont help that.



If you are using the Adjust One menu to adjust the lens, make sure your AF/FA menu is in Adjust One when you exit the menu. Otherwise it may not be applied.

The protocol is:

AF/FA Menu OFF................. No adjustment will be made
AF/FA Menu Apply ALL..........Apply all setting will be used for ALL lenses
AF/FA Menu Apply One.........Apply One setting will be used for lenses that have been individually calibrated. Mounting a lens with no calibration and the camera will uses the Apply All setting



The target must be at an angle, otherwise how can you see if there is front or back focus ?

The downside of a flat target parallel to the sensor is that you need more tests/shutter-cycles to determine front or back focus. Once advantage is that if have focus field curvature then you're getting a realistic impression of what your actual images will look like.

That's a good point about not saving, although I have checked for the "Saved" message - or at least have tried to. I have repeated the test. Here is: 0, Live View focus, and +10; I can't tell the difference:

[Crooski]

As PSCHLUTE wrote target should be in an angle
for example 45 degrees then you can see if the lens focuses in front of or behind the target

[tibbitts]

I may have resolved my initial question: this test was more controlled than my previous test or my 17-50mm tests yesterday, and I was not forcing the camera to refocus between attempts. I was literally not doing anything except changing (and yes, saving) the focus settings. I retried by sticking my hand in front of the lens after every setting change to force it to refocus completely between shots, I'm now getting different results for different settings, as expected. I'm not sure exactly why that's necessary, but it seems to make a consistent difference for me.

[pschlute]

and I was not forcing the camera to refocus between attempts

Yep. Forcing the camera to refocus , ideally from the infinity setting and from the closest focus setting each time is an absolute prerequisite of doing a AF/FA adjustment.

10 tests each time. 5 from infinity and 5 from closest focus position. Then go with the majority vote. PDAF is just not that accurate that a single test will work.

[tibbitts]

Yep. Forcing the camera to refocus , ideally from the infinity setting and from the closest focus setting each time is an absolute prerequisite of doing a AF/FA adjustment.

10 tests each time. 5 from infinity and 5 from closest focus position. Then go with the majority vote. PDAF is just not that accurate that a single test will work.

I don't think I had appreciated how much that's true before this last group of tests. Now when taking actual photos I'll try to remember to manually shift focus and let the lens refocus while taking several shots. Indeed the deviation is fairly frequent and significant, more than I had learned from my previous tests in the past.

I was surprised to also see variation in live view (CDAF) focusing, which I had never tested for before. I'd just assumed it would, by nature, be more accurrate and consistent. Then of course I searched the forum and found various instances of CDAF accuracy complaints - it's surprising I'd never noticed them before.