[Kobie]

Is this true?

AL

Unfortunately yes, it's true.

[jon.partsch]

The point I am trying to make is that it is simply how PDAF reacts to certain common lens issues. This is not a camera QA issue. If one does not want to have to deal with front/back focus, one has to ditch PDAF. With current tech, that means going to a non-reflex viewfinder (i.e. mirrorless).

I'm pretty sure this is wrong. Can you cite examples of these "certain common lens issues"?

If the PDAF sensors are on the imaging chip, then this is no longer an issue (though other focus problems can still occur). It is absolutely a problem with optical path to two sensors that are in physically different places within the camera. The same lens does not front/back focus the same way between different bodies, and many bodies actually have a method of calibrating the physical position of the PDAF sensor array if it comes out of alignment or was out of alignment from the factory.

It already exists. The Nikon D6 has that ability for automated AF Fine adjustments. You manually focus in live view then tell the camera "this is where the focus should be" The camera then corrects the AF for you. But, it's a what... $6000 DSLR?

Yes, this is part of the point I was trying to make ‐ the issue can absolutely be eliminated for DSLRs using tech, as Nikon is closest to doing. It will always be a potential problem with legacy manual lenses, especially with manual zoom lenses, but it should be possible to fully automate the correction process in camera for modern AF lenses, even creating an adjustment curve for zoom lenses so that they are perfectly corrected at any focal length. Does Nikon do this yet?

Is this true?

AL

Unfortunately, yes, but correcting fine focus for a zoom lens in the middle of it's zoom range will solve most of the problem in most cases. Some prefer to adjust fine focus for a zoom lens nearer to one end of it's focal length range, if they tend to use that area of the zoom range more. Of course, the problem is most noticeable at wide apertures.

-Jon

[brewmaster15]

Thanks Jon and Kobie. Thats something I did not realize and good to know.
AL

[gatorguy]

It already exists. The Nikon D6 has that ability for automated AF Fine adjustments. You manually focus in live view then tell the camera "this is where the focus should be" The camera then corrects the AF for you. But, it's a what... $6000 DSLR?

While that $6K Nikon D6 allows for somewhat automated lens adjustments (tho manual is sometimes needed) even they have limits when it comes to a zoom lens.The camera can store two fine-tuning values for a zoom: one for use when all the way out and another zoomed all the way in.

---

Last edited by gatorguy; 04-09-2021 at 09:33 AM.

[stevebrot]

If the PDAF sensors are on the imaging chip, then this is no longer an issue (though other focus problems can still occur).

It is an issue, but such systems use PDAF for speed and fine focus with CDAF (hybrid AF).



Steve

[stevebrot]

Yes, this is part of the point I was trying to make ‐ the issue can absolutely be eliminated for DSLRs using tech, as Nikon is closest to doing.

Agreed...the Nikon approach make it much more convenient and I wish the K-3iii had this tech. The patent will probably expire one of these days.


Steve

[stevebrot]

But, it's a what... $6000 DSLR?

This feature is present on the D500 and D850 and may have been migrated down to the D7xxxx series bodies as well.


Steve

[Rondec]

It already exists. The Nikon D6 has that ability for automated AF Fine adjustments. You manually focus in live view then tell the camera "this is where the focus should be" The camera then corrects the AF for you. But, it's a what... $6000 DSLR?

It feels like this sort of tech should be relatively easy to implement with a firmware tweak.

(I guess the question would have more to do if Nikon has patented their particular implementation).

[jon.partsch]

While that $6K Nikon D6 allows for somewhat automated lens adjustments (tho manual is sometimes needed) even they have limits when it comes to a zoom lens.The camera can store two fine-tuning values for a zoom: one for use when all the way out and another zoomed all the way in.

Okay, yes, thank you for reminding me about the details of Nikon's implementation. My assumption is that this provides merely a linear interpolation between these endpoints. That's pure speculation, and I have not handled any of the Nikon camera with this tech. I wouldn't be surprised if this adequately corrects all but the most egregious cases of F/B focus throughout the zoom range.

An ideal system would let you define the amount of correction at each end of the zoom range (or even automate this process for you), and then apply a polynomial function (curve) to set the amount of correction at every focal length based on the known optical properties of the lens design.

If this were done, then there would be almost zero advantage to an MILC design (other than it would probably be cheaper and easier to manufacture). Some will still prefer the compactness of a MILC design, but I need any camera I handle to have a beefy grip so that I can confidently hold and control the camera with a heavy lens attached. Incidentally, though not a MILC, this was my main gripe with the KP, and I will probably never own one for that fact alone (yes, I am aware of interchangeable grips, but resent having to buy an accessory just to be able to hold the camera).

Agreed...the Nikon approach make it much more convenient and I wish the K-3iii had this tech. The patent will probably expire one of these days.


Steve

The patent must be pretty broad in scope to prevent Ricoh from developing a similar technology.

It feels like this sort of tech should be relatively easy to implement with a firmware tweak.

(I guess the question would have more to do if Nikon has patented their particular implementation).

Yes, I assume that any camera, at least any camera that has fine focus adjustment capibility, can have this implemented in a theoretical firmware update.

[stevebrot]

The patent must be pretty broad in scope to prevent Ricoh from developing a similar technology.

It is broad enough that Canon does not offer similar.


Steve

[jon.partsch]

It is broad enough that Canon does not offer similar.


Steve

That's a good point. Bumping up against competitors' patents may be a big reason why Pentax cameras have lagged behind in many areas of autofocus performance. I guess we can also be thankful that intellectual property protections defend some of the wonderful areas where Pentax cameras have trounced the competition, like ruggedness and weather sealing, SR and Astrotracer, pixel shift resolution, etc.

~ Jon

[c.a.m]

I looked into Nikon's Auto AF Fine Tune a while ago.

Based on numerous user comments and questions, it seems that the technique is not foolproof and can be prone to issues. The procedure requires attention to detail and careful setup. It can be sensitive to the type of target and lighting, camera/lens physical stability, and the initial 'best focus' setting. Repeatability is difficult to achieve sometimes. Users have experienced 'Auto AF Fine Tune Failed' messages for no apparent reason.

While I understand that newer Nikon cameras or firmware provides for two separate settings for zoom lenses, that may not be sufficient to cover the full span of focal lengths on some lenses.

My impression is that if a user is fastidious with the setup and execution, the Auto AF Fine Tune can work well. However, it doesn't always yield an optimal AF Fine Tune setting and zoom lenses can still be problematic.

- Craig

[stevebrot]

The procedure requires attention to detail and careful setup. It can be sensitive to the type of target and lighting, camera/lens physical stability, and the initial 'best focus' setting. Repeatability is difficult to achieve sometimes.

Sounds a little bit like doing it manually!


Steve

(...has fielded a few too many AF fine adjust help requests...)

[c.a.m]

Sounds a little bit like doing it manually!

Sort of.

I haven't used the Nikon system in practice. From what I understand, though, the auto procedure can save time and tedium. For example, it's not necessary to make multiple runs manually, each one stepping through AF Fine Adjust settings like we do with Pentax cameras. However, some users recommend repeating the final auto step ten times or so for each lens, and averaging the resulting fine tune settings.

Anyways, I'm comfortable going through the Pentax AF Fine Adjustment procedure, and have refined my methodology to the point that I can calibrate a lens reliably and confidently in about 15 minutes. I understand that folks who might not be at ease with the manual approach would appreciate the automation -- if it works well.


- Craig

[stevebrot]

For example, it's not necessary to make multiple runs, each one stepping through AF Fine Adjust settings.

My understanding is that the camera does multiple exposures bracketing around the known good AF set in live view.


Steve