[StarTroop]

Thanks for the good work!

It seems so obvious now to use what is effectively a narrower reference point for focus-accuracy rather than blindly trusting the broad range of accuracy the green hexagon gives you. I suppose since you'd need to always focus from infinity or MFD for every refocused shot (particularly when making use of catch-in-focus), it won't work well with autofocus lenses and generally adds more physical motion to manually focus (maybe made up for by needing fewer attempts to achieve good focus). Have you done tests to see if returning all the way to infinite or MFD is required, or is it enough to simply start from any point you know is out of focus in the correct respective direction?

As for whether you calibrate for the further or closer point, I think the closer point is more practical if you tend to take pictures all around your focus range, including infinity. Otherwise you may often find the green hexagon turning on at infinity, even if focus is slightly beyond infinity. Starting at closest point with narrower DoF means that you will effectively expand DoF as you focus out until sharpness on target is acquired, rather than the other way where DoF keeps getting smaller.

Starting from MFD is exactly how the old Honeywell AF system in the Konica C35 AF worked. It didn't even have a motor, the lens was just wound to MFD as you cocked the shutter, and when the shutter button was pressed the lens would collapse toward infinity, and when the coupled rangefinder sensors detected focus it would trip the shutter immediately, or once infinity was reached. Catch-in-focus with AFFA calibration from MFD is basically the same thing.

I wonder, though, if Nikon's green hexagon equivalent is noticeably more accurate. It can actually tell you how close to the centre, and on which side of the focus range you are. Is that patented technology that Pentax can't imitate?

[c.a.m]

Thanks for the good work!

Thanks!

I suppose since you'd need to always focus from infinity or MFD for every refocused shot (particularly when making use of catch-in-focus), it won't work well with autofocus lenses and generally adds more physical motion to manually focus

Yes, this is generally true. I've calibrated about ten Pentax AF lenses, primes and zooms, using the basic AFFA procedure. Most of the lenses ultimately exhibit sharp focus regardless of whether the focus direction is from infinity or MFD. However, I have found several lenses -- HD DA 35mm Limited, DA 20-40 and my screwdriven DA* 50-135 -- that focus better if they are pre-focused to either side of the subject distance. I am a stickler for getting as sharp a focus as I can, so I do go through the tedium of referring to my little lens calibration card to remember which direction to focus from. It works for me because 99% of my subjects are stationary.

Have you done tests to see if returning all the way to infinite or MFD is required, or is it enough to simply start from any point you know is out of focus in the correct respective direction?

Another excellent point, which I opted to omit from the article to simplify the language and flow. In fact, it's not necessary to start strictly from infinity or the MFD. Starting sufficiently from one side of the subject distance works fine.

As for whether you calibrate for the further or closer point, I think the closer point is more practical if you tend to take pictures all around your focus range, including infinity. Otherwise you may often find the green hexagon turning on at infinity, even if focus is slightly beyond infinity. Starting at closest point with narrower DoF means that you will effectively expand DoF as you focus out until sharpness on target is acquired, rather than the other way where DoF keeps getting smaller.

Good idea in general. Where this approach falters is if the lens calibration shows the peak of the "from MFD" curve to be to the far left, as in my M 50/1.4. The best AFFA setting in this case might have been slightly beyond the limit of -10. I will review the AFFA settings of my lenses to see if I should change any to a “from MFD” direction.

I wonder, though, if Nikon's green hexagon equivalent is noticeably more accurate. It can actually tell you how close to the centre, and on which side of the focus range you are. Is that patented technology that Pentax can't imitate?

I've read a number of comments and blogs concerning the accuracy and precision of the Nikon system. Some folks say that it's quite accurate, but that doesn't seem universal. I'll try to find a couple of credible references and post them here.

Thanks again for your thoughtful comments. They add to the context of the subject, for certain.

- Craig

[StarTroop]

Yes, this is generally true. I've calibrated about ten Pentax AF lenses, primes and zooms, using the basic AFFA procedure. Most of the lenses exhibit sharp focus regardless of whether the focus direction is from infinity of MFD. However, I have found several lenses -- HD DA 35mm Limited, DA 20-40 and my screwdriven DA* 50-135 -- that focus better if they are pre-focused to either side of the subject distance. I am a stickler for getting as sharp a focus as I can, so I do go through the tedium of referring to my little lens calibration card to remember which direction to focus from. It works for me because 99% of my subjects are stationary.

For now I'll probably just stick to calibrating AF lenses to the centre, as I'm assuming (hoping) that they contain special calibration info that improves the camera algorithm, but in any case I don't often have issues with AF. Certainly pre-focusing those lenses in conjunction with AFFA probably offeres the best results, but it's not very practical on lenses without quick-shift. I'll have to experiment with my DA Ltds as I collect more, along with the 100mm WR.

Good idea in general. Where this approach falters is if the lens calibration shows the peak of the "from MFD" curve to be to the far left, as in my M 50/1.4. The best AFFA setting in this case might have been slightly beyond the limit of -10.

I hadn't thought about that. It would be a shame if the method were sometimes unfeasible due to the limitation of AF adjustments.


When I calibrated my manual lenses, I used a haphazard method with my camera handheld, using CiF to take the shots, and simply checking sharpness in playback. Since all my M series lenses needed values of -10 to -8 (but other manual lenses are closer to 0)), I'd been planning on recalibrating them more accurately to make sure there's no camera issue. Now that I'm thinking about it (although it could just be pure coincidence,) I think I may have inadvertently used a method similar to yours on these lenses, since I always focused out from MFD until CiF triggered the shutter (it seems to trigger on Hex-on, no matter what side). That may explain why I'd occasionally get out of focus shots in practice, since I sometimes focus from infinity when outside. Maybe a change in habits, to always focusing from MFD, will solve my occasional focus issues.

Honestly I'm getting a little excited just thinking about recalibrating with AFFA in mind. I find the interplay of the various systems in an SLR so interesting, and this method is about the closest thing to a "camera hack" that not only works, but is actually useful.

[noelpolar]

I've been playing recently with my 50/1.2.... and noted that focusing into the subject from a further distance focus (not necessarily infinity) was giving consistent results.... but I hadn't tried adding a -10 to +10 offset to further fine tune... so thanks for this.... will have a play now. Might save having to get a S screen.

[niels hansen]

c.a.m
Thanks a lot. You have learned me why my AF often is unreliable, but your procedure is very slow, I think that using liveview vil give the same result in a faster way ????

[robgski]

Thanks for the article!

[c.a.m]

focusing into the subject from a further distance focus (not necessarily infinity) was giving consistent results.

That's a good thing to know about the particular lens. An AF fine adjustment might further refine the focus accuracy and consistency.

so thanks for this.... will have a play now. Might save having to get a S screen.

Glad that the article is of interest. I have never used a retro-fitted screen, but understand that the 'S' variety can work well, as indicated by @BruceBanner in his current thread. Calibration may be required with shims, though.

Thanks a lot. You have learned me why my AF often is unreliable

Thank you!

but your procedure is very slow, I think that using liveview vil give the same result in a faster way ?

Yes, this approach requires fairly careful focusing in the field, and also some effort to find the best AFFA setting. Certainly, Live View with its Focus Peaking and magnification is a very good technique for focusing manually. I prefer Live View when I am using a tripod.

- Craig

[c.a.m]

Thanks for the article!

Thanks!

[JensE]

I've read the article back when you published it, but revisited it as I'm looking into (AF-) calibration right now due to some unsatisfactory results recently. Unfortunately, Trying to determine the plane of focus in the 'slanted ruler' type of setup was difficult at times, especially if LoCA and residual spherical abberations are in play. I'll revisit my difficult calibrations using a similar method as you described, just for AF. Thanks for that description alone.

With respect to manual focusing, I had questions that I wanted to clarify:

1. (How) Does the light intensity impact D+ - D0, or the range until the focus indicator turns off?
2. Similarly, what about the light color?
3. What about subject contrast?

I didn't perform an intense study, but at least did some checks using distance scales.


I first used the FA31mm on my KP at 1m distance (sensor - target), and the live-view magnified focus put the distance scale right on the 1m mark. However, for this combo, D+ and D- are so close that they are less than the thickness of the '1' stem on the focal scale. My AF calibration using 'slanted ruler next to focus target' method was spot on for this lens. I did however observe that the focus indicator only turns off at nearly exactly the f/4 DoF scale marker in either direction.

Next I put on deep red filter (090) for question 2. No impact whatsoever - no shift in the on- nor off-points. Again off-points coincided with the f/4 DoF markers.


I wasn't successful addressing the contrast target question - I couldn't get any indicator on less than stellar contrast focus targets. I'll have to redo this in daylight.



Next a Revuenon 50mm f/1.7. My default +5 correction was too much for this lens. Again, like the 31mm, a lens with little room for parallax errors on the focus scale and the 50mm has a 49mm filter thread, so that I could use my 6-stop ND filter. Without filter, the on-points D+ and D- were not as crisply defined as for the FA31, but rather close, maybe 1/3 of the distance to the f/4 DoF marker apart on the scale. They were not affected by putting on the ND64 (6-stop) filter, which brought the KP AF system close to its minimum working level.

[c.a.m]

With respect to manual focusing, I had questions that I wanted to clarify: (How) Does the light intensity impact D+ - D0, or the range until the focus indicator turns off? Similarly, what about the light color? What about subject contrast?

Hi Jens,

Thanks for posting your interesting findings.

I have not conducted any tests to specifically address light intensity, light colour, or subject/target contrast in the context of manual focus calibration or the behaviour of the focus indicator.

However, throughout my focus calibration work, I have observed that the colour temperature of the light source does not affect the AFFA settings on my K-5 II or K-3 II. I have used CFL, LED, halogen, tungsten, and natural light, with colour temperatures ranging from 2700 K to 6500 K. For example, for some of my lenses, I had calibrated with two or three different light sources, arriving at the same AFFA settings. On the other hand, if I recall correctly, the autofocus of the K-5 with its early firmware was affected by tungsten incandescent lighting.

I haven't noticed that light intensity has any effect on the range of the focus indicator. My artificial lighting is normally adjusted to provide for my desired shutter speed and aperture.

I would think that the contrast of the subject target is an important factor. While targets of opportunity may provide good results, I have tended to use a high-contrast target like the one I illustrate in the article. It helps to eliminate a variable in the process.

I first used the FA31mm on my KP at 1m distance (sensor - target), and the live-view magnified focus put the distance scale right on the 1m mark. However, for this combo, D+ and D- are so close that they are less than the thickness of the '1' stem on the focal scale. My AF calibration using 'slanted ruler next to focus target' method was spot on for this lens. I did however observe that the focus indicator only turns off at nearly exactly the f/4 DoF scale marker in either direction.

I don't own a copy of the FA 31mm to compare directly with your finding. However, I ran a controlled test on my K-3 II with my calibrated SMC Pentax 30mm f/2.8, which has a similar focus scale. Using magnified Live View, the sharpest focus was at the vertical stem of the '1' meter mark, and the in-focus range was very tight. Using viewfinder PDAF and focusing from the infinity side, the indicator illuminated when the '1' was aligned with the orange diamond marker, and turned off when it aligned with the left-hand f/4 DoF mark. So, I would say the behaviour is similar to your FA 31.

Thanks again for your observations!

- Craig