[loganross]

Hi there everybody!

As you might know from my introduction, I am a new owner of a K-3 and back in the Pentax world again. I've had the camera for a little over a week now and hadn't had much time to play with it. Yesterday I went on my first real "shoot" and it was basicly the first real test the camera got. I went to a formation dancing tournament for latin dances. The trainer of one team is a close friend of mine and this seemed to be a good chance to test my new K-3.

Now I know that Pentax isn't really first choice for shooting sports. The last camera I worked with however was a Canon EOS 300D with a relatively cheese lens and that had worked fine for me. I didn't expect my K-3 to have an AF like a Nikon D4, but I did expect it to outperform the EOS 300D!

My conclusion:
Either my camera is broken or the Pentax AF seriously sucks rocks! Sorry if this is a buzz-kill!

I missed a lot of moments yesterday because the camera took too long to focus. These dancers do move quickly, but not so much in the directions that make much of a difference to the AF. And even if they move towards the camera or away from it, they are not sprinting or as challenging to focus on as a tennis ball thrown at the camera. One thing dancers do very frequently is turn around. I shot a lot of butts yesterday, which in the case of the girls usually look pretty good. But they were not what I was aiming for. What I was aiming for was the pose 2 or 3s earlier. Sure I missed moments with the Canon, but those were less by far!

Just so you can see what I mean:
I shot about 360 photos yesterday, 139 of which were not during the actual dancing, but afterwards (handing out of trophies, posing, being happy about winning). The other 220 were of the dancing. I only shot one team which danced twice for about 5 minutes each round. Of these 220 photos I threw out 41 because they were completely out of focus. 57 I threw out because of "missed moments". Of the 139 after the dancing, only a handful had issues with the focus. I had time there to wait for the focus to happen - or not.

I don't know if this is relevant, but of the 139 photos three were completely over-exposed. All were shot with a flash. When I shot portraits of couples, I shot a fair amount of redundancy. Then it happened that of three practically identical portraits one was completely over-exposed.

Now don't get me wrong! Among those remaining photos there is still a great deal of crap, like where I messed up the composition, the shots weren't to flattering or the like. But those are mistakes I made. I just don't need the camera to make mistakes for me.

My gear:

• K-3 (obviously) with firmware v1.02
• kit lens DA 18-135mm f/3.5-5.6 WR (and some other letters)
• D-BG5
• Metz 58 AF-2

The place was a gym, pretty well lit, but not enough for the kind of shutter speed I wanted. This was not the kind of light were you could run into trouble with the focus because of low light. I set the camera to X, shutter speed 1s/180 and aperture f/5.6. The flash was set to P-TTL and aimed directly forward. The dancers had a distance to me of something between 2.5m and 10m.

The whole thing smelled pretty fishy to me, so I did a completely unscientific test today. I went outside in front of the house, sat on the stairs there and tested the focus speed on both the Canon and Pentax. It was early afternoon, so there was plenty of light, which was soft because it was overcast. I first aimed at a grating, about 60cm from me and then at a barn, maybe 100 or 150m from me on the other side of the road. I went back and forth a few times. The Canon 300D was fitted with an EF 28-80 3.5-5.6 IV Ultrasonic. I tried the Pentax out with the kit lens and with a Sigma 85mm f/1.4 EX DG HSM. Both AFs were set to single shot.

Result:
The old and cheesy Canon was significantly faster! I didn't use a stop watch because the Canon was just so fast that I couldn't have pushed the button that quickly. Using my "internal timer", I'd say the Canon takes about 0.1 to 0.2 seconds to focus between the two points. The K-3 takes nearly a full second to do the same trick. I can nearly always say the German word for twenty one, which has four syllables, before the beep. From time to time, I get cut off after the third syllable - never earlier. Even if the Canon hunts (goes completely past the focus point and back again), it is still much faster than the K-3 in best time!

What I also noticed:
When focussing, the K-3 at first moves smoothly, then stops and then seems to "tick" the lens again twice or more before reaching focus. My Sigma primes show this behaviour with two or three (sometimes even 5) audible "tocks" before focus is reached.

Is this all normal or could my K-3 actually be broken?

Kind regards,
Pix

P.S. Sorry about the long text! I guess I'm a little upset because to me this camera cost a lot of money and seems to be outperformed (in terms of AF) by an entry-level DSLR that is 10 years old (and the lens is even older).

See my thread here:
https://www.pentaxforums.com/forums/172-pentax-k-3/250297-compare-notes-k3-af...ness-cont.html

[PixelGeek]

Good evening, everybody!

Thank you all for your kind words and understanding!

Before I get started, one comment @loganross:
You are describing a completely different issue in your notes, if I didn't misunderstand you.

To the subject at hand...
This took pretty long, but I have my camera back (again). The first time I sent it directly to the Pentax service, this time to my dealer. I actually got it back last Monday but didn't have any time to play with it until today. I'm kinda a working guy.

I won't go over what happened the first time (when I sent the camera directly to Pentax). This time I sent it to my dealer. They checked it out, confirmed what I described in my email and sent it on to Pentax for verification. My dealer did compare my camera with a demo model the keep in the store and told me that although they saw what I did, their demo model worked just like mine.

On Monday when my camera was in the mail, I checked and found a service report from Pentax. They had aligned the lens (I wonder why that was necessary again), cleaned everything but could not find anything wrong with the camera (body) itself: Working within normal parameters. As a bonus, my dealer included their demo K-3 for me to try out and compare my camera too. I also got an offer that I could send back the camera if I still am not happy.

I sat outside on my porch again and did my little back and forth test again - and I recorded the sound again (sorry, had to use ogg this time):
http://media.advico.de/audio/kit_2014-03-16_phase.ogg
http://media.advico.de/audio/kit_2014-03-16_contrast.ogg

This is once again the kit lens (18-135mm), set at ~100mm, f/5.6, PDAF set to spot. I didn't bother with any of my Sigma lenses. For some reason, I decided to play with the live view and noticed something strange: It's bloody fast - and it sounds right! There are no micro-corrections using the contrast-AF. The lens just moves and locks. Ok, when I change focus from from near to far the camera moves the lens in the wrong direction and then back again. But even this feels pretty fast considering the "distance" the lens has to move.

Using my audio editor I got the focus times again (W is worst case, B is best case). Because I can't hear me pushing the shutter button, all timing is done from the first moment I can hear (or see in the graph) the focus motor running until right before the camera beeps. This means, the actual beeping is not included in the timing. This probably wouldn't add much (I didn't bother to check), but I left it out anyway.

PDAF
W: 1s, B: 0.55s

Contrast-AF
W: 0.9s, B: 0.37s

I think these times are quite remarkable! Although this is actually not the issue here, I think the contrast-AF times are quite excellent. At the same time, I think it is borderline shocking that the PDAF is slower than the contrast-AF - which is actually a first for me. I have not read, seen or heard one single review where something similar was described. This has me pretty much speechless.

I took the extra K-3 out on the porch with me. While I didn't record it's focussing like I did with mine, I can tell you that it felt exactly the same. The PDAF made the same micro-corrections, the contrast-AF was smooth.

All that being said, I have to conclude that my K-3 is not faulty. I am very happy to argue this point, but currently, the evidence (two trips to the Pentax service and a second K-3 that feels just the same in direct comparison) pretty much points to the K-3 (as a model) having a pretty slow PDAF, which is outperformed by the device's own contrast-AF.

Giving the camera back is not really an option to me. I have those two Sigma lenses and the Flash for Pentax. The shop that sold the camera didn't have the other stuff I wanted, so I got that elsewhere. That shop probably will not take the lenses and other stuff back after about 4 weeks, but I guess I could try if push comes to shove.

Currently, there is no other camera that I really want. None other has the internal image stabilizer, the LPF simulator or the magnesium alloy body. The D7100 has a little bit of magnesium, the 70D has none. The D800 could be an alternative, but it's in a completely different class and costs much more, which pretty much goes for the 5D aswell. The K-3 was pretty much the best I could get for my budget and what I wanted it for - save the AF.

I'd really like to read your opinions on these recent developments. If my camera is really within the usual parameters, I just don't understand why not a single review said something about the AF being this slow. One guy even used the K-3 for shooting wildlife! DigitalRevTV (YouTube) compared the D7100 to the K-5 II and said the Pentax AF outperformed the D7100's in low light - and these are pretty much Nikon guys! What am I doing wrong? At the moment, I'd just about listen to anything that sounds reasonable reasonable.

Kind regards,
Chris

[wizofoz]

Chris, I have just read all through this thread and have a couple of suggestions for you. I think your problem stems from a couple of sources. -

My understanding is that Canon have used their algorithm so that the 'beep' is heard faster, so that the perception of AF is faster, compared to the Pentax approach of trying harder to make the AF more accurate and giving it time to micro adjust ( We can always just se the camera to release priority and ignore the micro adjustments). So perhaps the combination of Canon beeping slightly before true AF is attained, and Pentax waiting for the final adjustment contributes to your perception. Perhaps.

What I'm saying is that maybe the AF of the K-3 is just as good or better than your older canon, but the 'perception' of when acceptable AF is aquired is different, I dont think I read anywhere where you set the K-3 to release priority to see what the time lag is between sighting your target and actually taking the shot (as opposed to waiting for the final beep)

And of course a more responsive lens would help, but you are saying the Canon lens is no speed demon, so that is a wash I guess.

Using P-ttl flash would also contribute delay as has been pointed out, but you have discounted the flash in your 'front step' comparisons.

I dont think I read your camera settings ( or I have forgotten if you did list them - it took quite a while to read through... )

I set my K-3 up so that I use the back AF button only for AF and disable it on the shutter button, I invariably set my AF mode to AF-C and I have my menu setting for focus priority only the first frame. if I were using flash in a multi exposure setting I would think about changing this to release priority all the time. I would probably use auto 27 point tracking in the dance situation, but i may revert to 9 points, or even spot depending on how the situation unfolds.

I would give your dancers one more try, but use a manual setting on your flash (or perhaps auto thyristor if it has one) and set it to release priority and practice using the back button for AF (the camera sets up so much better using the back AF button in my experience)

I don't have the 18-135 so I cannot replicate your situation exactly, but just picking up my camera, which happens to have the 16-50 mounted, I cannot tell the difference in AF speed between PDAF and CDAF so perhaps you do have a defective K-3. I even tried the notoriously slow 50-135 and still cannot tell any real difference between AF speeds in either mode.

Give it another try with more optimal settings and see if it satisfy's you. If not, there are other camera makes and models. I hope you find one that suits. Good luck

[cjsong79]

Hi again,

The AF slowness of current lineup of Pentax DSLR is not linked to the body, but linked to the lens. You can experiment by focusing on one point, and trying to focus again on exactly that point. It will be less than 0.2 sec. Now focus the lens at infinity and then focus at something close.... Depending on the lens thats going to be from 0.2 S (fastest canon lenses) to over 1S (slower pentax lens like 55-300), for 18-135 it should be quite fast at the wide end (0.2-0.3 s) and at tele end it should be quite a bit slower than canon lenses (close to 1S)

[DSims]

Don't even try to use your DA18-135 for sports - get screw-drive primes or a Tamron 70-200/2.8.

The Sigma 85/1.4 is the only lens you should be trying for sports right now - but I don't know how fast the HSM really is. The FA*85/1.4 works well - FA77/1.8 should too. F/FA135, FA*200, F/FA/DA*300 are other options.

[PixelGeek]

Don't even try to use your DA18-135 for sports - get screw-drive primes or a Tamron 70-200/2.8.

The Sigma 85/1.4 is the only lens you should be trying for sports right now - but I don't know how fast the HSM really is. The FA*85/1.4 works well - FA77/1.8 should too. F/FA135, FA*200, F/FA/DA*300 are other options.

Although I am thankful to any person who wants to help me out whem I write about a problem, I sometime do wish they would read the thread or at least the original post. Had you (and cjsong) done that, you would know, that this is not about the focussing motor. While admittedly the DA18-135 doesn't have the fastest motor I have seen, it would be plenty fast enough for me if the focussing would consist of one movement and a lock. What really costs time are those microcorrections (and the pauses between those). Because of them BTW, the Sigma 85mm and the 50mm EX primes are not faster than the kit lens - at least I can't tell the difference when using them.

Concerning the lenses you suggested...
None of them are of any real use to me because the focal lengths are way too long for for what I am doing. As I wrote in my OP, the subjects I am shooting are in some cases less than 1m from me and usually no further than 10m, the most common field being between 2 and 5m. Not all sports are outside and across a giant soccer or baseball field. Although I have and use prime lenses, they are of little use in this setting. I am pretty much stuck to the place I am sitting. Standing up is not an option because the audience could be a little annoyed with me if I get in their line of sight. I just can't move around enough to get anything with my primes. I have borrowed a DA*16-50mm 2.8 for next weekend. I am going to give it a shot, but I am not sure if that will be ideal for me, especially because the 50mm at the long end does seem a little short.

Kind regards,
Chris

[vonBaloney]

Although I am thankful to any person who wants to help me out whem I write about a problem, I sometime do wish they would read the thread or at least the original post. Had you (and cjsong) done that, you would know, that this is not about the focussing motor. While admittedly the DA18-135 doesn't have the fastest motor I have seen, it would be plenty fast enough for me if the focussing would consist of one movement and a lock. What really costs time are those microcorrections (and the pauses between those).

One movement and a lock is probably not even possible if you want to find the best focus, although it could just reach some threshold of acceptability and then stop. Problem is some lenses are better (some MUCH better) than others and so you'd have to pick a lowest common denominator which wouldn't please anybody. Focus isn't a math problem where it can just calculate the answer and then just GO TO THAT SPOT. It is a search, and so it can never know if it has hit the optimal spot until it goes past it, and so then has to backtrack. I believe modern AF systems do get some assistance in the form of data from the lens that will help focus that particular lens to make it as efficient as possible, but it is still a search so I don't see how you could totally get rid of micro-corrections and still find the optimal point. Right?

[PixelGeek]

One movement and a lock is probably not even possible if you want to find the best focus, although it could just reach some threshold of acceptability and then stop. Problem is some lenses are better (some MUCH better) than others and so you'd have to pick a lowest common denominator which wouldn't please anybody. Focus isn't a math problem where it can just calculate the answer and then just GO TO THAT SPOT. It is a search, and so it can never know if it has hit the optimal spot until it goes past it, and so then has to backtrack.

Actually, PDAF is a mathamatical process, although the calculated result is not perfect. This means, the direction the AF has to move should be detected every time (if there is enough light and something to focus on) and the amount of movement required to reach focus should be pretty close. You make it seem a bit like a trial-and-error method which it is far from.

I believe modern AF systems do get some assistance in the form of data from the lens that will help focus that particular lens to make it as efficient as possible, but it is still a search so I don't see how you could totally get rid of micro-corrections and still find the optimal point. Right?

Well, getting rid of microcorrections totally is probably impossible. That would mean a perfect AF and I doubt that we will ever get there. We are getting off-track here a bit, but this is an interesting discussion, so I don't mind having it, as long as it stays a discussion and noone starts to rant because I critisised Pentax.

Canon and Nikon seem to be doing a better job with their AF. I don't know how or why, but I could venture a guess. The first time the PDAF take a measurement the amount of required movement may not be to exact. This depends on how far off focus the lens is. Imagine, the lens is completely off focus. One way to get an exact hit would be to take a measurement, calculate, move to that rough position and take another measurement and so on, until you hit focus. This method would probably require three or so corrections. You could of course keep taking measurements while the focus is moving and thus get several calculated focus positions, each getting closer to perfect. If this system works well, you should be close to being able to focus in what seems like one movement (and possibly is).

But I actually don't expect my K-3 to be that good. It would be fine with me if the PDAF worked like the contrast AF, with the advantage of knowing the right direction. In my post from March 16th I put in a link to two sound files. Firefox can play there directly. Listen to the contrast AF and notice how quick is it when if moves in the right direction. There is one correction at the end and it's fast, while the PDAF has three corrections which feel a bit slower too. On average, PDAF is slower than contrast AF and that's just weird.

Regards,
Chris

[vonBaloney]

Actually, PDAF is a mathamatical process, although the calculated result is not perfect. This means, the direction the AF has to move should be detected every time (if there is enough light and something to focus on) and the amount of movement required to reach focus should be pretty close. You make it seem a bit like a trial-and-error method which it is far from.

Oh no, I know that -- even a simple binary search algorithm is pretty efficient, and I'm sure it is more sophisticated than that. Modern lenses that are capable of giving detailed data to the body should be able to do pretty well. (And apparently the process of having your lens re-calibrated for front/back focus often involves nothing more than editing some numbers in the lens rom data.) I don't really know how it works exactly (but I do know about search algorithms) -- I'm assuming the oldest auto-focus models are basically "dumb" (i.e. not giving the camera much or any data), but many are still able to focus quite quickly. (As-as the AFA 1.7x adapter if you are in the ballpark already.) But a good modern system with a modern lens should do things like "know" where the lens is focused already before being asked to focus anew, have some sort of built-in equation about the lens and the amount of rotation needed at a particular distance to make an x-sized adjustment, information about aberrations that will throw it off, etc. Again, I don't really know how it works but I'm sure research is on-going in all of these areas. Everyone says that Canikon is better, but how is this achieved? Is it only on modern lenses that it is better? (i.e. if you put the oldest compatible AF lens on a modern Nikon body, does it still beat the pants off Pentax?)

[traderdrew]

Don't even try to use your DA18-135 for sports - get screw-drive primes or a Tamron 70-200/2.8.

The Sigma 85/1.4 is the only lens you should be trying for sports right now - but I don't know how fast the HSM really is. The FA*85/1.4 works well - FA77/1.8 should too. F/FA135, FA*200, F/FA/DA*300 are other options.

The newer Sigma 70-200 F/2.8 looks like a good lens for sports. I used the older 70-200 macro version for sports and I would say it definitely outperformed my DA* 60-250 F/4 in speed. I wish I more of my lenses had that hypersonic motor.

If you look at my flickr and 500px.com photostreams for birds, you wouldn't think I had an autofocus problem. Part of it is how I use it. One tip I can give others is if you see a bird in your viewfinder, don't always try to capture it in perfect focus right away. Track it and when it is in prime range, don't hesitate to take the shot. However, it really depends on the type of lens you are using. I came from a background where I had to make things work and not complain about them. Yes there are better cameras out there for birds but at what cost?

The K-3 is an autofocus improvement over my K-5 for sure. I suspect lenses with better T stops improve autofocus accuracy.

By the way everyone, Yes, definitely align your best lenses with Lens Align. I wish Pentax would build a memory into the camera on where each lens needs to be set after they are aligned.

[DSims]

Although I am thankful to any person who wants to help me out whem I write about a problem, I sometime do wish they would read the thread or at least the original post. Had you (and cjsong) done that, you would know, that this is not about the focussing motor. While admittedly the DA18-135 doesn't have the fastest motor I have seen, it would be plenty fast enough for me if the focussing would consist of one movement and a lock. What really costs time are those microcorrections (and the pauses between those). Because of them BTW, the Sigma 85mm and the 50mm EX primes are not faster than the kit lens - at least I can't tell the difference when using them.

Concerning the lenses you suggested...
None of them are of any real use to me because the focal lengths are way too long for for what I am doing. As I wrote in my OP, the subjects I am shooting are in some cases less than 1m from me and usually no further than 10m, the most common field being between 2 and 5m. Not all sports are outside and across a giant soccer or baseball field. Although I have and use prime lenses, they are of little use in this setting. I am pretty much stuck to the place I am sitting. Standing up is not an option because the audience could be a little annoyed with me if I get in their line of sight. I just can't move around enough to get anything with my primes. I have borrowed a DA*16-50mm 2.8 for next weekend. I am going to give it a shot, but I am not sure if that will be ideal for me, especially because the 50mm at the long end does seem a little short.

Kind regards,
Chris

My terse response wasn't just being flippant. If you had more experience with a wide variety of lenses on the Pentax system you'd know the lens choice has a lot to do with it.



BTW, dance is exactly what I was shooting a week ago, on Friday and Saturday. I used lenses from 24mm to 300mm. 24 and 31mm were the most difficult, because the dancers need to come really close for the most effective photos (and unlike my other lenses, I had AF accuracy issues with my FA*24/2). I was either sitting in the front row (in a gym), or sitting on the floor in front of the front row (on the basketball court sideline) to get even closer. I sat on the floor most of Saturday. Overall I like that angle better, looking up slightly at the dancers. It was my first time shooting dancers, but at least I've shot other action and am familiar with my camera system. I'm still organizing the gallery, and may go back and adjust the exposure on some. I threw out about 50% from Friday, but only about 25% from Saturday (with the exception of the 24mm lens, which did much worse). What remains is a lot of OK photos and a few pretty good ones. So you can have a reasonable "success" rate. Surprisingly, I was actually able to enjoy looking at the Saturday photos in the site's rather slow slide show mode for over an hour (I thought I'd never last 5 minutes) because they seemed to tell a story, so I guess they're decent. You can see them here: EventTime Photos

The most difficult part for me was keeping the dancers in the desired side of the frame. This was especially hard because I didn't know their routines. I normally choose an upper corner AF focus point, but I had to resort to an upper center focus point because they moved from side to side so quickly. Somehow I still managed to keep the dancer I was focused on out of the mundane-looking dead center of the frame much of the time, and hopefully facing toward or moving into the center of the frame, as is usually desired. I usually feel exhausted after shooting action, because each frame takes a bit of work to compose and track the subject.


The other thing that matters a lot is the camera settings and how you use it. Typically you want AF-C; you may want to use the AF button on the back, and possibly disable the shutter button half-press AF; you need to tell the camera whether it shoots right when you press the button or only after it believes it has "perfect" focus (hence the micro-adjustments to the AF). You usually want to manually select the focus point yourself, but the K-3 is even more capable than previous bodies, so you might try other options as well. And I'm sure there are a few other settings I'm forgetting right now. There's even a chance you can make the DA18-135 work sometimes with the proper settings and technique, so it's worth a try. But it might still be an exercise in futility.


If I had to choose one lens for you to use it would be the FA*85 or the Tamron 70-200/2.8. I have the DA*16-50, and I can't imagine using it in this situation (I believe the AF is too slow, for one) but give it a try if you like. The F or FA135 also does well, and probably the FA77 (I don't have mine anymore). Other possible lenses where I'm not as sure about the AF performance include the DA70, DA50/1.8, and even DA35/2.4 if you can get close enough. I'd choose any of these over the DA18-135 or my DA*16-50. Primes are actually very practical in these situations, especially if you have 2 bodies with a different lens on each. And unless you're happy with a lens like the Sigma or Tamon 70-200, it's about the only way to get really good results on a Pentax, because of peculiarities of the system (it may also be a good way on Canon or Nikon, but few people do it under 200mm, since they optimize their 70-200 lenses for AF speed).


This is why I say try the Sigma 85/1.4 again if the AF can perform quickly enough using the proper camera settings.

---------- Post added 03-31-14 at 12:51 PM ----------

The newer Sigma 70-200 F/2.8 looks like a good lens for sports. I used the older 70-200 macro version for sports and I would say it definitely outperformed my DA* 60-250 F/4 in speed. I wish I more of my lenses had that hypersonic motor.

If you look at my flickr and 500px.com photostreams for birds, you wouldn't think I had an autofocus problem. Part of it is how I use it. One tip I can give others is if you see a bird in your viewfinder, don't always try to capture it in perfect focus right away. Track it and when it is in prime range, don't hesitate to take the shot. However, it really depends on the type of lens you are using. I came from a background where I had to make things work and not complain about them. Yes there are better cameras out there for birds but at what cost?

The K-3 is an autofocus improvement over my K-5 for sure. I suspect lenses with better T stops improve autofocus accuracy.

By the way everyone, Yes, definitely align your best lenses with Lens Align. I wish Pentax would build a memory into the camera on where each lens needs to be set after they are aligned.
Last edited by traderdrew; Today at 09:11 AM.

The Sigma 70-200/2.8 OS HSM can definitely be used here as well. I tried it (the latest version) this previous weekend at the dance championships. On the K-3, it's reported that the Tamron can now come really close to it in AF performance. But it costs more than the Tamron, and I don't like the image rendering as much. Images are sharp, but lack life compared to Pentax primes, I think. The Tamron isn't perfect, but I'd prefer it over the Sigma, especially for the price. Nevertheless, the Sigma is a good, well-built lens with good AF performance.

Thanks for the confirmation about the AF speed of the DA*60-250 for sports, since I haven't been able to try one in that situation. I like the images from it better than the Tamron or Sigma 70-200s, but I think the slower AF speed combined with the f/4 aperture could be a deal-breaker for fast action shooting (and the dance routines I was shooting certainly qualified as fast!).

+1 on the Lens Align.

[cjsong79]

hi
its a little sad to hear that my reply is not helpful, i did read all your previous posts, and i actually created my forum membership because i wanted to reply to this thread - as i have been struggling with the same question for the last few years !

i will re-clarify : an old rebel body does not perform closed loop focus so it is normal there is no hesitation. I am sure K-3 does a closed loop algorithm. Recent canon entry models perform closed loop algorithm

But again, it is the lens which is the real culprit, because makng micro adjustment 2-3 times before settling in will be very noticeable with screw lenses, while with a fast USM (eg canon), you wont hear a thing or feel any vibration

i have shot tamron 17-50 with both my pentax k-5 and canon bodies, the focus speed is basically the same.

same on the canon side, if u use lens like 50mm f1.4 or old 35mm f2 (which is not super quiet) you will notice its very obviously doing some last second back/forth before settling down the AF (you dont feel this with lens like 24-70, 17-55, 17-85, 15-85)

so again, what i am trying to tell you is that

1. it is normal that an autofocus algorithm does not finish with a single calculation/move (open loop), and does a few more moves. this additional move is more visible on camera that does closed loop AF, and has screw drive lens which is unable to make quick and undiscernable micro movements, though on some cases its actually quite fast (35mm f2.4), but just noisy

2. The problem of speed of single shot AF responsiveness of Pentax is not in the body (it peforma on par with canon if using the same third party lens), so future model is unlikely to address this issue, but introducing a new lens lineup whose focusing motor is comparable with canon will address this issue (unlikely to happen). As of today, the fastest focus on pentax is sigma 17-50 / 70-200 lineup where the speed is slightly slower than most canon USM zoom, but faster than pentax zoom

3. Pentax is not the right system to shoot where the limiting factor for getting the shot is AF speed. it is adequate for normal day to day situation, but is very difficult to use on situation where things are moving, there is a reason why action shooters mostly use higherend canon/nikon.

I personally think the best would be to try another lens (sigma 17-50) and if thats also a problem, to switch to canon

CJ

[DSims]

its a little sad to hear that my reply is not helpful, i did read all your previous posts, and i actually created my forum membership because i wanted to reply to this thread - as i have been struggling with the same question for the last few years !

Thanks for the very informative and useful post - I certainly learned a few things!

The problem of speed of single shot AF responsiveness of Pentax is not in the body (it peforma on par with canon if using the same third party lens), so future model is unlikely to address this issue, but introducing a new lens lineup whose focusing motor is comparable with canon will address this issue (unlikely to happen).

The only part I disagree with is that I think Pentax will address this issue relatively soon, but only on a very limited number of (but definitely not the Limited) lenses. I think they were hinting at this in the recent interviews. My guess is they'll try to address this in the upcoming *70-200 lens on the roadmap - and possibly only that lens for now.

I personally think the best would be to try another lens (sigma 17-50) and if thats also a problem, to switch to canon

I agree with this as well - it could be worth switching to Canon if it doesn't work out. But I also think it's worth giving some screw-drive lenses a chance, since (as you pointed out with the DA35/2.4) they can focus rather fast, and the K-3 is the best platform yet for screw-drive lenses.

[traderdrew]

I was out in the field yesterday with my K-3 and my 500mm Sigma lens. I had some autofocus problems and then I cleaned my contacts and my mount and it's back working well.

[PixelGeek]

My terse response wasn't just being flippant. If you had more experience with a wide variety of lenses on the Pentax system you'd know the lens choice has a lot to do with it.

My slight irritation by your response had nothing to do with the fact that is was short. What I find frustrating in this discussion is the fact that things I write are being ignored. Imagine an argument like this:
him: The chair is green.
her: I disagree! The table is blue.
This argument is not going to be settled any time soon.

I do not doubt for a second that you guys are really trying to help me out here and I hope you believe that I really believe, I want to learn something. But for that to work we have to be talking about the same things. I have stated several times that neither the 50mm Sigma prime nor the 85mm Sigma prime is any faster (at focussing) than the kit lens (DA 18-135mm) - and yes I did try all of them. If you suggest the 85mm SIgma when I write about slow AF, then one of us is missing something. I will go into the focal length issue a little further down.

Somthing similar applies to a post after yours concerning microcorrections and screwdrive focus. The poster says this combination makes the microcorrections very obvious while SDM or HSM tend to "hide" them. I currently have no Pentax screwdrive lenses and all of my SDM (HSM) lenses show very noticeable microcorrections.

Basicly, it's fine if you disagree with me, but then please disagree with me!

Another detail, that might be unclear:
Photography is a hobby of mine. This means, I watch my budget and although I probably could spend more money on my stuff, the money is a very important factor to me. I also tend to buy less lenses and try to get good ones. I actually bought the two Sigma lenses for portraits, not for action. I do not make any money with my photos, I take photos because I enjoy doing so and it relaxes me. Unless I suddenly turn rich, I will probably never buy a second body.

BTW, dance is exactly what I was shooting a week ago, on Friday and Saturday. I used lenses from 24mm to 300mm. 24 and 31mm were the most difficult, because the dancers need to come really close for the most effective photos (and unlike my other lenses, I had AF accuracy issues with my FA*24/2). I was either sitting in the front row (in a gym), or sitting on the floor in front of the front row (on the basketball court sideline) to get even closer. I sat on the floor most of Saturday. Overall I like that angle better, looking up slightly at the dancers. It was my first time shooting dancers, but at least I've shot other action and am familiar with my camera system. I'm still organizing the gallery, and may go back and adjust the exposure on some. I threw out about 50% from Friday, but only about 25% from Saturday (with the exception of the 24mm lens, which did much worse). What remains is a lot of OK photos and a few pretty good ones. So you can have a reasonable "success" rate. Surprisingly, I was actually able to enjoy looking at the Saturday photos in the site's rather slow slide show mode for over an hour (I thought I'd never last 5 minutes) because they seemed to tell a story, so I guess they're decent. You can see them here: EventTime Photos

I've taken a look at the link an there are some nice shots. However, these are not the kind of shots I am going for. I used to dance as a sport too, so I know what is important to dancers. I like the "portraits" you short, but with very few exceptions, I always try to get the whole body in the shot, because I want to capture more of the dancing while your photos seem to go more for the personality. This is not a bad thing, just a different emphesis. Here are some examples of shots I made last time...






Note: These photos actually were sharp when I submitted them.

Now you can hopefully see why I can't work with 85mm. For that to work, I'd need a distance of at least 5m to the subject for a full-body shot - something I just can't do most of the time. I would never have gotten that line of dancers with 85mm. Additionally, because I work with a flash, the distance shouldn't be that big either.

I get that the kit lens isn't ideal - not by a long shot. But if the focal length of a lens just doesn't let me catch the subject I wish, then talking about its other qualities is pointless. I have used a full-format Canon before with the 24-105mm f/4 L IS. I would like something similar for my Pentax. That would mean something like 16-70mm, if the long end goes a little further, good. I know there are 17-70mm lenses, but these are not DA* (for Pentax) or EX /for Sigma).

The other thing that matters a lot is the camera settings and how you use it. Typically you want AF-C; you may want to use the AF button on the back, and possibly disable the shutter button half-press AF; you need to tell the camera whether it shoots right when you press the button or only after it believes it has "perfect" focus (hence the micro-adjustments to the AF). You usually want to manually select the focus point yourself, but the K-3 is even more capable than previous bodies, so you might try other options as well. And I'm sure there are a few other settings I'm forgetting right now. There's even a chance you can make the DA18-135 work sometimes with the proper settings and technique, so it's worth a try. But it might still be an exercise in futility.

I rarely use AF-C, because it too often misses (for my taste). I know most of the routines, so I usually know when I want to take the shot.

If I had to choose one lens for you to use it would be the FA*85 or the Tamron 70-200/2.8. I have the DA*16-50, and I can't imagine using it in this situation (I believe the AF is too slow, for one) but give it a try if you like. The F or FA135 also does well, and probably the FA77 (I don't have mine anymore). Other possible lenses where I'm not as sure about the AF performance include the DA70, DA50/1.8, and even DA35/2.4 if you can get close enough. I'd choose any of these over the DA18-135 or my DA*16-50. Primes are actually very practical in these situations, especially if you have 2 bodies with a different lens on each. And unless you're happy with a lens like the Sigma or Tamon 70-200, it's about the only way to get really good results on a Pentax, because of peculiarities of the system (it may also be a good way on Canon or Nikon, but few people do it under 200mm, since they optimize their 70-200 lenses for AF speed).

As I have said, the focus motor is not the issue here. I have seen reviews of the 16-50mm where people say the motor is slow. This is not my problem. The speed of the contrast AF I could work with, no worries.

Cheers!
Chris