[Christine Tham]

Okay, here are some of the results from yesterday.

First of all, here's an image (snow leopard) that I would consider to be in focus - it was also a technically challenging one for the camera, so I am glad it nailed it:


This one's marginal, but on balance I'll consider it a pass:


This one of the binturong is also a pass:


This one is front focused (camera AF was aimed in the middle of the sun bear):


This is is back focused (camera AF was on the middle goat, but it's the goat on the right that's in focus):


This is an example where the camera got it right:


I could post more, but I wanted to give a sense that the camera focus accuracy shifts from subject to subject, and seems to depend on the subject colour (which would point to the colorimetric sensor as the culprit).

I haven't provided any examples, but I've also noticed that in a batch of photos taken in continuous mode, the focus will shift from shot to shot even though neither the camera nor the subject is moving - hence the AF mechanism in not consistent.

[Christine Tham]

Here's a link to the album on Picasa if you want to examine the images in higher resolution (up to 2048 pixels across the long edge - the highest that Google doesn't count towards "free" usage).

https://picasaweb.google.com/chris1.tham/20111228TarongaZoo?authuser=0&feat=directlink

[creampuff]

Any image - I didn't even bother to look - will only proof that the camera-lens combo works fine in combination. What does this tell about camera or lens indidiually... nothing.

Therein you're wrong. My brand new DA* 55mm f/1.4 didn't appear tack sharp with my K-5. Similarly my Sigma 70-200mm f/2.8 HSM II wasn't giving me the sharpness I expected with my K-5.
After AF calibration by Pentax Japan I had no focus problems with my camera body with different lens combinations both indoors and outdoors, both lenses included.
If don't know what you're talking about, perhaps it would be wise not to make meaningless comments.

[philbaum]

Okay, here are some of the results from yesterday.

First of all, here's an image (snow leopard) that I would consider to be in focus - it was also a technically challenging one for the camera, so I am glad it nailed it:


This one's marginal, but on balance I'll consider it a pass:


This one of the binturong is also a pass:


This one is front focused (camera AF was aimed in the middle of the sun bear):


This is is back focused (camera AF was on the middle goat, but it's the goat on the right that's in focus):


This is an example where the camera got it right:


I could post more, but I wanted to give a sense that the camera focus accuracy shifts from subject to subject, and seems to depend on the subject colour (which would point to the colorimetric sensor as the culprit).

I haven't provided any examples, but I've also noticed that in a batch of photos taken in continuous mode, the focus will shift from shot to shot even though neither the camera nor the subject is moving - hence the AF mechanism in not consistent.

Christine,
I'm not sure what you are hoping to accomplish with any of these photos. You mentioned earlier that you were using a variety of AF modes including AF multi-point. Expecting AF multi point to be consistent is not realistic, i gave up on using that a few years ago, unless its for tourist snap shots and then it doesn't matter. When i really care about taking an image for print display/selling, i work with a single af point, usually center, even calibrate for that single point, and then I can reproduce camera performance. It looks like your camera can focus in and produce sharp images, but none of us know how you are calibrating and if you're using multi points, who can read anything into these images.

Perhaps you can find someone in your area that can work with you on the calibration and use, because right now it seems like your use of the camera and expectations are inconsistent. Machines like cameras are often very consistent - we just don't understand their limitations. AF on the middle of a back bear is not a reasonable expectation, etc.

---

Last edited by philbaum; 12-29-2011 at 07:54 AM.

[jsherman999]

Christine,
...

Perhaps you can find someone in your area that can work with you on the calibration and use, because right now it seems like your use of the camera and expectations are inconsistent. Machines like cameras are often very consistent - we just don't understand their limitations. AF on the middle of a back bear is not a reasonable expectation, etc.

I was going to say - focusing in the middle of an area that has little contrast disparity is going to be hit & miss with any camera. I usually hate to say 'user error' because I think it gets used too much to cover equipment deficiencies, but in this case...


.

[Christine Tham]

AF on the middle of a back bear is not a reasonable expectation, etc.

Why? Works on every other camera I have. Used to work on the K-5.


And what's wrong with Multi-point AF? Multi-point AF on the K-5 is very useful for sports shooting (or shooting of moving subjects), because the camera can then use more than one AF point to gauge the right focus - this was one area where I was quite impressed with the K-5 when I initially bought it.

However, specifically for the images that were front focused or back focused, centre AF was used. Multi point AF was used for the photos of the snow leopard because it was moving, and as you can see those images are in focus.

If you have "given up on using multi-point AF" years ago because you found it inconsistent then I suggest you probably need to broaden your horizons. Try using multi point AF on a Canon 1D - you will be amazed.

Anyway, the real point I was making (which admittedly is not demonstrated by those images) is that the camera's focus actually changes from shot to shot in a continuous batch, which is a new behaviour, and not typical of past behaviour or behaviour of other cameras. It is partially demonstrated by the last two photos of the mountain goat - in both cases centre AF was used to target the goat, but one was focused and another was back focused.

[JinDesu]

Side note, how does multi-point AF determine what to focus on anyways?

I assume that multi-point would work on the bear if you point at the middle of the bear, but I can't see center point working on that bear no matter what camera.. that center cross-type AF point would just see a nice dull mass of black..

[Christine Tham]

that center cross-type AF point would just see a nice dull mass of black..

Bear fur isn't completely black, there's enough highlights to allow AF to work.

[Wheatfield]

Bear fur isn't completely black, there's enough highlights to allow AF to work.

But it is extremely low EV, and probably has a fair amount of red in it. You are probably invoking the well known low, warm light inaccuracy that the K5 is known for.

[Christine Tham]

You are probably invoking the well known low, warm light inaccuracy that the K5 is known for.

Yes. I was trying deliberately to see if the the focusing is affected by the colour of the subject, to see if the colorimetric sensor calibration is off.

It would seem from my photos that there are some evidence that colour affects AF accuracy - for example the mountain goat is back focused and the bear is front focused. But certainly not conclusive.

What's more disturbing is the lack of consistency across shots (not demonstrated in the examples, because I have hand picked shots from continuous batches).

[creampuff]

Poor bear! Fortunately, it's oblivious to the many AF tests done on its fur.

[Christine Tham]

Poor bear! Fortunately, it's oblivious to the many AF tests done on its fur.

This is what it thinks of people who say it's not a "reasonable expectation" of any half decent camera being able to focus on it's fur:

[Raffwal]

I think there are three factors at play here:

1) The more the industry gives us, the more we expect. We're almost all pixel-peepers (no offence!) who tend to focus on rather small differencies or problems.

2) Cameras such as K-5 are really consumer electronics of today. While we may feel we're paying a lot for the camera (I certainly do), its price in today's dollars/euros/your-favorite-currency is most likely in relative terms significantly less than a SLR in the olden days. The industry has to cut corners to stay competitive.

3) As each brand has its strengths and weaknesses, AF simply isn't a strength of Pentax. Period.

[Emacs]

I don't know whether Pentax truly suffers more than other bodies from this (although from reading many threads here and elsewhere I get the feeling that AF accuracy is not counted among the brand's advantages over other brands, to say the least) but it seems to me what you're observing is a common characteristic for phase detection AF in general.

I noticed also that the AF is much more accurate with my LX-5 point and shoot (even allowing for the much greater depth of field) than with my K-7 and K-5 -- however, it's also slower, and hunts more.
Personally I prefer accuracy to speed, and I don't care as much as many do about hunting and noise and so on -- I just don't see the point of fast AF if the camera isn't focusing correctly so much of the time -- it's like me declaring myself to be a fast, smooth hurdle jumper -- by just running parallel to the hurdles and never over them! I simply hate phase detection and wish we could have the more accurate contrast detection that LV uses, in the viewfinder. Maybe there's some other advantage to phase detection that I'm unaware of; perhaps the knowledgeable members of this forum will enlighten us.

I've basically resigned myself to stopping down often more than I want to, and manually focusing as often as is practicable (I'm totally dependent on quick shift.)
I drove myself nuts with focus adjustments for the K-7; with the K-5, I've decided to just assume that AF is not my friend, and work around it; life is too short to be obsessing over focus charts.

I suggest you to switch to sony nex either nex-7 or nex-5n with EVF.
The MF is much more accurate with it than with K-5 OVF.
In fact, I get more properly focused shots with my 43Ltd and 5N than with K-5 and this lens.

[ZizZ^]

I'm always so supriced by the amount of problems/complaints with autofocus. If you do enough research before you buy a camera you can know what to expect. At least thats what i did. And to be honest i'm more satiesfied with the K5's focus then the D7000 for example. I used the latter for 3 weeks before returning it. I got back, front, or wathever miss focus at least as much in bad light situations as with the K5. 39 nine focus points is all good and well if the technology is up to par. For me it felt like a gimmick wich i rather didnt use to get pictures focused properly. I do dance pictures with the K5 all the time in dark and moving situations. If you learn the camera and how it handles focus it's all but a matter of experience and practice to get it right.

At least for me the focus works as i expected and does a fine job in general. And i wonder why does not a single professional review compain so much about bad focus? Of course they mention less accurate/slower focus in low light but not a single one puts the autofocus system in general as a "con". Maybe because they are realistic about dslr expectations?