[derekkite]

If you set to multiple points, aim at a moving subject holding the AF button and not shooting you will see that it tracks very effectively. The red dots sparkle as it follows the subject, the AF mechanism operates keeping a sharp image in the viewfinder. Where it fails is when it loses the input from the various sensors when the mirror lifts and a shot is taken. At high speed continuous shooting about half will be soft, more of less depending on the relative movement of the subject.

I suspect the internal data communication is too slow, especially from the metering array ( also evident in the pttl delay). A fix would be to speed up the processing to allow the short sample time between shots to be adequate. Having accurate lens positioning seems to help as well.

But in the mean time giving enough time by technique can help. I'm finding that following the subject with AF enabled and shooting a series of individual shots when it is focus gives good results. It isn't smart enough or fast enough to do it by itself.

[RedBoomer]

Centre point ? You've disabled predictive AF, so as soon as you lose target or the camera/lens can't keep up with the action you will get OOF images If you used 9 point expanding you'd get predictive AF and auto handoff if you slightly lost the subject from the centre point during the pan the camera would hold target for you

Awaldam - I appreciate your knowledge on how the Pentax system works. Pentax have not done a very good job of explaining how best to use their system so information such as yours is much appreciated. It all makes sense to me as well as Derekkite's suggestion that the processing is too slow to handle hi-continuous and the focus accuracy suffers as a result.
To confirm my recommendation of SEL 4 (centre point ) - SEL4 is the expanded AF mode 27-point and I select the centrepoint mainly because it means you can switch from portrait to landscape orientation without changing the target point. I am not disabling Predictive AF but using it. SEL 4 is what appears in the top lcd when SEL-Large is chosen in the Info settings for the expanded af mode.

[awaldram]

Awaldam - I appreciate your knowledge on how the Pentax system works. Pentax have not done a very good job of explaining how best to use their system so information such as yours is much appreciated. It all makes sense to me as well as Derekkite's suggestion that the processing is too slow to handle hi-continuous and the focus accuracy suffers as a result.
To confirm my recommendation of SEL 4 (centre point ) - SEL4 is the expanded AF mode 27-point and I select the centrepoint mainly because it means you can switch from portrait to landscape orientation without changing the target point. I am not disabling Predictive AF but using it. SEL 4 is what appears in the top lcd when SEL-Large is chosen in the Info settings for the expanded af mode.

Ah sorry

I think of them as expanded 27 , expanded 9 and expanded single

You probably technically more correct sing SEL4

When you said center point I assumed ! you meant single expanded or single point serves me right for not checking SEL4

In reality you do them same as me but use 27 point, I use 9 point because it is noticeably quicker to react and hand off (less processing required), I also usually use center but sometime top middle for breast stroke to aid framing for head shot close up.

I agree with you both processing power needs to upped to cater for the modes properly but I think Pentax,Canon and Nikon are all guilty in their midsummer models of giving feature sets that the CPU can't drive fast enough to utilize fully.

Canon and Nikon probably do it to leverage upgrades and Pentax to keep price competitive, Hopefully they'll re asses 'good enough' forTheir FF offering.

To improve tracking my settings across models
k10d - never used af-C as to slow
k20d - single point af-C non tracking pan and hold
k5 - 5point AF-C tripod panning
k3 - 9 point expanding

[biz-engineer]

Any idea of what repeated moving target would be ideal for self training / understanding of the limits if the camera AF ?

[awaldram]

@ awaldram: Thanks a lot for the detailed clarification. So, if I understand correctly, since longer focal lengths yield less DoF, full frame might potentially work better for AF accuracy and tracking?

Yes as DoF is greater on FF then the phase variance would be more apparent for any give aperture , This would be the same as having a stop more light prior to blackout.

So an F4 lens would focus as per an f2.8 lens on aps-C

Once you go past the baseline of the sensor gains stop so an f2.8 lens on FF would outperform any lens available on aps-C under low light af-C

Any FF camera without f2.8 baseline sensors will not outperform a k3/k5ii with an f2.8 lens. This is irrelevant of cpu power etc as the f2.8 sensor will allow phase detection below light levels the FF could see (you can see that in the EV-3 spec)

---------- Post added 13-10-15 at 12:51 PM ----------

Any idea of what repeated moving target would be ideal for self training / understanding of the limits if the camera AF ?

I use traffic in quiet roads (busy enough not to pxss off drivers) where I know what speed they'll be doing you can get a good feel for focal length Vs AF speed Vs target size and get plenty of panning practice.

Bridges over motorways provide head on fun for long focal lengths, Also trains into stations is good for practice.


All are capable of saturating even the best tracking if you get focal lengths wrong for subject distance

[Kharan]

Yes as DoF is greater on FF then the phase variance would be more apparent for any give aperture , This would be the same as having a stop more light prior to blackout.

So an F4 lens would focus as per an f2.8 lens on aps-C

Once you go past the baseline of the sensor gains stop so an f2.8 lens on FF would outperform any lens available on aps-C under low light af-C

Any FF camera without f2.8 baseline sensors will not outperform a k3/k5ii with an f2.8 lens. This is irrelevant of cpu power etc as the f2.8 sensor will allow phase detection below light levels the FF could see (you can see that in the EV-3 spec) [COLOR="Silver"]

It works both ways. The shallower DOF of 35mm means focus errors are more evident. BUT, there's no inherent magical light advantage to FF - it just needs longer glass than crop to get the same results, which means in turn that DOF is shallower for any given identical aperture. The trade off, typically, is that 35mm cameras have more advanced focus modules right in the center of the frame (sweet spot), and better low light performance, whereas APS-C cameras are more pixel-dense and have better AF point coverage, but the outer points suffer more.

Also, a something to keep in mind is that you can't just increase aperture forever and expect better, faster and more precise AF - there's a well-known sweet spot for phase detection that lies between f/2 and f/2.8. Any larger, and there's bound to be focus errors because of the tolerances involved. This is not a big problem for Pentax, but is quite evident in some Canon and Nikon cameras, and is also part of what's wrong with the Sigma 18-35mm.

[6BQ5]

Yesterday I experimented with the three different Expanded Area AF focus modes. I enabled these modes, set my DA 18-135mm to 70mm, and panned the camera slowly around a walking trail filled with trees and bushes. I was impressed with how much more active the K-3's focusing mechanism was than the single focus point selection I was using before. The red focus indicator was indeed sparkling inside the defined square space. All of the images were at least as sharp if not sharper in some cases. I also tried all three burst rates. Low seemed more practical. Medium and High created so many images that the camera would literally freeze and stay "busy" for about 2 minutes as all the RAW image data was written to the SD cards.

My DA 18-135mm set to 70mm gives a focal length ration of f/4.5 wide open.

I tried shooting with SR on and off. Keeping it on gave more soft images than keeping it off. This sounds about right to me. I thought I read somewhere, some time ago, that SR goes crazy trying to compensate for a panning action. It shifts the sensor to compensate until the movement reaches its end. Then, it moves the sensor all the way back to the other side so it can shift back again. The result is a strange vibration.

[biz-engineer]

Yes as DoF is greater on FF then the phase variance would be more apparent for any give aperture , This would be the same as having a stop more light prior to blackout. So an F4 lens would focus as per an f2.8 lens on aps-C Once you go past the baseline of the sensor gains stop so an f2.8 lens on FF would outperform any lens available on aps-C under low light af-C Any FF camera without f2.8 baseline sensors will not outperform a k3/k5ii with an f2.8 lens. This is irrelevant of cpu power etc as the f2.8 sensor will allow phase detection below light levels the FF could see (you can see that in the EV-3 spec)

As you say it, it makes sense.

I use traffic in quiet roads (busy enough not to pxss off drivers) where I know what speed they'll be doing you can get a good feel for focal length Vs AF speed Vs target size and get plenty of panning practice. Bridges over motorways provide head on fun for long focal lengths, Also trains into stations is good for practice.

awaldram , thanks a lot for the hints. I was also thinking of using cars but I was afraid that someone would feel insecure from seeing me as a spy. But using trains from far away with a long lens, should be fine.

---------- Post added 13-10-15 at 18:06 ----------

[slip]

Interesting question. I also have a K-3, and I also asked myself how to improve AF lock and tracking. One of things I noticed is that when out of focus, the AF move into a default direction: from far to close. So, by pre-focusing at infinity (quick shift makes it easy), or just behind the target (if the target position is known in advance), the K-3 is able to lock fairly quickly without hunting, then tracks the target. The K-3 AF is hunting when the out of focus is initially in front of the target, then the AF goes forward and again backward to find the target, which takes a lot of time. That's also why I guess that Ricoh implemented a AF position memory on the DFA 150-450, for instance, to preset the AF to a certain position. I use focus preset a lot with my K-3 + DA*300 combo, it works very well.

All this time I have been prefocusing in front of the subject. I will give your method of prefocusing slightly behind the subject a try!
Thanks for the tip!

Randy

---------- Post added 10-24-15 at 10:53 AM ----------

If you are looking for a less expensive Sigma 70-200 2.8 the older APO version takes a small hit on sharpness at 2.8 but the focus is very fast even though it is a older screw type.
It is almost too fast sometimes as if the focus points even slightly fall off the subject it will immediately focus on the background. Less forgiving then a slower focusing lens like the *50-135

Randy

---------- Post added 10-24-15 at 11:16 AM ----------

After shooting some backyard birds I find using auto 9 point to be a little more effective then select center point

Randy

[Nicolas06]

Yes as DoF is greater on FF then the phase variance would be more apparent for any give aperture , This would be the same as having a stop more light prior to blackout.

So an F4 lens would focus as per an f2.8 lens on aps-C

Once you go past the baseline of the sensor gains stop so an f2.8 lens on FF would outperform any lens available on aps-C under low light af-C

I agree with you that the max you can expect depend of your AF sensor. So there 3 f/2.8 AF sensor in a K3, the other are f/5.6 and would not benefit.

Also i don't think there is any adventage for FF vs APSC for the same AF sensor. The sensor doesn't look for the full frame but a small area. So no reason to expect better performance out an f/2.8 FF sensor and an f/2.8 APSC sensor.

Last but not least, I think the performance of the lense wide open actually count, regardless of what is the widest apperture. FA50 f/1.4 is so-so focussing wise because the contrast and resolution is quite low at f/1.4 and the sensor has to work with that... You don't always get better performance with a wider apperture: I remember that the 17-70 f/4 focus much faster and accurately than the 50-135 for example. Hey they even share the same SDM motor and the 50-135 is DA* while the 17-70 is not. Things are more complex in reality.

[biz-engineer]

I agree with you that the max you can expect depend of your AF sensor. So there 3 f/2.8 AF sensor in a K3, the other are f/5.6 and would not benefit. Also i don't think there is any adventage for FF vs APSC for the same AF sensor. The sensor doesn't look for the full frame but a small area. So no reason to expect better performance out an f/2.8 FF sensor and an f/2.8 APSC sensor. Last but not least, I think the performance of the lense wide open actually count, regardless of what is the widest apperture. FA50 f/1.4 is so-so focussing wise because the contrast and resolution is quite low at f/1.4 and the sensor have to work with that... You don't always get better performance with a wider apperture: I remember that the 17-70 f/4 focus much faster and accurately than the 50-135 for example. Hey they even share the same SDM motor and the 50-135 is DA* while the 17-70 is not. Things are more complex in reality.

There are generalities that are valid for most cases, and particular cases that lead to a different results. There are always people who point out a particular case to invalidate the general rule. So, we get things like "no, FF is not better than APSC if if if if if...". Okay fine. I have not issues with depth of the field and autofocus with my Nikon Coolpix 2100 or Canon Powershot 710 IS. But when I used K200D, the image quality was so much better that there was no need to compare.

[Nicolas06]

There are generalities that are valid for most cases, and particular cases that lead to a different results. There are always people who point out a particular case to invalidate the general rule. So, we get things like "no, FF is not better than APSC if if if if if...". Okay fine. I have not issues with depth of the field and autofocus with my Nikon Coolpix 2100 or Canon Powershot 710 IS. But when I used K200D, the image quality was so much better that there was no need to compare.

I don't get your point here... You said FF is better for that... It is not, I say it and you say compact camera garbage.

I have the right to buy an FF or MF, Pentax or not and to do it for other reason than to hope that f/2.8 AF sensor will perform better on the FF, thanks. It is not because you are for FF that you have to find that EVERYTHING in an FF are better including the fact that the FF cameras are more expensive, bigger/heavier than their smaller sensor counterpart !

For me the reason to not go FF are my APSC is good enough for most things and that FF bodies do not solve my main issue: size/weight of the camera + lenses. For me the reason to go FF/MF would be slightly improved image quality in some selected cases and if money was not an issue, then I might be more interrested to try the 645Z than the Pentax FF...

[biz-engineer]

I don't get your point here... You said FF is better for that... It is not, I say it and you say compact camera garbage. I have the right to buy an FF or MF, Pentax or not and to do it for other reason than to hope that f/2.8 AF sensor will perform better on the FF, thanks. It is not because you are for FF that you have to find that EVERYTHING in an FF are better including the fact that the FF cameras are more expensive, bigger/heavier than their smaller sensor counterpart ! For me the reason to not go FF are my APSC is good enough for most things and that FF bodies do not solve my main issue: size/weight of the camera + lenses. For me the reason to go FF/MF would be slightly improved image quality in some selected cases and if money was not an issue, then I might be more interrested to try the 645Z than the Pentax FF...

Ok, my point was that AF is supposed to be better on FF, everything else being the same (which may not be, for instance difference between a 6D and 7DII). Generally speaking, including for other posts (such as pixel shift advantage over ff etc...), the "gear ratio" is not the same when we consider different formats in 1.5x sensor size increments (roughly one stop), and since lens apertures, ISO, depth of field effects, range is more than one stop (1bit, or 6dB, whatever you like), there is overlap of performance. There is a lot of IQ and AF perf overlap between APSC and FF, and there is less overlap between APSC and MF. So, anyone can always take a camera setting of two camera formats where they overlap and say "no, the smaller format can do the same" and conclude that the larger format is not worth. I read that all the time, especially from normhead and others, who are taking very peculiar cases to defend a general position (so I suppose they set their K-3 in M mode where it is as good as FF (f5.6 ISO100), and never change the setting :-) ).

[Nicolas06]

Funily on the thread on the Pentax FF, what I find most interresting is the new controls. It look like we now have 3 dials so you can control for example exposure compensation without a push of a button or control ISO + apperture + speed directly. It seems many settings can be managed easily with the new ergonomics instead of messing up with the menus. That a good thing.

I also hope that now I can switch instantly from exposure braketing to normal shooting mode... What I hate with current exposure braketing is that I can't change itt that easily.

The fully articulated screen look like a very nice feature too. I like also the possibility for 5 user modes.

[slip]

Any idea of what repeated moving target would be ideal for self training / understanding of the limits if the camera AF ?

for practice I go to local softball games. they wouldn't take the AF to the limit but they are great place to learn different techniques and there are plenty going on on summer weekends.
I use K3 with the DA200 f2.8 @3.2 with the camera set to c-focus. focus priority on the first frame and shutter priority the rest. I used expanded 9 and 27 center point and have seen a slightly better success rate with the 27. 6 frames per second helps to keep the focus on track.
once the focus is locked the camera does a good job predicting were the next shot will take place (of course I follow the lens around with the action)
success rate is high with players running from around the plates but is less effective for people running towards you

Randy