[rawr]

A work-around solution for the 35ART is to calibrate it using the Sigma lens dock.

That's not always effective. Sigma's lens firmware with the new Art series lenses for Pentax (and other mounts too, apparently) is still not quite right. Lens dock adjustments don't really help much, since the problem is deeper. Sigma really do need to go through their lens firmware with a fine tooth comb, particularly the AF code sections. Hence some of these Sigma lens AF issues are not really a Pentax problem per se.

[Class A]

Lens dock adjustments don't really help much, since the problem is deeper.

There are hundreds of thousands Sigma lens users who do not have AF issues.
Sigma produce so many lenses for various mounts they can even afford to drop a mount (K-mount) from their programme.

I'm not denying that there are Sigma lenses with AF issues, but there are genuine Pentax lenses with AF issues as well. I'm not saying that Sigma do not produce more lenses with AF issues (for whatever reasons) but the idea that there is a deep problem with every single Sigma lens has absolutely no merit at all

Sigma really do need to go through their lens firmware with a fine tooth comb, particularly the AF code sections.

According to everything I know about AF systems -- I've read white papers, patents, etc. -- there is no "AF code" in a K-mount lens' firmware. HSM lenses may contain a controller that enacts the requests made by a camera, but the responsibility to obtain AF focus is solely the camera's and the latter also determines in what way that objective is achieved.

Please point us to a reliable source, if you know anything to the contrary.

Again, I'm not saying that there are no Sigma lenses that require calibration. It is even possible that Sigma have released batches of lenses that feature incorrect lens parameters; I have no knowledge of that but it is a possibility and could explain the 18-35/1.8 phenomenon (a lens that works fine for many but caused issues for some).

In any event, it seems rather safe to exclude the possibility that Sigma has to attend to some "AF code" in their lenses.

Back on topic:
It seems that K-1 allows downgrading the firmware version as well as upgrading it. Member lennyl has made a respective remark in the comments section of the V1.40 firmware release announcement. That would be fantastic news, AFAIC, if it could be confirmed.

Did anyone other than lennyl observe a shift in AF adjustment values after the update to V1.40? It is not the first time someone reports that AF adjustment values have to be changed after a firmware update. Often, I think, such observations can be chalked up to changing conditions (e.g., changing the distance to the focus chart) when performing the AF adjustments, but lennyl seemed to have been rather systematic about this and even reported more AF hesitation ("hunting") with V1.40. I'm still sceptical that V1.40 brought any changes to the AF system, but it is of course possible.

[Paul the Sunman]

Did anyone other than lennyl observe a shift in AF adjustment values after the update to V1.40?

Not me. I was out shooting today with the K-1 using DFA 70-200, DFA 28-105, DA 15, and Sigma 35 Art. Focus was spot-on for them all. Actually, none of them need or needed adjustment.

[Class A]

Not me.

Thanks, Paul!

[Class A]

I've just tried V1.40 and while the night vision LCD display option looks great and I'd love to keep it, I just tried some quick AF tests with my FA 43/1.9 and they seem to confirm what lennyl reported. My tests were quick and dirty but every time the V1.30 AF result with the FA 43/1.9 wide open was visibly better than that of V1.40. I tried two focusing distances and the results were consistent.

I did not try to use AF adjustments to see whether I could bring the V1.40 results to the same level; I guess this should be possible. I'd have to do a proper test to see whether

1. there is actually a real difference between V1.30 and V1.40 AF performance. My tests were really quick and dirty and I feel reluctant to even report my impressions.
2. whether explicitly setting and resetting the AF adjustment values makes a difference (I had mine set to "OFF" and that had been fine for V1.30).
3. if there is a consistent change in AF adjustment values across lenses.

It would be great if others could do some AF checking (regarding FF and BF) with V1.40 and compare the results to V1.30.

There is definitely one very good bit of news: A downgrade to V1.30 was indeed possible without any issue. If Ricoh will continue to allow this, it will make it much easier to just try a new firmware version in the knowledge that one can always go back to a known old status, should one have any reason to believe that a new version brought any disadvantage. Thank you, Ricoh!

[rawr]

the responsibility to obtain AF focus is solely the camera's and the latter also determines in what way that objective is achieved.
Please point us to a reliable source, if you know anything to the contrary.

Ummm, one doesn't have to look any further than, for example, the Sigma lens firmware updates page. A quick scan of the release notes for various Sigma lens firmwares will, for example, reveal:

This firmware update standardizes the data processing to improve AF accuracy and resets the focus adjustment values.
It has refined the AF algorithm to reduce the operating noise.
It has improved the smoothness of AF as it starts up and stops.
It has improved the Auto Focus speed has been improved on "Standard" and "Speed-priority" modes.
It has improved the focus accuracy when it is mounted on a Canon EOS-1D Mark III or Canon EOS-1D Mark IV.
It has improved the AF accuracy when the lens is attached with the SIGMA MOUNT CONVERTER MC-11 EF-E.
It has improved the AF accuracy when the lens is used for shooting movies as well as Live View on EOS DIGITAL SLR cameras
It has improved the AF accuracy when attached to the TELE CONVERTER TC-1401.
This firmware update standardizes the data processing to improve AF accuracy and resets the focus adjustment values.
It has improved the precision in its focusing performance when the AF frames in peripheral areas in the viewfinder are selected.
It has improved the focusing performance of Continuous servo (AF-C) mode.

I don't know why you refuse to accept this. It's there in black-and-white.

[biz-engineer]

Originally posted by Class A Quote the responsibility to obtain AF focus is solely the camera's and the latter also determines in what way that objective is achieved. Please point us to a reliable source, if you know anything to the contrary. Ummm, one doesn't have to look any further than, for example, the Sigma lens firmware updates page. A quick scan of the release notes for various Sigma lens firmwares will, for example, reveal: QuoteQuote: This firmware update standardizes the data processing to improve AF accuracy and resets the focus adjustment values. It has refined the AF algorithm to reduce the operating noise. It has improved the smoothness of AF as it starts up and stops. It has improved the Auto Focus speed has been improved on "Standard" and "Speed-priority" modes. It has improved the focus accuracy when it is mounted on a Canon EOS-1D Mark III or Canon EOS-1D Mark IV. It has improved the AF accuracy when the lens is attached with the SIGMA MOUNT CONVERTER MC-11 EF-E. It has improved the AF accuracy when the lens is used for shooting movies as well as Live View on EOS DIGITAL SLR cameras It has improved the AF accuracy when attached to the TELE CONVERTER TC-1401. This firmware update standardizes the data processing to improve AF accuracy and resets the focus adjustment values. It has improved the precision in its focusing performance when the AF frames in peripheral areas in the viewfinder are selected. It has improved the focusing performance of Continuous servo (AF-C) mode. I don't know why you refuse to accept this. It's there in black-and-white.

What AF refers to in lens terms is the actuation part of the AF servo. No lens focuses on its own, the most part of AF control is in the camera body. However, I believe the lens firmware is simply the controlled response of actuation, i.e the tweaking of dynamic characteristic of the mechanical motion of the AF lens element(s).

You are both right, but what differs is your understanding of the term "AF".

[Winder]

Did anyone other than lennyl observe a shift in AF adjustment values after the update to V1.40?

Yes. My lenses started back focusing. I had to recalibrate. I have a shoot this afternoon. After the shoot I will update to 1.41 and check AF again.

[rawr]

No lens focuses on its own, the most part of AF control is in the camera body.

Of course. But what you have with modern digital lenses and cameras are two computers trying to work with each other. Not always co-operatively.

[Class A]

I don't know why you refuse to accept this. It's there in black-and-white.

I was talking about the "firmware" in K-mount lenses. Your quote makes references to Canon camera models.

It seems that the lens model you quoted from is the Sigma 17-70mm F2.8-4 DC MACRO OS HSM and while that lens is indeed available in K-mount, it is entirely possible that the firmware release notes only reference an AF algorithm for Canon versions of the lens.

For the lens to have control over AF, it would have to communicate with the camera about the results of the PDAF measurements that the camera performs. It seems unlikely that Pentax took this approach for screw mount lenses and I'd be surprised if they changed the approach with SDM lenses. As a matter of fact, a Pentax patent describes an algorithm that runs in the camera which checks whether a lens is capable of performing SDM. If the latter is the case, then this simply changes how focus changes are enacted (via the camera's motor or the lens' motor), but in any event the feedback loop is still run by the camera.

While I don't have proof that K-mount lenses never take control of AF, it seems highly unlikely to me. For sure, we don't have proof that a K-mount Sigma lens controls AF, just because a Sigma release note talks about AF in general (making frequent references to Canon while doing so).

EDIT: It is also entirely possible that even EOS lenses do not contain an "AF algorithm" other than the logic required to actuate the in-lens motor. The main feedback loop probably still resides in the camera.

[Class A]

But what you have with modern digital lenses and cameras are two computers trying to work with each other. Not always co-operatively.

I hope you will admit that this is pure conjecture on your part.

You do not have any knowledge about what the responsibilities of the lens firmware are, do you?

Again, Sigma would not be the highly successful and highly praised lens manufacturer they are, if their lenses contained code that "fought" with the camera.

It is much more likely that there are clear responsibilities and that the overall behaviour is cooperative than anything else. This does not rule out that some parameters (regarding AF step size, for instance) may be wrong, that initial AF calibration data regarding zoom setting and focus setting may be wrong for a certain mount, etc. If any of the latter is wrong, the camera will have a hard time or will even be incapable of producing correct focus. However, you have no proof of code in the lens firmware that is ultimately responsible for focusing.

[Class A]

Yes. My lenses started back focusing. I had to recalibrate.

Thanks a lot for reporting.

That's the third anecdotal data point that suggests that either V1.40 is a step backwards regarding AF, or that at least some measures must be taken (explicit AF adjustment resetting or new AF adjustment calibration).

After the shoot I will update to 1.41 and check AF again.

I guess you meant to write "1.40". There is no "V1.41", is there?

[Winder]

Thanks a lot for reporting.

That's the third anecdotal data point that suggests that either V1.40 is a step backwards regarding AF, or that at least some measures must be taken (explicit AF adjustment resetting or new AF adjustment calibration).


I guess you meant to write "1.40". There is no "V1.41", is there?

Yes. 1.40

[rawr]

Your quote makes references to Canon camera models.

Nikon was there too. And Pentax. And probably even Sigma or Sony mounts are in that list.

Obviously the Sigma lens itself has no CDAF or PDAF module built into it, so the lens can't perform AF disconnected from the camera. But the camera body hands over responsibility for delivering AF at the lens mount, and from that point the lens firmware dynamically tries to implement the instructions emanating from the camera. If the lens firmware is imperfect, the lens won't interface optimally [responsively, precisely] with the signals passed from the camera body. Seems straight forward to me.

[Class A]

But the camera body hands over responsibility for delivering AF at the lens mount,...

Again, you are assuming that, are you not?

Again, everything points to the notion that a K-mount camera always stays in control of AF and merely instructs a lens with a built-in motor (HSM, SDM, ...) to move in certain ways.

If the lens firmware is imperfect, the lens won't interface optimally [responsively, precisely] with the signals passed from the camera body.

Of course. If the lens does not enact the movement commands from the camera as requested, there is a problem. However, there are many Sigma lens copies that have AF performance that is unproblematic. Your idea of a deep problem is not compatible with this observation, is it?