[reh321]

There appears to be a trend that any criticism of the K-1 (or even Pentax in general) is met with attacks on the messenger. I guess PF is going to turn to the DPReview norm of behavior.

Yes, I'd be happier if people simply recognized that this is the wrong test for this camera {see my comments in #18 above} and then forgot about the whole thing.

Improving AF on the K-1 would be "great", but I'm guessing that several other improvements would be "more great".

[Quark]

DPReview just posted a very negative review of the K-1 continuous autofocus. I have owned a (full frame) K-1 for 2 weeks and have owned various Pentax DSLRs over the years (also currently own K-3). I have always used single (not continuous) autofocus and generally get good focus results. In order to protest the DPR review, I just tested continuous autofocus on my K-1 as follows: Pentax 300mm f4 DA* lens with aperture set at F/4 (for shallow depth of field). K-1 in autofocus continuous mode, sel 9 (red grid), continuously shooting shutter at medium speed. The focus hold setting was set to low. First frame action was set to focus priority. Action in AF.C Cont. set to focus priority. The ISO was set at 800 and the shutter speed (aperture priority exposure) was about 1/2000 sec. Like the DPR test, I had a person ride a bike towards me at relatively low speed. I focused the grid on the rider's head, depressed the shutter button and held it down for about 20 frames as the rider moved towards me and I maintained the focus grid pointed at the rider's head. I repeated this test several times. I was shocked and disappointed to find that the rider's head was in focus in only 30% of the frames. Therefore, apparently the negative DPR autofocus test results are valid. Of course, using an aperture like F/8 would substantially increase the percentage of in-focus frames. However, that would just mask the basic functional problem.

I am very concerned about the poor performance of this important camera function and fear that it will limit the K-1 market share. I'd like to ask the following questions:

1) Does anyone have success with continuous autofocus on the K-1 or K-3? Is there some trick to making it work reliably (other than slow apertures)?

2) Does anyone know why technically the Pentax continuous autofocus performance is so poor and inferior to, say, Nikon D750? Is it due to less powerful (and less expensive) computational hardware in the K-1 compared to the D750? Is it due to inferior tracking algorithms? Is it due to the lack of some other proprietary hardware?

3) Is there anything that can be done to make the Pentax K-1 continuous autofocus more accurate and competitive?

These questions may seem simplistic but the problem is simple - the continuous autofocus does not work properly and reliably. Can we K-1 owners expect Pentax to fix this function in the K-1 or must we just except it as is? If the later, then Pentax should at least explain why and not advertise the capability (for example, claiming next generation autofocus hardware).

Other than this continuous autofocus issue, I think Pentax has created a wonderful and outstanding camera.

TEST CONDITIONS I did some further tests of the K-1 continuous autofocus tracking of a moving bicycle subject. There were three cases: 1) Full Frame, Select-9 AF mode; 2) Full Frame, Auto-9 AF mode; 3) APS-C Crop mode, Select-9 AF mode. Common settings for all three tests: DA*300/4 at F/4, DNG file format (no JPEG files), SR off (not that it matters), hand held, shutter about 1/2000 sec, AF hold off, composition level adjustment off, medium speed burst mode shutter, first frame action in Cont C set to focus priority (i.e. system locks onto focus before firing shutter), Action in AF.C Cont set to focus priority (i.e. system does not fire shutter unless it registers focus lock).

TEST RESULTS 1) FF, Sel-9: 21 images, 6 in focus, 15 out of focus; 2) FF, Auto-9: 26 images, 8 in focus, 18 out of focus. In both cases there tends to be a run of in-focus images (2 to 6) and then strings of out of focus. 3) APS-C, Sel-9: 24 images all in focus. There is a completely different feel to the tracking and shutter action in APS-C mode versus FF mode. In APS-C mode, it feels like the camera will go on forever (or at least as long as data buffer capacity is available) staying locked onto focus, firing the shutter and delivering 100% in focus images as it tracks the rider.

Some have asked for image files with EXIF data attached. I will include some in a separate post (I need to go to a different computer to get the files).

In another separate post I will describe what I thinking is going on technically. In summary, I think the K-1 optical and sensor hardware is fine to support reliable continuous focus tracking.

[Nicolas06]

Yes, I'd be happier if people simply recognized that this is the wrong test for this camera {see my comments in #18 above} and then forgot about the whole thing.

Improving AF on the K-1 would be "great", but I'm guessing that several other improvements would be "more great".

To me even if it is not my priority, it is still one of key features to improve for the FF.

[David&karen]

Thank you all and esp. Quark for all the time and effort involved to do this testing, looking at picture data, and brain storming to figure this out. We had a thought, which we think will not help, but we saw somewhere that someone posted that they put the AFc mode into FPS priority instead of Focus priority and that increased the in focus hit rate buy like a factor of 2 or more. Personally it goes against logic that it would work, but has anyone here tried it?

David

[Nicolas06]

Thank you all and esp. Quark for all the time and effort involved to do this testing, looking at picture data, and brain storming to figure this out. We had a thought, which we think will not help, but we saw somewhere that someone posted that they put the AFc mode into FPS priority instead of Focus priority and that increased the in focus hit rate buy like a factor of 2 or more. Personally it goes against logic that it would work, but has anyone here tried it?

David

If I understood it well, it depends the kind of subject you have.

[David&karen]

Frankly, I don't understand it at all, that's why I asked; it probably will not help but??

David

[monochrome]

To me even if it is not my priority, it is still one of key features to improve for the FF.

To me it is like video. I guess it should be better because somebody seems to demand it, even if it is only reviewers who do. But I won't use it much either way.

It's as if they think people can't weigh alternatives and make decisions. If video on a FF dSLR was important I'd sell this Pentax stuff and get some kind of 5D and a couple lenses. If it was really that important to me to have a fast tracking AF.C FF camera I'd sell all this stuff and get a D750 and a couple fast zooms. That's why I'm so frustrated by this whole thing. The AF is not Poor. The K-1 doesn't seem to do predictive z-axis continuous autofocus very well. So noted. Thank you DPR.

[Nicolas06]

To me it is like video. I guess it should be better because somebody seems to demand it, even if it is only reviewers who do. But I won't use it much either way.

It's as if they think people can't weigh alternatives and make decisions. If video on a FF dSLR was important I'd sell this Pentax stuff and get some kind of 5D and a couple lenses. If it was really that important to me to have a fast tracking AF.C FF camera I'd sell all this stuff and get a D750 and a couple fast zooms. That's why I'm so frustrated by this whole thing. The AF is not Poor. The K-1 doesn't seem to do predictive z-axis continuous autofocus very well. So noted. Thank you DPR.

I agree the AF is not poor, it is more than adequate for most use (that was not the case for me for K5)? It is not the best sure and you have to know if it is important for you or not. I think for most people it is not, but I am not everybody.

[monochrome]

I agree the AF is not poor, it is more than adequate for most use (that was not the case for me for K5)? It is not the best sure and you have to know if it is important for you or not. I think for most people it is not, but I am not everybody.

i didn't have a K-5 but I recall people saying the center point was the size of a dinner plate

[Nicolas06]

i didn't have a K-5 but I recall people saying the center point was the size of a dinner plate

I had K5, I have K3. The center AF point are the same size to me on both. In both body, the center AF point is a centered cross with horizontal and vertical pattern that detect things up to the border of the "( )" drawing.

The difference is on K3 the center AF point is truely the biggest one. The off center AF point are smaller. In practice this make K3 a bit harder to lock focus on something (a tiny bit...) but when the AF point take the focus, because it is small it is very likely what you see in your viewfinder arround the red dot.

On K5 first version, not only was the AF quite poor in low light overall, but the off center AF point were huge. If you were using one of theses to pick up the eye for example it could actually pick the nose or the ear so you had to be able to check the focus by yourself on the viewfinder and fine tune it as necessary or to use a quite narrow appear so that it would not really matter.

I brought a K3 when I understood why I failed my AF shallow dof shoot. I didn't want to have to rely on MF as typically it is slower and in my case I don't see well enough using the OVF to understand where the focus really is.

You can actually check how big an AF sensor is by yourself. You just need to do single point focus, set a wrong focus position and try to focus on an uniform subject like a white wall and move slowly to a single constrasty object (like a straight line). You now you reached the border of the focus point when the camera start to refocus properly.

[Rupert]

Lots of comment on the AF system....I am not a sports or action shooter so I have no useful input and trust others are more qualified in this area to make a judgment.

I can tell my personal story, which is very important and meaningful to me. I mainly shoot with a 50-500 Bigma...currently the latest HSM OS version. I have been a Bigma shooter for around 10 years. In that time I have gone from the DS to most of the other flagship models and now to the K1.

All previous models were a nightmare at AF in one way or another. Gaining focus lock in a crowded scene with a bird/squirrel in the brush was near impossible. Even in good conditions, focus lock was slow and the big ol' slow Bigma might hunt forever before locking focus...if it ever did. Many hundreds of shots were missed....maybe thousands over the years. I chalked it up to the Bigma being long and slow...even the new model exhibited the same problems unless in very good light.

So I get my new K1 and immediately mount my most used lens...the Bigma. I set it to Cop Mode for the best advantage at long shooting.....most is at 500mm. What happened next was simply amazing to me. It locked focus instantly, in every circumstance, almost every time and even in very low light. I found it unbelievable, but it was real and it still is. It does no hunting...none, it locks on the subject and you have your shot immediately.

Birds are mostly jittery, they don't linger as you take a coffee break while your Bigma gains focus lock. Case in point here...this Jay showed up and I knew from his appearance he was not going to stick around for more than a few seconds. The Bigma was mounted and ready on my desk, but in the past I would not have bothered ......knowing he would be gone before I could lock on focus. Not so with the K1...I picked it up, turned it on, focus was instantaneous and I fired.


So for me the K1 is a dream come true. It does everything I had only dreamed of and a lot more. My camera is for me and my use as I use it......I am fully and completely satisfied with the K1 and the AF system....as well as the SR, WB and wonderful color rendering.

My regrets to those that are not, but if you are a mostly a still shooter this camera is a Pentax miracle.

Old Norm once wrote an article of education regarding the AF systems on various cameras. The bottom line was that you can get most anything you want if you are willing to pay the price. At the price of the K1 we are getting a tremendous bargain.....you want more......open your wallet and spend the bucks, it's all out there but not at a K1 price.

Best Regards!

[Winder]

The K-1 has pretty average C-AF. It is on par with mirrorless cameras like the Sony A7II. My D-FA* 70-200mm is rather slow to focus and I get rather disappointing results. The Sigma 85mm F/1.4 is surprisingly fast and I get much better results. I would expect that a lens like the 28-105mm would perform rather well. Pentax is still 2 generations behind the competition when it comes to C-AF. I would say that it is equal to the Canon 5DII and faster than the Sony A900, but its not going to compete against the D750 when it comes to C-AF. If you are shooting something that is moving fast, then the K-1 is not going to keep you happy. The AF is however more than capable of professional level work for studio, landscape, architecture, portraiture, and weddings.

[Nicolas06]

So I get my new K1 and immediately mount my most used lens...the Bigma. I set it to Cop Mode for the best advantage at long shooting.....most is at 500mm. What happened next was simply amazing to me. It locked focus instantly, in every circumstance, almost every time and even in very low light. I found it unbelievable, but it was real and it still is. It does no hunting...none, it locks on the subject and you have your shot immediately.

I am pretty sure most of that you experienced was already available in the K3 then K3-II. You only had a K5-II before that had same AF as K5, only better in low light and a bit more precise for 3 AF point for f/2.8 lenses.

When I gone from K5 to K3, it was really night and day for AF performance.

I am sure the K1 is better. A bit on AF, and much better for high iso and packed with features but just to say that K1 was not the first camera with acceptable AF, K3 started it 3 years ago

[Quark]

TEST CONDITIONS I did some further tests of the K-1 continuous autofocus tracking of a moving bicycle subject. There were three cases: 1) Full Frame, Select-9 AF mode; 2) Full Frame, Auto-9 AF mode; 3) APS-C Crop mode, Select-9 AF mode. Common settings for all three tests: DA*300/4 at F/4, DNG file format (no JPEG files), SR off (not that it matters), hand held, shutter about 1/2000 sec, AF hold off, composition level adjustment off, medium speed burst mode shutter, first frame action in Cont C set to focus priority (i.e. system locks onto focus before firing shutter), Action in AF.C Cont set to focus priority (i.e. system does not fire shutter unless it registers focus lock).

TEST RESULTS 1) FF, Sel-9: 21 images, 6 in focus, 15 out of focus; 2) FF, Auto-9: 26 images, 8 in focus, 18 out of focus. In both cases there tends to be a run of in-focus images (2 to 6) and then strings of out of focus. 3) APS-C, Sel-9: 24 images all in focus. There is a completely different feel to the tracking and shutter action in APS-C mode versus FF mode. In APS-C mode, it feels like the camera will go on forever (or at least as long as data buffer capacity is available) staying locked onto focus, firing the shutter and delivering 100% in focus images as it tracks the rider.

Some have asked for image files with EXIF data attached. I will include some in a separate post (I need to go to a different computer to get the files).

In another separate post I will describe what I thinking is going on technically. In summary, I think the K-1 optical and sensor hardware is fine to support reliable continuous focus tracking.

Under the test conditions in the referenced post, the Pentax K-1 performs highly accurate continuous autofocus of moving targets (bicycle rider) in the APS-C crop mode but less so in Full Frame. Being uninformed about the engineering design details of the K-1, I can only speculate as to the cause of this result. It appears that the sensor hardware including the autofocus phase detection system and the “Pentax Real Time Scene Analysis System with 86,000 pixel RGB light metering sensor” is fully capable of providing accurate autofocus of moving subjects. Problems seem to arise when capturing and storing full-frame image sensor data in rapid bursts (say 1 or 2 frames per second). Apparently the digital computer logic becomes a bit overwhelmed with the data rate and gives priority to the image sensor data.
As a practical matter, if there are problems autofocusing on moving targets with the K-1, one alternative is to switch to APS-C crop mode and work with the smaller field-of-view.
If Pentax wishes to improve the continuous autofocusing performance of the K-1 in full frame, they could split the digital data processing functions into two CPUs, one handling exposure and autofocus sensor data while the other handles image sensor data. Perhaps this already is the architecture and the control CPU just needs to be faster. Powerful ARM processors, used for example in smart phones, cost under $10.
In summary, I believe the Pentax K-1 autofocus related hardware is fully capable of DSLR state-of-the-art performance and just needs faster control data processing to reach full potential.

[Nicolas06]

In summary, I believe the Pentax K-1 autofocus related hardware is fully capable of DSLR state-of-the-art performance and just needs faster control data processing to reach full potential.

Fully capable? Maybe. State of the art? No 8-9 year ago Nikon had already 50+ point sensor AF module and the latest greatest is 150+ AF points. There also mirorless body with AF point all over the frame.

K1 AF is good enough. While not its strength it is not an issue for most and if they are not into action/sport they don't really have to care.

Anyway, if the K1 only perform well in APSC mode this is problem, it now need to be compared to a D500 rather than D750 / D810 ...