[Tommot1965]

That's true and this is why I write my report. The whole thing is a bit complicated (no simple truth here after 250 test shots) which is why it takes me a couple days to write it up. Actually, the focus problems are neither erratic nor following a very simple rule. However, a fuzzy rule exists which I am writing down and which may help to avoid the problem.

Im looking forward to seeing your report....

[Christine Tham]

That's true and this is why I write my report. The whole thing is a bit complicated (no simple truth here after 250 test shots) which is why it takes me a couple days to write it up. Actually, the focus problems are neither erratic nor following a very simple rule. However, a fuzzy rule exists which I am writing down and which may help to avoid the problem.

Thanks, Falk, looking forward to reading your report.

For me, the problem is so slight (in the order of a few mm) and occurs under very specific conditions (and on specific lenses), which I have yet to encounter in actual shooting.

My rule would be - if there's any doubt, shoot using Live View, or somehow force AF assist light on (I noticed that once the light comes on, it will stay on for the next few shots around EV1-3 where the problem starts manifesting itself).

[Tommot1965]

It would be if the aperture was set to 1.4 or 1.2

Anyone can take a pic with a smaller aperture and wider DOF and claim there isn't a problem.

That pic is about EV3 and it's front focussing. lets see the same shot at EV2 or below at f/1.4

+1......

but it does show a work around for now.....not that helps much, but it does make the K5 usable in these Low light conditions

[Tommot1965]

Thanks, Falk, looking forward to reading your report.

For me, the problem is so slight (in the order of a few mm) and occurs under very specific conditions (and on specific lenses), which I have yet to encounter in actual shooting.

My rule would be - if there's any doubt, shoot using Live View, or somehow force AF assist light on (I noticed that once the light comes on, it will stay on for the next few shots around EV1-3 where the problem starts manifesting itself).

yes , for me that would work...I was test shooting the other night, and the AF assist lamp illuminated a car number plate at around 15 metres from the camera....so that quite a distance for the AF lamp to give the Phase detect AF a chance of correct focus

[Christine Tham]

Christine, I think all the uproar about this is that the K7 does not suffer from FF at all, its '+' sensor doing its job perfectly, even with the AF assist lamp disabled. I know, I had one, as I had a K10, K20 and Kx (and they suffered from FF).
And you can search here, there was no such report at its time...

Again, I think that's anecdotal evidence rather than established fact.

There are a few people here claiming their K20D focuses perfectly in low light.

And I seem to recall when the K7 came out there were more than a few reports of focusing issues.

My take on it is that I don't expect ANY camera to focus perfectly all the time - none of mine behave flawlessly under all conditions. If you do have one, keep it and treasure it always!

The K-5 seems to have a different focusing algorithm compared to earlier Pentax cameras - it behaves quite differently from my K10D. As I mentioned before, it always seem to favour the closest object it can find in its sensor whereas the K10D tends to hunt around.

Which would mean the K-5 would APPEAR to be focusing further in front compared to previous cameras. I don't think this is a flaw - merely a characteristic of the focusing algorithm. I'm not saying that a slight FF in low light is correct behaviour - that is clearly a flaw in the algorithm.

However, under typical shooting conditions, the K-5 focuses very predictably and consistently, similar to the Canon EOS 1D Mk III which is good from my perspective - I prefer fast and consistent over hunting around.

[jolepp]

Enclosed is a real life snap that is intented to demonstrate the problem I'm having in practice. As "evidence" of a problem with my K-5 this is not worth much beginning with the fact that lens is a Tamron 17-50mm which needs a -10 adjustment to focus good with my K-5 in daylight and could maybe do with an extra click or two.

The AF had to do with tungsten from a bulb hanging from the ceiling while the image sensor had a blast from the then new to me AF280T bounced from the ceiling. I used center point AF on the eyes and then recomposed at a distance of around 1.2m (I was sitting on the other side of the table there while the little guy was having some chocolate cake). This has been cropped in an attempt to make the FF present more evident.

Again, this is merely to demonstrate the kind of thing that got me suspecting that there might be something amiss with my K-5 as opposed to proving anything. With a real life snap like this too many incidental things can affect the result, however, I have deleted loads of the likes of this .

[edit: added a resized version of the entire image as attachment #2]

[Christine Tham]

With a real life snap like this too many incidental things can affect the result, however, I have deleted loads of the likes of this .

LOL - I have deleted loads of pictures like that from my K10D as well. Very few from my K-5 though.

You are right - too many potential issues for this to be conclusive (same as for a lot of other photos demonstrating "problems" or "no problems" in this thread).

Given that Falk has previously shown us how large the center AF sensor is (I can't remember the diagram he linked to, but the center AF sensor area is almost 1/4 of the screen) it is quite likely the sensor has focused on the spoon even though you were aiming at the eyes.

Remember the K-5 will prefer to focus on the closest object within the sensor hairlines, not necessarily the object at the notional "point" as depicted on the screen.

This is exactly what a pro camera like the EOS 1D will also do - focusing on the closest object is by design (and I seem to recall stated as a design objective in the manual) - because the camera is guessing the object you really want is the closest, and it may have moved whilst the camera was trying to focus so it's not really the object you are pointing at.

[jolepp]

LOL - I have deleted loads of pictures like that from my K10D as well. Very few from my K-5 though.
...
it is quite likely the sensor has focused on the spoon even though you were aiming at the eyes.
...

I guess I was trying to say that this is the kind of result I tend to get in lowish tungsten light (with the Tamron 17-50 1:2.8, which seems to work well enough in daylight and was pretty ace with the K-x even in tungsten light). I don't think the center sensor area is quite large enough for the camera to have legitimately focused on the spoon/hand as it in effect has; I added a downsized version of the full image to the original post above to give a better idea of this. This particular image is nice for demonstration purposes as there happens to be something in front of the face instead of the typical thin air. Actually, "illustration" might be a better word: whatever went wrong with this particular one is not so important, the point is demonstrating/illustrating the problem I have in general. For this I might as well have drawn a picture with the little guy's Crayons - if was good enough with that kind of thing, which I'm not .

[Smeggypants]

Remember the K-5 will prefer to focus on the closest object within the sensor hairlines, not necessarily the object at the notional "point" as depicted on the screen.

Will it? Where is this reliably documented?


And in any case given the same subject with two different light levels the distance of various objects in the scene is irrelevant.

IOW, if your questionable statemnt above is true then the "K-5 will prefer to focus on the closest object within the sensor hairlines, not necessarily the object at the notional "point" as depicted on the screen." at normal light levels too, and seems to focus on what is at the centre of the red square.

[GregK8]

Out of curiosity, what happens if you use one of the other AF points instead of the center, say the upper one. I'm not suggesting a work-around, just wondering if the phenomena is affected by which sensor is used.

[Bramela]

It would be if the aperture was set to 1.4 or 1.2

Anyone can take a pic with a smaller aperture and wider DOF and claim there isn't a problem.

That pic is about EV3 and it's front focussing. lets see the same shot at EV2 or below at f/1.4

This shot of same subject at 1.4 (widest I have.) Focussed on word "inside" again. FF????????

K5/M50 1.4. 30thsec/f1.4/iso800 Handheld camera.

[GregK8]

This shot of same subject at 1.4 (widest I have.) Focussed on word "inside" again. FF????????

K5/M50 1.4. 30thsec/f1.4/iso800 Handheld camera.

Well, someone is going to point out that this shot is at EV3 not 1 or 2.

[jolepp]

Out of curiosity, what happens if you use one of the other AF points instead of the center, say the upper one. I'm not suggesting a work-around, just wondering if the phenomena is affected by which sensor is used.

I haven't tried that, ... it would be interesting if it did make a difference. Do you have a particular idea as to why it might ?

[Christine Tham]

I don't think the center sensor area is quite large enough for the camera to have legitimately focused on the spoon/hand as it in effect has

Well, this diagram illustrates the size of the centre sensor (and all the other sensors) - courtesy of Falk I think:

[GregK8]

I haven't tried that, ... it would be interesting if it did make a difference. Do you have a particular idea as to why it might ?

Well, if we are supposing the center AF point is so big that you *may* have focused on the spoon even though you said you focused on the eyes, I was thinking that if we took the center sensor out of play it might be easier to determine exactly what is being focused on.

** I am qualifying the *may* because you did not agree to it, and it is just one theory among many.