[liam bede]

FA77 and DA50 lenses are not especially great for AF but if the subject isn't moving fast and the apperture isn't too large with K3 is should work just fine. I'd expect a sigma 24-70 f/2.8 to perform significantly faster through.

Hi there, I'm interested in why this is. Also, which other sigma lenses would you say are faster focusing on the k3 (especially in low light and/or at wide apertures)? Many thanks

[stevebrot]

Hi there, I'm interested in why this is.

Welcome to the Pentax Forums!

I own both of those lenses and shoot them on the K-3. My assessment is that neither is particularly slow on that camera with the FA 77 being the slower of the two due, probably, to the higher mass of the optics and focus mechanism and fairly long focus throw. The DA 50/1.8 is not particularly slow. I measured in at a little over 1s from MFD to 10ft in a well lit interior room.* The same task using the DC motor in my internal focus Sigma 17-70/2.8-4 (C) @ 50mm took about 2s. FWIW, the 17-70 (C) has a DC motor and a reputation for quick focus.

Steve

* EV(100) 7

[Nicolas06]

Hi there, I'm interested in why this is. Also, which other sigma lenses would you say are faster focusing on the k3 (especially in low light and/or at wide apertures)? Many thanks

FA77 and DA50 are made with portraiture in mind. So precise AF is more important than fast AF. They also use screw drive that is far from the latest greatest focussing technology even if Pentax pushed it more than their competitors.

All lenses focus much better on K3 than on K5 because the AF system has much improved toward K5. Much faster to focus, far more decisive and much better low light performance. The AF point are also smaller (except the center AF point) and so the focussing can be much more precise to where the focus is made on the frame.

But technology wise only the most recent DA DC lenses and DFA lenses have modern focussing capabilities. The "old" DA* line use SDM that is rather slow. The sigma own implementation may not be faster than latest DFA lenses (DFA 15-30, 28-105, 24-70, 70-200 and 150-450) but all are made with FF in mind and very expensive. The sigma silent motor lenses are performing quite well and may be better for fast focussing if the DFA line is out of reach for you.

[stevebrot]

FA77 and DA50 are made with portraiture in mind.

Both are often used for such, but I don't know that either was ever marketed as a portrait lens. With the Pentax AF system, the camera does not make a trade-off of speed for accuracy depending on lens mounted. Accuracy trumps unless camera is set to release priority (when supported).

They also use screw drive that is far from the latest greatest focussing technology even if Pentax pushed it more than their competitors.

All lenses focus much better on K3 than on K5 because the AF system has much improved toward K5. Much faster to focus, far more decisive and much better low light performance.

Yes, screw drive is considered passe', but mostly due to the noise. Focus throw and the strength of the AF motor are both more important for speed and apply to both screw drive and in-lens systems. Under optimum conditions, the K-3 is faster than earlier bodies when used with screw drive lenses due to a much stronger focus motor. In sub-optimum conditions and/or with in-lens-motor, the K-3 still has as advantage due to improved AF system in general. As you noted, the K-3/K-3IIAF is more decisive overall.


Steve

---

Last edited by stevebrot; 09-25-2016 at 10:11 AM.

[Nicolas06]

Both are often used for such, but I don't know that either was ever marketed as a portrait lens. With the Pentax AF system, the camera does not make a trade-off of speed for accuracy depending on lens mounted. Accuracy trumps unless camera is set to release priority (when supported).

On a given camera, the max strength of the motor is fixed. Then it is a lens design decision to apply more or less demultiplication, to decide on the size/weight of the focus group and the length of the focus. The worst being typically a macro lens because one need to focus up to 1:1.

If lot of demultiplication is applied and the focus group is light enough, the focus will be faster, but less precise.

Modern in lens focus design are more than just a motor inside the lens. Feedback loop mechanism with sensors inside the lens allows to check how the lens process the camera order and fine tune acceleration/decelaration and greatly reduce the risk of front/back focus and increase precision. Smaller focus group allow for smaller motor that can then be more reactive and work better with contrast AF.

There a nice article on lens rental on the evolution of the evolution of AF. Basically before the best lenses AF bundled with PDAF where fast but less precise than MF or LV AF. Now they are both fast and precise and lenses made for contrast AF are no longer slow to focus. This is both software and hardware related.

Yes, screw drive is considered passe', but mostly due to the noise. Focus throw and the strength of the AF motor are both more important for speed and apply to both screw drive and in-lens systems. Under optimum conditions, the K-3 is faster than earlier bodies when used with screw drive lenses due to a much stronger focus motor. In sub-optimum conditions and/or with in-lens-motor, the K-3 still has as advantage due to improved AF system in general. As you noted, the K-3/K-3IIAF is more decisive overall.


Steve

Myself I have nothing against screw drive and don't care of the noise on the ltd. I see the value of screw drive in particular for a bit smaller design and a longer lifetime and no risk of in lens motor failure. Changing the body is easier than changing all the lenses.

But I also know there some limitations. I am sure most of them could be done keeping the in-body motor, but not all and the camera would have to work within the limitation of some old lenses (too much strength could damage the lens for example).

[stevebrot]

If lot of demultiplication is applied and the focus group is light enough, the focus will be faster, but less precise.

Stepping motors...they use stepping motors. Even the so-called wave motors actuate in discrete steps. Translation? They don't overrun. The base precision is based on the size of the steps. AF detection precision is related to the prism angles and number of photo-cells for each point in the detector.

Modern in lens focus design are more than just a motor inside the lens. Feedback loop mechanism with sensors inside the lens allows to check how the lens process the camera order and fine tune acceleration/decelaration and greatly reduce the risk of front/back focus and increase precision. Smaller focus group allow for smaller motor that can then be more reactive and work better with contrast AF.

Much of the above is quite true, though I do believe you are overthinking some of it. My understanding is the body gives the lens a rough notion of direction and movement. And while the in-lens motor is controlled by the lens, the feedback loop to attain focus is back through the body. The lens moves, the body reacts. In order for the body to give specific instructions as to movement, it would need to have intimate knowledge to map its focus histogram to lens movement parameters for the current focal length. I am not saying this is not how it works, but it would be nice to have a reference. Link?

As for front/back focus, it is my understanding that is related to the optical quirks of a particular lens and not its drive mechanism. After all, manual focus with the same lens requires the same fine adjustments. Drive precision may affect far-near vs. near-far front/back focus, but that is not the common cause. When such exists, an intermediate adjustment is used.

There a nice article on lens rental on the evolution of the evolution of AF. Basically before the best lenses AF bundled with PDAF where fast but less precise than MF or LV AF. Now they are both fast and precise and lenses made for contrast AF are no longer slow to focus. This is both software and hardware related.

I, for one, would appreciate a link to this article. I did several searches at LensRentals and came up with nothing. The idea of lenses being designed for CDAF as opposed to PDAF is new to me.

Sorry for being so contrary, but you really tossed a lot of stuff out there.

Myself I have nothing against screw drive and don't care of the noise on the ltd.

I guess not, if your signature is up-to-date All of those lenses are screw drive and the DA 15 is a bit of a speed demon. At least that is my memory from when I demo'ed that lens. I think we may have chased the OP away with all the talk. If so, that is a shame.

My advice has been and continues to be to buy for the optical qualities unless one has a high priority for quiet operation or speed. If one needs speed, then actual comparison testing with the intended body might be in order. Many (most) motor-in-lens lenses are not that quick and those are not limited to Pentax-brand. Likewise, not all bodies are created equal in their ability to drive the lens.


Steve

[Nicolas06]

Automatic LV AF vs Manual LV AF vs PDAF: https://www.lensrentals.com/blog/2012/07/autofocus-reality-part-1-center-poi...shot-accuracy/
Evolution of autofocus on 5DmarkII, 5Dmark III, asking PDAF to focus several time and evolution of lenses: https://www.lensrentals.com/blog/2012/07/autofocus-reality-part-1-center-poi...shot-accuracy/

More on Canon lenses: https://www.lensrentals.com/blog/2012/07/autofocus-reality-part-3a-canon-lenses/

To be short first article compare PDAF to CDAF to manual focussing. And concluding that CDAF or manual AF is better. The second AF figure out that when BOTH a modern lens AND a modern Camera, the PDAF is as precise as CDAF or manual focussing. But if you only have the new body or only have the new lens, this doesn't work. The third article confirm it.

Notice that it was first links in google when I did "lens rental autofocus" search in google.

I'll have to check back the optimized for CDAF vs PDAF thing. Sigma released a statement in an interview that CDAF require a redesign of the lenses to have smaller lighter focussing group to be more reactive and that why they couldn't just release DSLR lense son mirrorless by adding some extention tube and adapting to the mount: AF performance would be terrible (a bit like K01). CDAF can't guess if the good focus is nearer or farther. CDAF as a consequence need to make few small micromovement to optimize the constrast until is done. That why, historically it was so slow: it require many steps. PDAF on the contrary can compute the direction and the amount of focus shift to apply and directly give the order to the lens without needing to check again because the sensor is more advanced. That why historically PDAF is much faster: it can be done in 2 steps: 1 measure, 2 give order to move to right focus position.

[stevebrot]

Thanks for the links!

to be short first article compare PDAF to CDAF to manual focussing. And concluding that CDAF or manual AF is better.

I know and have read the article and it is basically conventional wisdom around these parts. The poor precision in PDAF (20% missed focus with the f/1.4) with both lenses is attributable to the f/2.8 limitation of the PDAF sensor.

The second AF figure out that when BOTH a modern lens AND a modern Camera, the PDAF is as precise as CDAF or manual focussing. But if you only have the new body or only have the new lens, this doesn't work. The third article confirm it.

The second link is the same as the first. I think you may have intended part II of the series (LINK), where he tested an older (1988) 24/2.8 against a new 24/2.8 on both 5DII and 5DIII bodies. Cicala did not provide bullet-point summary, but I can do so here:

• The older lens performed equally poor (accuracy and precision) for CDAF, PDAF, and manual focus. This was attributed to spherical aberration, a penalty of the lens design. This was true on both cameras.
• The newer lens performed better for CDAF and manual focus on the 5DII. PDAF on that body was better, but still poor.
• The newer lens performed similarly for CDAF and manual focus on the 5DIII, [B]but much better for PDAF precision and accuracy /B] in the same range as CDAF
• A new 28/2.8 lens also did very well on the 5DIII

Cool! The newer lens is better optically and the newer body is able to leverage that improvement for both CDAF and PDAF while the older body was not. Aside from the aperture sensitivity issue (not present at f/2.8), PDAF precision is affected by the focus point sensor design, both sensitivity and density of the linear detector array. Improvements in that area have been incremental for all makers.

The third part of the article (link 3) confirms that 4 out of 5 of the newer lens designs across various focal lengths have PDAF precision equivalent to CDAF on the 5DIII. Cicala's conclusions were that improvements to the body PDAF system were to thank. I suspect there may have been mechanical improvements in the lenses as well, though he did not go there.

I'll have to check back the optimized for CDAF vs PDAF thing.

Thanks! This is the part I was not aware of.

Other folk reading this may be interested in Cicala's earlier article where he attempts to explain how AF systems work:

https://www.lensrentals.com/blog/2010/07/how-autofocus-often-works/

Thanks again for digging up the links.


Steve

[Nicolas06]

I couldn't find back the exact Sigma interview despite spending quite some time on it. I think it is a bit too old.

Still I found something different, not exactly matching but still interresting in that direction: Sigma Q&A with Kazuto Yamaki: Why no Micro Four Thirds lenses, and is a full-frame Foveon feasible?

Extract:

E: Another question from our own staff, one of our technical editors. He's wondering if there's a firmware update or re-chipping service for some of the older Sigma lenses that have problems working on newer bodies. Now I think that both of these, I think I know what it is, and it's not so much a chip issue. This particular editor has a 10-20mm F4-5.6 EX DC HSM lens that won't auto-focus in live view on some of the newer Nikon bodies. And then in our lab, we use your 70mm f/2.8 DG Macro wherever we can because it's so sharp. So when we're testing cameras, that's pretty much as sharp as a lens can get. But again, our lab tech has found that the 70mm won't auto-focus in live view mode on some cameras. So our editor was asking if there's a chance to update the firmware, or put new chips in them?

But now that I ask the question, I'm also thinking that the problem with live view focusing is that the lens has to be able to move and stop very quickly. The camera has to be able to check focus, and then it has to move and stop again very quickly. So would those issues just be that some lens motors aren't suitable for contrast-detection focusing? Or is it something that could be fixed with the firmware?
KY: There's two issues. First of all, those lenses are made for still cameras. They're not made for live view autofocus. Live view autofocus is very similar to video usage. But the Hyper Sonic Motor -- or ultrasonic motors in general -- are not very suitable for that kind of purpose. Stepping motors or voice-coil motors are more suitable.
DE: Oh, I see. So ultrasonic maybe isn't able to accelerate very quickly, and stop very quickly.
KY: Yes, right, that's one of the problems, but also the other problem is firmware. Maybe we can optimize the autofocus function for live view mode, so we will continuously improve performance.

This is another hint that CDAF require different focussing technology than PDAF. If you notice, the latest 55-300 from Pentax isn't using screw drive or even DC, there a reason for that. An many mirrorless lenses don't use a mecanical motor anymore at all...