[Class A]

So... kudos to Sigma & Ott+Wyss and their SSC service (allows for one free maintenance service per year).

Great to hear that there are still companies that care!

I had my Sigma 100-300/4 serviced for a slight front-focus and was told that it is an easy fix depending on the serial number of the lens. I had to pay a very small service charge because I had bought my lens used and it was not under warranty anymore, but now I'm a very happy user as well.

[jeallen01]

Sigma Imaging UK can also be heartily recommended - they serviced my 150-500 HSM OS last year when I asked them to see if they could improve the AF speed (which is NOT impressive, and much slower than my "new" version 100-300 F4) and they spent some time on it and replaced the motor and so - BUT they did not charge me ANYTHING (not even the return courier cost!) because they could not be sure that they had managed to fix the issue! Awesome

[Edgar_in_Indy]

There are two batches of the 100-300 Sigma. At least some lenses of the older batch are known to have a "focus jump" at f6.3. It means that the lens will show severe backfocus from f4-f5.6. From f6.3 onwards the lens will focus correctly. They fixed that issue in the newer batch of lenses. You can distinguish the two batches by the look of the rubber on the focus- and zoomrings.
Old: http://www.lenstip.com/aparaty_image/2190_sig100-300_1.jpg
new: http://www.dyxum.com/images/Lenses/269/269_1.jpg

I upgraded from the old version to the new version when I had the opportunity. I mostly did it because I like how the newer version looks (more modern), but now I'm glad I did! Incidentally, both versions worked fine on my K-30. Then again, my old version had been serviced/repaired by Sigma, NY in the past, so maybe they updated something in it at that point.

Anyway, if I ever decide to sell my copy, I'll be able to tout the fact that it's the newer version. I would hate to let my Sigma go, but I have had my eye on the Olympus 40-150mm 2.8 Pro for my m43 system...

EDIT: Come to think of it, the first person I sold my old 100-300mm to complained of AF accuracy problems on his K5-IIs. After he returned it, and I confirmed it still worked fine on my K-30, I sold it to somebody with a K-3 who left enthusiastic feedback about how great the lens was.

BTW, this kind of thing is one of the reasons I'm going toward mirrorless. Lens calibration issues are a non-issue with mirrorless.

[Class A]

Lens calibration issues are a non-issue with mirrorless.

Only if focusing happens stopped down, otherwise focus shift would still be a problem, and even then you may have typical PDAF problems with modern mirrorless cameras that use on-sensor PDAF.

I don't have first hand experience with these new systems but it appears to me that you'd have to pay with speed to get the best accuracy and may need to focus stopped down with some lenses.

[Edgar_in_Indy]

Only if focusing happens stopped down, otherwise focus shift would still be a problem, and even then you may have typical PDAF problems with modern mirrorless cameras that use on-sensor PDAF.

I'm not sure how common focus-shift on stop-down is, but I suppose that could be a problem. But then again, that type of behavior should be consistent and predictable, so Olympus, Panasonic, Sony, etc. would surely have that as part of the lens profile, and compensate for it.

But in regards to your second point, having the PDAF points on the actual sensor should eliminate the other issues that arise from the sensor and the PDAF sensors being out of perfect calibration. When the PDAF points are on the actual sensor, then that problem is solved.

From my own experience with a few different mirrorless cameras, I have found the AF to be much more predictable than any of my DSLR's.

[Class A]

But in regards to your second point, having the PDAF points on the actual sensor should eliminate the other issues that arise from the sensor and the PDAF sensors being out of perfect calibration. When the PDAF points are on the actual sensor, then that problem is solved.

A difference in optical path length between sensor and regular PDAF sensor is just one (trivial) source for PDAF issues.

If it were the only challenge then all PDAF issues could be addressed by a single AF adjustment value that would be the same for all lenses, focus distances, apertures, etc.

[Edgar_in_Indy]

A difference in optical path length between sensor and regular PDAF sensor is just one (trivial) source for PDAF issues.

If it were the only challenge then all PDAF issues could be addressed by a single AF adjustment value that would be the same for all lenses, focus distances, apertures, etc.

But the variations due to aperture, focal length, and focal distances are all covered in my first point about lens profiles. We know that the manufacturers include profiles and behind-the-scenes corrections in the firmware these days. In some cases they even bake optical corrections into RAW files. So it's likely that the other attributes are included in the lens profiles as well.

And this assumes that a lot of lenses even need these kinds of corrections. Like I said, I don't know how prevalent focus-shift due to stopping down is, but I've not seen where it's an issue for mirrorless.

[Class A]

But the variations due to aperture, focal length, and focal distances are all covered in my first point about lens profiles.

First, as the variations occur amongst copies, as opposed to between models, the profiles would have to be copy specific. Some of the variations are actually captured in lens memories, but, second only some aspects of lens variations are captured.

I'm further assuming that there are tolerances for lens mounts and sensor positioning as well.

Your experiences with mirrorless AF may be positive, but most PDAF experiences are also mostly positive and I don't know how slow/fast your experience was and how challenging the lenses were.

[Edgar_in_Indy]

First, as the variations occur amongst copies, as opposed to between models, the profiles would have to be copy specific. Some of the variations are actually captured in lens memories, but, second only some aspects of lens variations are captured.

This sounds like speculation. It seems like a lens would have to be a pretty bad copy to require its own unique profile. But sure, there's always going to be a certain number of bad copies. That's what the warranty and/or exchange policy is for.

I'm further assuming that there are tolerances for lens mounts and sensor positioning as well.

I don't think variations in mounts or sensor position would come into play. AF is performed on the sensor, so there will not be a discrepancy between the AF sensor thinking everything is sharp, and the image sensor capturing something less sharp. That's what happens with DSLR's, because focusing is not done on the sensor.

Your experiences with mirrorless AF may be positive, but most PDAF experiences are also mostly positive

You don't have to spend a lot of time on a DSLR forum to hear people talking about this lens or that lens back-focusing, front-focusing, or focusing inconsistently. These kinds of complaints are a staple. There's a reason that DSLR makers have begun including AF adjustment in their bodies. There's a reason that people have played the "lens lottery", and have returned multiple copies before getting one that works well with their particular camera body.

As far as I know, mirrorless cameras don't even have the AF adjustment "feature", which is logical since they aren't subject to AF sensor / image sensor discrepancies.

I don't know how slow/fast your experience was and how challenging the lenses were.

I'm currently using the Olympus 75mm 1.8 on my Panasonic GX-7, and that lens has a very shallow depth of field, so any problems would be very evident. I also have the 20mm 1.7. In the past, I've owned the Lumix LX-100 which has an m43 sensor and a fast zoom lens, and I also owned the Sony RX-1, which has a FF sensor with a 35mm f2 lens.

[Class A]

This sounds like speculation.

Maybe it does, but it isn't.

If your viewpoint were correct, one AF adjustment value per camera would be sufficient. Your model cannot explain the various issues that appear in practice, whereas mine can.

N.B., warranties are for definite defects. Some AF inconsistencies may or may not be fixed under warranty, depending on whom you are dealing with. Many of us know the service comment "The lens performs according to factory specifications" even when it has some clear optical defects. Some tolerances even have to be accepted as reasonable.

[Edgar_in_Indy]

Maybe it does, but it isn't.

If your viewpoint were correct, one AF adjustment value per camera would be sufficient. Your model cannot explain the various issues that appear in practice, whereas mine can.

What is the "one AF adjustment value per camera" you're talking about? Are you talking about DSLR's, or mirrorless? Are you saying that since using the micro-adjustment in a DSLR does not always cure all AF problems in a DSLR, then mirrorless cameras also must suffer AF calibration problems?

Focus-shift differences caused by varying focal lengths (in a zoom) or by subject focus distance would not effect a mirrorless camera, since the differences would show up on the image sensor, where AF is performed. The only differences that would not show up on the sensor during AF would be focus shift caused by stopping-down, since AF is usually performed with the aperture wide-open.

I don't know that such a focus-shift is prevalent (that's the speculation part I'm referring to), but even if it were, it should be predictable for a given lens design, and would be part of the AF profile for that lens. And even if individual copies had minor variations in the degree of focus-shift during stop down (more speculation), it would probably not be evident in photos since the focal plane widens as you stop down.

I've owned several DSLR's and several mirrorless cameras, and I've not seen the same kind of unpredictable AF from mirrorless that I have from DSLRs. When everything is calibrated and working right on a DSLR, AF is outstanding. But there are simply more variables when AF is not performed on the actual image sensor.

[Class A]

What is the "one AF adjustment value per camera" you're talking about? Are you talking about DSLR's, or mirrorless?

I'm talking about DLSRs.

Are you saying that since using the micro-adjustment in a DSLR does not always cure all AF problems in a DSLR, then mirrorless cameras also must suffer AF calibration problems?

Yes, if the mirrorless camera uses an "on-sensor PDAF" AF approach. The latter makes sense, as it helps speed up focusing.

If a mirrorless camera only uses contrast detection AF (CDAF) (or adds CDAF in the final stages of a PDAF-supported AF approach) then it won't suffer from the typical PDAF issues (that cannot be addressed by using just one camera-specific AF adjustment value). However, in this case, the mirrorless camera won't have fast AF (until manufacturers figure out better CDAF algorithms). The fact that PDAF supporting sensors have been developed seems testimony to me that using CDAF on its own is still a tough proposition.

Focus-shift differences caused by varying focal lengths (in a zoom) or by subject focus distance would not effect a mirrorless camera,

They will, if the mirrorless camera uses a sensor with built-in PDAF sensels for focusing (and does not do further, time-consuming, CDAF steps in order to fine-tune focusing).

[Edgar_in_Indy]

I'm talking about DLSRs.


Yes, if the mirrorless camera uses an "on-sensor PDAF" AF approach. The latter makes sense, as it helps speed up focusing.

If a mirrorless camera only uses contrast detection AF (CDAF) (or adds CDAF in the final stages of a PDAF-supported AF approach) then it won't suffer from the typical PDAF issues (that cannot be addressed by using just one camera-specific AF adjustment value). However, in this case, the mirrorless camera won't have fast AF (until manufacturers figure out better CDAF algorithms). The fact that PDAF supporting sensors have been developed seems testimony to me that using CDAF on its own is still a tough proposition.


They will, if the mirrorless camera uses a sensor with built-in PDAF sensels for focusing (and does not do further, time-consuming, CDAF steps in order to fine-tune focusing).

Well, most mirrorless cameras still use only CDAF, which is where I'm coming from. I've never experienced a hybrid PD/CDAF yet, so I can't speak to that first hand. But my impression is that in many cases the PDAF is used to get 90 or 95% of the way there, and then CDAF is used at the very end to verify focus.

But I think there are some instances where only the PDAF sensors are used (in the highest speed burst mode with CAF, perhaps). In which case you probably do lose the extreme precision of CDAF. But even so, you still eliminate some of the traditional problems associated with having the PDAF sensors in a different physical location than the image sensor. In DSLR's, AF accuracy is dependent on the position of the PDAF sensors in relation to the image sensor, as well as the perfect alignment of the mirrors which direct a portion of the incoming light to the PDAF sensors.

[daimonwn]

Does inbuilt AF adjustment work with this lens?
I have a front focus on both K-5 / K-1 but inbuilt camera AF adjustment hasn't any effects unfortunately.
Can someone check this trouble with their own lens?

[Edgar_in_Indy]

Does inbuilt AF adjustment work with this lens?
I have a front focus on both K-5 / K-1 but inbuilt camera AF adjustment hasn't any effects unfortunately.
Can someone check this trouble with their own lens?

So you're saying that with AF adjustment set to either extreme, the lens performs the same?