[TomB_tx]

I don't think your experience is unusual - one of the reasons I still avoid autofocus after 50 years shooting manual focus... but dSLRs all have lousy viewfinders for manual. I want a camera to always fire the shutter the instant I press the release, so it has to be in-focus at that time. At least with manual focus the tracking and focus is still in my control, and the shutter timing is up to me.
If your lenses depend on the screw-drive motor in the camera they will be slower than more modern designs with SDM or other focusing motors in the lens. (Not that buying new lenses is a good solution.)
And - since I didn't see it mentioned - I assume you were using the viewfinder and not live-view, which is much slower focusing.

[PixelGeek]

I don't think your experience is unusual - one of the reasons I still avoid autofocus after 50 years shooting manual focus... but dSLRs all have lousy viewfinders for manual.

Dude, 50 years of photography is a long time!

What experience exactly do you mean? Slow AF?
You are right about one thing: DSLRs (at least those with crop sensors) often have lousy viewfinders. Even worse to me are the lenses however. There often is no real grip and the rotation from macro to infinity is often only 45°. They are just no longer really designed for human hands.

If your lenses depend on the screw-drive motor in the camera they will be slower than more modern designs with SDM or other focusing motors in the lens. (Not that buying new lenses is a good solution.)
And - since I didn't see it mentioned - I assume you were using the viewfinder and not live-view, which is much slower focusing.

All lenses in question have USM, SDM or HSM. I have heard however that the screwdrive in the K-3 is pretty fast just the same. In any case, what we are talking about here is not an issue with a slow focussing motor, but rather with the camera adjusting the AF several times before locking the focus.

And yes, I was using the viewfinder at all times (und thus the PDAF).

Cheers!
Chris

[monochrome]

what we are talking about here is not an issue with a slow focussing motor, but rather with the camera adjusting the AF several times before locking the focus.

It has been my long-held impression that Pentax historically intentionally programed their autofocus to behave just the way you describe - i.e. very rapid 'in-range' movement, then several micro adjutments to 'perfect' focus before the confirmation signal. The explanation for conparatively slow confirmation (or focus lock) has been higher hit rate - greater accuracy (I recall you aren't asking about accuracy though). I've also read here that Canon in particular seems to have intentionally written their AF code to beep 'when close' so it seems faster, but isn't as accurate. I ahve no way to know about Canon AF.

Given that there is a new 86,000 px RGB AF sensor in the K3 I'm not sure we have enough experience and observation yet to render a judgement for you about the K3 AF. However (and again) my experience with my lenses has been a very fast focus lock wtih no incremental adjustments, which is distinctly different from my prior Pentax cameras. I've only had the rare occasion where the camera chose the wrong direction to start, then ran the entire focus throw and back before lock - and that was using an FA50/2.8 Macro as a hand-held general 50 indoors. Not the best use for that lens.

[Pepe Guitarra]

Honestly, I would expect that Ricoh will come with some more agressive improvements, as they have done with the K3, but to get to the AF that Canon 1DIII or even 5DIII have, will take time. Even those cameras, professionals have to set the AF for the specific case they are shooting. Otherwise, the lens will hunt one way or the other.

It has been my long-held impression that Pentax historically intentionally programed their autofocus to behave just the way you describe - i.e. very rapid 'in-range' movement, then several micro adjutments to 'perfect' focus before the confirmation signal. The explanation for conparatively slow confirmation (or focus lock) has been higher hit rate - greater accuracy (I recall you aren't asking about accuracy though). I've also read here that Canon in particular seems to have intentionally written their AF code to beep 'when close' so it seems faster, but isn't as accurate. I ahve no way to know about Canon AF.

Given that there is a new 86,000 px RGB AF sensor in the K3 I'm not sure we have enough experience and observation yet to render a judgement for you about the Pentax AF. However (and again) my experience with my lenses has been a very fast focus lock wtih no incremental adjustments, which is distinctly different from my prior Pentax cameras. I've only had the rare occasion where the camera chose the wrong direction to start, then ran the entire focus throwand back before lock - and that was using an FA50/2.8 Macro as a hand-held general 50 indoors. Not the best use for that lens.

[lister6520]

Admittedly I have not tried using release priority because I have never needed it for dancing shots in the past. My EOS 5 (no 'D') only allowed that when using AF-servo (AFC in Pentax terms, I believe) and that mode really blew. The one shot setting worked fine for what I did and I kept using it with the EOS 300D. I've only had the K-3 for a little over a week now and since I'm a working guy I haven't had the chance to try out too many things. I basicly thought, what had worked with the 300D should work with the K-3 aswell.

You wrote, your Sigma lens does the same things mine does. Does that also include the long focus times? As I wrote, I took audio recordings of the focus sounds from my camera. I loaded these into Audacity and with the visualisation I could determin the time it took for the camera to lock focus (by selecting the point of the first AF sound to the end of the beep). One and a half seconds were no exception. The 300D took well unter 0.2 seconds every time - and it was never meant to be a sports camera either.

Regards,
Chris

I depends how you define focus time. It very quickly reaches good focus, I would say somewhere around a quarter of a second or less but can then spend a considerable time perfecting that focus, even well over a second sometimes when lighting is poor. In bright sunlight the initial focus takes about the same but then the microadjustments add very little or none at all to the overall focusing time.

However I must stress that even if I release the shutter as soon as the lens first reaches the 'approximate' focus the shot is still accurately focused - it is not that much an approximation. The refinement it then goes on to spend quite some time doing may or may not improve the focus enough to matter. Generally unless you pixel peep or need to crop down a lot it won't make a difference. Whenever I am shooting 'action' type photos I don't wait for the beep, instead I just shoot when the lens initially stops moving.

So in practical terms I would say focus does not take a long time but waiting for the final beep does take an awfully long time. If I were Pentax I would make an option to beep when the first focus is reached and then possibly a second different one on completing the micro adjustments.

[PixelGeek]

I depends how you define focus time. It very quickly reaches good focus, I would say somewhere around a quarter of a second or less but can then spend a considerable time perfecting that focus, even well over a second sometimes when lighting is poor. In bright sunlight the initial focus takes about the same but then the microadjustments add very little or none at all to the overall focusing time.

Well, I define focus time as the time from when I press the button to when the camera actually releases the shutter. I do not agree with your time statements however. The measurements I took were done in the early afternoon. The sky was overcast, but there was more than enough light to get the focus right. I made some audio recordings of my camera focussing during my tests. Since I can't upload mp3 (or any other audio format), I'll just add two links:
http://media.advico.de/audio/kit.mp3
http://media.advico.de/audio/sigma50.mp3

First is the kit lens: smc PENTAX-DA 18-135 mm F3,5-5,6 ED AL [IF] DC WR
Second is: Sigma 50mm F1,4 EX DG HSM
The two files are about 330KB each, so no really big deal.

As you can clearly hear, the focus takes much longer than a second - even in daylight! The microadjustments are quite numerous and there is a noticeable pause between them. This doesn't really make sense to me. A phase detection auto focus should be able to determine exactly how far the lens has to be moved to reach focus. The AF stepping motor should be able to hit that spot without the need for corrections. It remains a mystery to me, why the camera should need three or four of them.

However I must stress that even if I release the shutter as soon as the lens first reaches the 'approximate' focus the shot is still accurately focused - it is not that much an approximation. The refinement it then goes on to spend quite some time doing may or may not improve the focus enough to matter. Generally unless you pixel peep or need to crop down a lot it won't make a difference. Whenever I am shooting 'action' type photos I don't wait for the beep, instead I just shoot when the lens initially stops moving.

Although that may be a workaround, I personally wouldn't call it anything more than that, for two reasons:

• If you don't get a beep, you don't know on what the camera actually focussed, because you don't get a red light either. We have all these lovely focus points now but we are going to make our camera focus on only one in order to be sure which part of the motive is in focus?
• I actually like the focus priority. That way, I can be sure the picture has a really good chance of being in focus. Timing the shutter moment like you suggested can be hard, if you can't hear the lens move (or not). Dancing is rather noisy.

So in practical terms I would say focus does not take a long time but waiting for the final beep does take an awfully long time. If I were Pentax I would make an option to beep when the first focus is reached and then possibly a second different one on completing the micro adjustments.

They could (in practical terms) just get the focus right the first time. But since we are talking about practical things...

I did compare my K-3 to other cameras, namely the 300D and the 700D - and said that my K-3 is slower. Both are entry level DSLRs. The 700D costs less than half of what the K-3 costs. The 300D doesn't really have a price tag anymore but it's 10 years old now! Then I get answers in the forum telling me that Pentax is not as fast as an EOS 1, EOS 5 or D4. All those cameras cost more than three times as much as the K-3. Even the last post prior to yours (by Pepe) contained that reference. Do I have a (brain) malfunction because I do not get this line of argumentation?

Regards,
Chris

[JimmyDranox]

Is very clear. Your camera is not functioning normally. Sigma 50/1.4 on K-5 is much faster. Under a half of second. And I've used a K-3 at one meeting, and was at least as fast as my K-5, maybe faster. And in daylight, Sigma rarely hunts, even if is cloudy. I saw Sigma hunting like this only in night, outside, on difficult subjects.

[stevebrot]

Is very clear. Your camera is not functioning normally.

...What he said.

Your camera is still under warranty and might even be within the period of time where the dealer may allow you to return it for free exchange. Most good dealers will even eat the shipping both ways. I would contact the dealer first.

As for this thread...the more you protest, the more people are going to wonder why you don't just send it back. The longer you wait, the more it may cost you. If you don't want to keep the K-3, don't. If you do want to keep it, by all means get a replacement


Steve

[JimmyDranox]

...What he said.

Your camera is still under warranty and might even be within the period of time where the dealer may allow you to return it for free exchange. Most good dealers will even eat the shipping both ways. I would contact the dealer first.

As for this thread...the more you protest, the more people are going to wonder why you don't just send it back. The longer you wait, the more it may cost you. If you don't want to keep the K-3, don't. If you do want to keep it, by all means get a replacement


Steve

Right so. Way?

If you think that you are the lonely one, remember what has happened with Nikon D600.

[derekkite]

There are a few individuals on this forum who got duds, got a replacement, and found a massive improvement. If it isn't accurate and fast, get another one, because it is for me and many others.

[PixelGeek]

As for this thread...the more you protest, the more people are going to wonder why you don't just send it back. The longer you wait, the more it may cost you. If you don't want to keep the K-3, don't. If you do want to keep it, by all means get a replacement

Yeah, I kinda forgot to mention something. My bad. It was rather late yesterday when I wrote that post.

I have already contacted my dealer. She requested that I send the whole kit to Hamburg where one of the companies that does Pentax service in Germany is. I got it back after a little less than a week. Result: The kit lense was realigned on warranty. Nothing was done with the camera body because the technician says it's within the specs set by Pentax (Ricoh). The focus speed with the kit lens did improve quite a bit. It is now about as fast as the Sigma is. To me this is still miles away from blazingly fast - and a long way away from the speed with which the EOS 700D focuses (I'd say about a factor 3).

I called my dealer and told her as much. She said it's possible the technician made a mistake and asked me to return the camera for an exchange. The camera is currently with her and I am hoping to have the replacement by the end of the week. What I tried to get from you guys (I guess) was a clear conscience. I just wanted to have the feeling, I am not just the most ball-busting customer this shop has in 2014 but that I really do have a defective camera and a valid claim.

Sorry for being so unclear!

Cheers!
Chris

[bwDraco]

I'm echoing what others here are saying: The AF behavior you're experiencing cannot possibly be normal. This could suggest a hardware problem with the AF system or its optics, or a problem with the lens itself. I would suspect the former and would recommend that you either get the camera replaced under warranty or have the AF unit adjusted or repaired to fix the issue.

In good light, it typically does not take more than half a second for my K-5 and DA 18-135mm to acquire a focus lock. The amount of hesitation you experienced from the AF system should never occur with the K-3 under any reasonable lighting conditions, especially given that since the K-5 II, the AF algorithms have been tuned specifically to avoid this.

As for sports/action photography, I feel the K-3 has lots of untapped potential in the hardware. The 27-point AF system and 86k pixel RGB metering system should be capable of outperforming the Canon EOS 70D by a significant margin, but the AF algorithms seem to be limiting the camera's performance. C&N have well over a decade of experience designing high-performance AF systems for sports and action, but Ricoh-Pentax had only a few years of experience working on a sports-capable AF system. To drive the point home, the release notes for the last two firmware updates for the K-3 both state that improvements have been made in continuous shooting performance when using phase-detection AF.

--DragonLord

[PixelGeek]

I'm echoing what others here are saying: The AF behavior you're experiencing cannot possibly be normal. This could suggest a hardware problem with the AF system or its optics, or a problem with the lens itself. I would suspect the former and would recommend that you either get the camera replaced under warranty or have the AF unit adjusted or repaired to fix the issue.

Well, as I wrote, the camera already was with the Pentax service. They sent it the body back saying there is nothing wrong with it, only the lens was dealigned (or so they say). Currently the whole kit is with my dealer and I am hoping that she sees the situation the same way you do.

What really gets me in this discussion is the huge delta of what people are saying about my observations. While you (and others) say that there is something wrong with my camera, others in this thread seem to find that 1 second or even 1.3 seconds of focus time (which is 5 times the amount you "set") is not only normal but apparently totally acceptable. That's the reason why I kept this thread going for so long and why I felt so unsure about what to do. The response was so mixed. I'd like to know your thoughts about this (especially why the viewpoints are so different)!

BTW. A regular focus time of more than 1s would be completely unacceptable for me when I spend more than 1000 Euros on an SLR body, regardless of whether that is normal or not. In a way, I guess I have fallen in love with my K-3 a little. It feels good, it's good to operate, it feels very sturdy and has some great features. I am actually very relieved that my model is broken, because that means I can keep all those goodies without having to worry about the slow AF.

As for sports/action photography, I feel the K-3 has lots of untapped potential in the hardware. The 27-point AF system and 86k pixel RGB metering system should be capable of outperforming the Canon EOS 70D by a significant margin, but the AF algorithms seem to be limiting the camera's performance. C&N have well over a decade of experience designing high-performance AF systems for sports and action, but Ricoh-Pentax had only a few years of experience working on a sports-capable AF system. To drive the point home, the release notes for the last two firmware updates for the K-3 both state that improvements have been made in continuous shooting performance when using phase-detection AF.

I get the point you are making. Still please remember that I was comparing my K-3 to the EOS 700D, not the 70D, which is quite a different class of camera!

Having a real pro-grade Pentax (full frame competitor of the D4 or EOS 1D X) would be great. Even the mid-level SLRs from Nikon and Canon have profited from having these "big brothers". Additionally, the prestige of such a camera would really boost the sales (IMHO) of Pentax SLRs. At the dancing tournament I wrote about, I was the only one there with a Pentax (which in a way is kinda cool too) and I saw a lot of cameras there - even one jerk trying to show off his Hasselblad. YouTube is also a good indicator. If you search for the Nikon D7100 or Canon EOS 70D (which have not been around much longer), you are completely smothered by reviews and comparisions. For the K-3 I can see something like three or four real reviews (where people actually use a production model) and not a single real comparison. Boosting the sales could be a good idea. But, I'm getting side tracked here.

Once I hear from my dealer, I'll let you guys know.

Kind regards,
Chris

[cjsong79]

Hi Chris

I've been reading this thread with interest, as one day I may change my body (K-5) to a K-3 or an equivalent canon body... While I have not shot every single DSLR in the world, I have experienced both pentax and canon DSLR to a fair extent since 2006, and AF was something that i have been thinking a lot during the last few years. I will try to focus my comments purely on the aspect of AF speed on single shot mode without flash.

I think part of your frustration might be solved by starting to gather data from test sites that have been done in a controlled way (not all test site results are "useful"). I think imaging-resources, will give you valuable information on A) the speed of the camera algortihm to determine focus, and B) the pre-focus shutter-lag .... it exists in its "performance" tab. Then if you go to a website called lenstip.com, it gives you some useful information on the C) speed of the lens itself.

After going through those test data (+ reflecting with my actual experience with various pentax/canon cameras), I have made some reflections to my understanding of the two systems in terms of AF performance

(A) AF module performance : First of all, within CANON cameras, somewhere in 2009 the AF algorithm seems to have changed (and this is also seen in a few articles in the web), so that canon is performing what we call "closed loop" AF check in newer camera. I also suspect that this is only applied on the single shot AF of CANON and not on AI servo. This means that on single shot mode, the older CANON cameras may actually be faster than the newer models (though likely less accurate). On pentax side, i noticed a subtle change happening between K-5 model and models after K-30 (where it actually became a little slower since K-30), i don't know if it is linked to applying a stronger closed loop process, but K-5 had a lot of complaints on AF accuracy which was solved from K-30 and i think they are linked. Let me give you some figures....

Pentax models before K-5 : around 0.1 s (only AF module lag)
Pentax models after K-5 : around 0.1 - 0.2 s (K-3 should be around 0.15 s)
Canon earlier entry models : 0.1-0.2s
Canon recent entry models : 0.2-0.3s
Canon performance models : 0.1s (eg. 7D, 50D, 70D)

(B) Pre focus lag : This obviously applies only on situation where you have already aquired focus, but i wanted to hilight there are some meaningful performance difference on prefocus lag between pentax and canon/nikon. Pentax model are more like 0.1s, canon/nikon entry is 0.1s, but the higherend models (d7000, 70d) are more like 0.05s, this difference for me was subtly annoying on some subjects (such as kids facial expression changing quickly)

(C) Lens speed : 18-135 is one of the faster lens Pentax has, it should rack from close to infinity in around 0.7 sec (some pentax and nikon lens can take 1s). What we call relatively fast AF lens (e.g. sigma 17-50, or canon 24-105) should be able to do this in 0.5 sec. Super fast lens (e.g. canon 15-85, 85.8, 17-55, 24-70, 70-200) can do this in 0.2-0.3 sec. All in all, most of the lens that are mountable in Pentax are on average quite a bit slower than Canon lens, including not so advanced lens like 17-85, 28-135

So between K-3 + 18-135 and 300D + 28-80 (i don't know this lens), the AF performance should be faster in 300D (not by a margin of 1 sec, but maybe 0.5 sec if we are moving the whole length of the focus range). This gap will be much smaller if you have already prefocused on a similar distance (which i do all the time for pentax), and K-3 should theoreticlally be much more accurate if you have calibrated your AF microadjust.

I personally used mostly a 450D a lot during 2008-2010 with various lenses and I know it is quite a bit faster (a little with kit lens, and significantly with good lens) than K-5 for example (which i also used extensively during the last few years). The gap widens slightly further with more advanced canon models.

I think though in real-life your problem experienced might be more linked to the flash performance of Pentax system. And on daylight conditions with 18-135, i have had no problems to nail focus on fast moving kids (but i couldn't do this with slower pentax lens like 43.9 or 55-300).

My conclusion is the AF wise, canon is just quite a bit ahead of pentax. Pentax is usable but you have to adapt (e.g. prefocus frequently to nearby distance) and make sure you use above average performing lenses (not in terms of optics but in terms of AF speed). On the otherhand nothing to do with AF, but while the keeper rate is lower in pentax on the AF part, i throw away more images from canon (even 5d mk ii) on lowlight than K-5 because K-5 has a very good shadow recovery on post process (and i don't like dark pictures and i tend to lift the shadows in my PP), so the overall keeper for me is complex to determine =)

[cjsong79]

and to add one more comment, within a zoom lens the AF performance variance is worse in pentax between wide / tele end (though 18-135 is not super serious on this regards). make the test on the wider angle end or middle part of the zoom of 18-135 and the gap should be smaller. Though it doesn't solve the fact that focusing on tele end of zoom is a struggle for pentax systems.....