[BruceBanner]

I recently read a post on Facebook from a Pentaxian and he was alluring to the idea that some lenses just have better AF ability than others, that it's not Pentax camera bodies that are the problem for AF but rather a significant amount of the responsibility should be on the lens attached.

Interesting, I never thought of it that way. I mean for sure we have a review section here and we talk about AF accuracy and response and what not, but really opinions seem to vary quite a bit in this regard (indeed for example he found the DA 50/1.8 to have a higher success rate than the DA70 despite these forums rating the DA70 as being better for AF). I have owned a few lenses in my time but I can't really honestly say if this lens or that was better for AF accuracy and speed.

The same fella was suggesting that the more newer lenses that Pentax have released (DFA 50/1.4) as well as the DFA 15-30 (that he got to test) would have better AF accuracy compared to 3rd party lenses (Tamron, Sigma etc) and older Pentax lenses. He also proposed the idea that instead of switching brand for AF it might be wise (and cheaper) instead to invest in one of the newer lenses that Pentax has developed, it may very well offer the AF accuracy and response that you've been looking for.

Well... it did get me thinking. Do we (or can we) somehow create a database or list of lenses that seemingly perform very well in terms of AF accuracy and response? How could we go about that? Do we just look at the review section one lens at a time, take note of what the review section thinks it's performing for AF rating and do it that way?
What about the camera the lens is attached to, does this matter? Could the KP with the FA77 for example yield better results (in terms of AF accuracy) than the FA77 on the K-1?

Thoughts?


Bruce

[Adam]

Problems with AF accuracy are often user error (either through poor shooting technique with wide-aperture/telephoto lenses, or incorrect/missing fine calibrations). Apart from this, sharpness and contrast are important- and this is why most modern lenses do so well in terms of accuracy.

The speed of the lens may also play a role (the K-1, KP, and K-3 can take advantage of F2.8 lenses with the center point and the points directly above and below it; others see no difference between F5.6 and faster). Also, depending on the conditions, you may benefit from the low-light capabilities of the AF system in the K-5 II and newer.

Finally, in my opinion, screwdrive lenses are at a slight disadvantage compared to SDM/DC/PLM lenses because the focusing drive isn't as precise. Similarly, lenses with a long throw may not be able to reach the intended focus point before it's too late.

For all these reasons I believe that newer lenses will generally do better. But there are many factors at play and aggregate anecdotal evidence therefore isn't really a good way to come up with a score.

[Wild Mark]

I will second the idea of missing fine tuning as a contributor to poor AF performance. Users need to bear in mind that lenses out of the box are not all equal.

As for AF accuracy regarding new/ old, I tend to agree overall (but not totally). It is another factor and perhaps slightly less than the issue above.

[microlight]

Absolutely, Bruce; I’ve always thought that peoples’ difficulty with AF was probably more to do with lens rather than body. There’s no doubt that the DC-powered 20-40 is faster and more accurate (and quieter!) than the 16-45 that I used previously, and the PLM 55-300 is in a different league AF-wise to the previous screw-drive version.

[BarryE]

Yes, allowing for user error, they do differ. I recently compared my DFA 50 macro with a newly purchased FA43 as I was not convinced I wanted to keep the FA43. Also, I compared the FA35 f2 with FA31.

Allowing for the macro's slow AF due its extended throw, it was consistently better at focusing than the FA43. In the field I also find the 43 misses quite a bit - subjective, I know, but it was clear in a more controlled environment. Interesting, when comparing the 31 to the 35, as I was set-up to make these tests, the FA31 missed more often too, though the misses were not as frequent as the 43's.

So from my testing, I'm reasonably sure that the cheaper DFA 50 and FA35 are better at focusing than the Limiteds. I wanted, for my purposes, to get a handle on the 43's abilities and now I know and can compensate. Do I care ? Not really, as I can adapt and check.

Ref the FA35 f2. This little lens come so close and often exceeds the FA31 in usage, I do wonder why it doesn't get more recognition. It's so light and unobtrusive too. Love it ...

[beholder3]

But there are many factors at play and aggregate anecdotal evidence therefore isn't really a good way to come up with a score.

^This. +1

All autofocus discussions ignore the huge multidimensional space of scenarios of what is being discussed.

1. Precision for macro work favors long focus throws.
2. Generally geared autofocus drives (screw, DC, micro SDM) are at a disadvantage because of the gears unavoidable wiggle room.
3. Old used many years gears are worse than new lens gears
4. (fast) lenses with focus shift lack precision.
5. Precision discussion needs to also take into account lens speed/DoF/subject distance, which needs to be the same to compare.
6. Precision discussion needs into account "pixelpeeperness" with regards to camera pixel count. Obviously someone with a 8 MPx camera does not see the same as someone with a 36 MPx camera.
7. Precision discussion needs to have one subject scenario where the center spot AF point is being used and no afterwards tilting of the camera (recompose). Obviously if you recompose the results are skewed. If you dont use the F2.8 sensors the results are skewed. If you use extreme edge line sensors the results are skewed.
8. Precision is only repeatable from a tripod as physical human movement can be more than enough to affect results.
9. Precision comparisons also need to be done with the exact same subject as differing subject contrast edge situations will make a huge difference.
10. Precision comparisons also need to be done with the exact same lighting situation as differing light situations will make a huge difference. Low light and low contrast will introduce negative effects.
11. Camera settings such as focus priority will have impact
12. User behaviour will have an impact with regards to button mashing. There are people who slowly press the shutter button giving the AF system time. Then there are people who just press it in one go. Obviously the latter will lower AF results.
13. Finally all this makes only sense in a controllable AF.S setting as AF.C introduces tons of other non-repeatble effects.

[BruceBanner]

Ok, good feedback thus far but what I was really alluding to was the idea of PF to do some serious testing and post results, akin to what they did recently with the K-1 vs K-1ii AF Review.

We have to do the best we can to weed out all the variables, so that means everyone contributing or using the review database data probably opens up way too many variables to consider any information as valid. We need strict testing guidelines, AF settings consistent across the tests etc, but we also need a few camera models and of course a tonne of lenses to test!

Guess it will never happen...

[Digitalis]

Pentaxian and he was alluring to the idea that some lenses just have better AF ability than others, that it's not Pentax camera bodies that are the problem for AF but rather a significant amount of the responsibility should be on the lens attached.

While it is a nice idea to think that our shiny new cameras/lenses aren't at fault when the AF stuffs up - we have to also consider the possibility that when both are calibrated and corrected to the point where nothing could possibly go wrong, and it does: there must be an third possibility - user error. When absolute precision needs to be achieved, the human element is invariably what makes things go pear shaped.

Do we (or can we) somehow create a database or list of lenses that seemingly perform very well in terms of AF accuracy and response? How could we go about that?

This would be a very, very bad idea. It would be making unsupportable claims about lens performance, ignoring there always will be statistical outliers. With the majority of cameras I use*, I focus manually. Because at least I know where the blame lies in the event of mis-focusing.

* In truth: in the confines of the studio, having a camera with the worlds fastest AF isn't going to make for a more compelling photograph when you're taking photos of steaming hot ramen noodles in a bowl. I do use AF some of the time for moving subjects, I don't fully rely on it. There are literally thousands of examples of good sports and action photographs that were taken without the modern convenience of Auto-focus. I'm not saying my way of working is the best: each to his own. But it is a technical choice I make.

[Dericali]

In response to the facebook post, I would agree. That was a problem for the K1 when it first released, a lot of reviewers on decent websites tested the camera using only screw drive FA lenses and then complained about the autofocus being slow. Similarly the K1/K1 M2 autofocus comparison on this website didn't compare performance on any DFA lenses.

Personally I haven't used a lot of the older FA and DA lenses on the K1 (just 31mmLTD) but I have used most of the new lenses.

The Pentax DFA* 70-200 is easily the best in terms of autofocus performance. The main ways it excels are speed to focus and ability to achieve a lock in low contrast, low light situations.

The DFA* 50mm is an interesting case, I find it very quick to move in terms of its motor but I find it often struggles to achieve a focus lock in anything approaching remotely low contrast... a lot more than lenses such as the 24-70mm or the above mentioned 70-200. When I use the 50mm I get the feeling that this lens is waiting for a better performing body.

The 28-105 kit lens is decent too for autofocus. I really hope that the next Pentax FF body has more precision and spread in its autofocus point map.

[Pentax Syntax]

Yes, allowing for user error, they do differ. I recently compared my DFA 50 macro with a newly purchased FA43 as I was not convinced I wanted to keep the FA43. Also, I compared the FA35 f2 with FA31.

Allowing for the macro's slow AF due its extended throw, it was consistently better at focusing than the FA43. In the field I also find the 43 misses quite a bit - subjective, I know, but it was clear in a more controlled environment. Interesting, when comparing the 31 to the 35, as I was set-up to make these tests, the FA31 missed more often too, though the misses were not as frequent as the 43's.

So from my testing, I'm reasonably sure that the cheaper DFA 50 and FA35 are better at focusing than the Limiteds. I wanted, for my purposes, to get a handle on the 43's abilities and now I know and can compensate. Do I care ? Not really, as I can adapt and check.

Ref the FA35 f2. This little lens come so close and often exceeds the FA31 in usage, I do wonder why it doesn't get more recognition. It's so light and unobtrusive too. Love it ...

I second the statement that the performance of the FA35 f2 is great and it is no surprise that Pentax decided to offer in camera lens corrections for it and the FA50 1.4. The rendering of the FA35 f2 is different than the FA43 or FA31 limiteds. I have heard it described as more "clinical". While not quite sure what that means, I think they were referring to peripheral sharpness and bokeh differences. In any case, the FA 35 is a fast focusing lens which produces great results.

[automorphism]

I recently read a post on Facebook from a Pentaxian and he was alluring to the idea that some lenses just have better AF ability than others, that it's not Pentax camera bodies that are the problem for AF but rather a significant amount of the responsibility should be on the lens attached.

The AF speed is definitely dependent on the AF method of the lens. Newer lenses with a built-in motor will be better at driving the lens elements. That in particular contributes to the speed of locking on to a fixed target. But then there is also the tracking algorithm for AF-C, which is dependent on the algorithms programmed inside the camera. In this regard Pentax lags behind. Basically, overall AF performance is proportional to both the body software and lens AF motor (or with screwdrive, how well screwdrive can drive the lens).

[awscreo]

I can say for certain that my dfa 28-105 focuses much faster and more precise that any of my screw drive lenses.

Worst offender for me personally - Da50 1.8, it misses a lot (tried calibrating it, but the values are quite different in close distance and farther subjects), although Tamron 70-200 is pretty close to it. If I'm lacking some special technique on how to use af properly, I'd love to learn about it. But again, the dfa lenses I have focus very fast and accurately.

Tamron 90 2.8 macro focuses like a champ, fast and accurate, despitr it being a screw drive lens.

[aaacb]

I have lenses with screwdrive (limited primes and a couple others), DC (the DA 16-85), and PLM (the DA 55-300). I've had an SDM lens (60-250). All this with the same camera, a K3II.

For AF-S, some screwdrive lenses are less consistent than others (the DA35/2.4 which I had was towards the bottom, the FA 35/2 and FA43 is towards the top), but overall accuracy is about the same. For AF-C, both sports and concerts, the difference between a screwdrive lens (DA 70 or DA 40) and the PLM is night and day, with the SDM I had somewhere in between. At least for my k3ii camera, most of the time between the lens and the camera, the lens is the limiting factor. Even so, I have many sports and concert shots taken with screwdrive lenses, and I wouldn't give up on them for AF improvements alone.

I wonder how much wear and tear affects AF performance between the different implementations, especially consistency.

[BruceBanner]

While it is a nice idea to think that our shiny new cameras/lenses aren't at fault when the AF stuffs up - we have to also consider the possibility that when both are calibrated and corrected to the point where nothing could possibly go wrong, and it does: there must be an third possibility - user error. When absolute precision needs to be achieved, the human element is invariably what makes things go pear shaped.



This would be a very, very bad idea. It would be making unsupportable claims about lens performance, ignoring there always will be statistical outliers. With the majority of cameras I use*, I focus manually. Because at least I know where the blame lies in the event of mis-focusing.

* In truth: in the confines of the studio, having a camera with the worlds fastest AF isn't going to make for a more compelling photograph when you're taking photos of steaming hot ramen noodles in a bowl. I do use AF some of the time for moving subjects, I don't fully rely on it. There are literally thousands of examples of good sports and action photographs that were taken without the modern convenience of Auto-focus. I'm not saying my way of working is the best: each to his own. But it is a technical choice I make.

This is what I found out for myself this year. I decided to stop using AF entirely and master MF for all my lenses, even on jobs/events! I decided in order to increase my skill set for MF I had to be very strict with myself and resist the urge to toggle that AF button on. Indeed I believe the experience worked, and to this day I use MF more than AF, like a great deal more!
However I did swap over my Samyang 85/1.4 for the FA 77/1.8 to gain back AF and a lighter more compact lens. I did get caught out on a couple of events where I felt AF would have really helped, just like you say.

But the Facebook post made me think about my future purchases, for example, instead of saving for a FA43/1.9 should I continue saving and just aim for the DFA 50/1.4, not because I am necessarily needing that focal length and aperture (nor would I relish the additional size and weight) but rather the lens perhaps just has a far higher AF success rate in AF.C mode etc.

I would really like to see a proper Pentax review of AF between more classical lenses (the FA Limited's for example) vs some of the newer glass (DFA's).

The AF speed is definitely dependent on the AF method of the lens. Newer lenses with a built-in motor will be better at driving the lens elements. That in particular contributes to the speed of locking on to a fixed target. But then there is also the tracking algorithm for AF-C, which is dependent on the algorithms programmed inside the camera. In this regard Pentax lags behind. Basically, overall AF performance is proportional to both the body software and lens AF motor (or with screwdrive, how well screwdrive can drive the lens).

So is there a general consensus that certain lens AF tech is better than other for focusing? I mean I understand some are quieter, some give quick shift etc, but what if all you care about is precision and speed, what lens tech is supposed to deliver best?

[leekil]

So is there a general consensus that certain lens AF tech is better than other for focusing? I mean I understand some are quieter, some give quick shift etc, but what if all you care about is precision and speed, what lens tech is supposed to deliver best?

There was a Lensrentals blog post some time ago in which they tested a slightly older Canon camera's focus (tracking?) against something like a 70-200 Mk II, and it came up reasonably good. But then they tried it against a lens significantly newer than the camera, the Mk III version of the same lens, and the focus was much more accurate. They attributed it to a more precise focusing mechanism in the lens, that was able to give the camera's autofocus mechanism & algorithms the accuracy it needed. I think the same is probably true with the newer Pentax lenses vs. the older ones. And I believe there has been some discussion here about the superior tracking capabilities of newer lenses, which might be due to both more accurate control of the lens elements, and/or lenses being designed so that smaller & lighter elements are moved during focusing.