[magkelly]

You've got 3 different cameras and supposedly the same problem on two of them. I have a K-x and I've borrowed a K-r and a K-5 to play with a couple of times. My K-x is an awesome little camera, love it to pieces, but the AF on the K-r is a bit better and the AF on the K-5 just completely blows either of the other two away. AF in low light is a problem for any camera. Even the best pro cameras will blip a bit if there's not enough light for the camera to AF well. It's a fact of photographic life. Low light means you really have to work for a shot, use a really fast lens, and maybe even MANUALLY focus. But FYI, my teacher's 2 K-5's get better shots in low light than any cameras I've ever seen and btw he has several Nikon and Canon cameras as well. If you're using the same lenses on more than one body I'd suggest you'd look to the lenses you're using more for answers rather than the bodies. Or to your technique because the K-5 actually IS usually capable of quite a bit in low light. I'm not saying yours couldn't have an issue, but I seriously doubt that both a K-5 and a K-r would both be giving you that much difficulty if perhaps there wasn't something else going on or that the K-x would be beating them both at it. That just doesn't make sense. The sensors in the K-r and K-5 are actually better than in the K-x supposedly.

[Zafar Iqbal]

K-5 beats the shit out of K-x i low light. No doubt about it. My current issue is with focusing under normal/bright (sunny) light.

I mentioned earlier a success rate at about 50%. I was out last evening again. Lower light, but not "low" except for the very end of the walk.

My success rate then was noticeably higher than 50% using 2 of the same lenses I've always been using - *and* with micro adjustment set to 0. Most of the pics with the 50% success rate had -9 dialed in as it gave better focus.

It's a highly confusing matter.

[Anvh]

And funny enough you're the first one i hear the lowlight works great but the daylight focus is way off XD

[Zafar Iqbal]

And funny enough you're the first one i hear the lowlight works great but the daylight focus is way off XD

I could focus in virtually pitch dark (through the VF) with K-5. Never could, never can with K-x. K-5 could however not focus in *brighter* (but still low light) artificially lit conditions, whereas K-x would nail it every single time and ask for more.

It was looked at under the warrantee so that should make it clear it did suffer from a problem. It's only unfortunate the camera then because useless during day time.

The issue was more complicated (and confusing) than natural vs. artificial light.

[TomTextura]

Sorry you're experiencing so many problems. I'm not very impressed with my K-5's AF performance but not all that too frustrated either. I have to dial in a +6 micro adjustment to my FA77 for it to be reliable. The rest of my lenses do okay without any micro adjustment. I do hope that the next AF iteration from Pentax is better by a good amount but I know that even then it won't be perfect.

That said, it sounds to me like you need to employ a much more controlled testing methodology. I suspect that photo walks with settings being regularly changed won't be adequate for getting to the bottom of the problem. I recommend getting a focusing chart, using a tripod, and keeping detailed notes of all the setting changes you make so that when you go to look at the pics on your monitor for analysis you can be certain what was happening at the time of the shot. If you need different lighting conditions and colors, fit that in to your test but stick with using a tripod and keeping detailed notes. Once you've run your camera and lenses through controlled focus testing and dialed in what you find to be the most effective adjustments, then you can go out on a photo walk to see how it performs in the real world with the new adjustments. Keep your test and real world shots separate though.

Definitely avoid focus and recompose as it's pretty much a flawed technique. Even when applying little tricks to make it work better you are still leaving yourself open to a high possibility of getting the shot out of focus.

I'd recommend using the Selective AF Point setting. The far right and left points are the least reliable but that still leaves you with 8 points other than the center one to choose from. If what you really want the focus to be on can't be lined up with one of those AF points then you'll have to rely on MF or take your chances with focus and recompose. I leave my camera set on Selective focus almost all the time and have the best luck with it. I'm still working on my manual focusing skills because I don't always want to compose my shot based on the available AF points and because I too end up with missed focus more often than I would like.

[TomTextura]

Another thing... It's my opinion that wide apertures are not much good if you don't have a fairly close subject that you want to separate from the background, or conversely, a moderately distant subject that you want separated from the foreground. Wide apertures are also good if you are trying for an artistic narrow DOF but this will still be at relatively close distances. If you are just capturing a scene without a specific subject that fills a fair amount of the frame then you will pretty much always want to stop down to get sharper images. If you aren't going for subject isolation or working in really low light conditions, wide apertures aren't going to do you any favors when it comes to getting sharp focus.

[Marc Sabatella]

Yet I think it pretty clear the 3rd pic hows the focus is in the air somewhere between the street light and bus stop

The scene tell is so complex, there are just way too many variables to know what happened. That's what I mean about a basically flawed methodology. There is simply no way to know where the camera tried to focus, and hence no way to evaluate the results.

The K-5 was back focusing by so much, that I couldn't bring the focus in front of the building or even just right on it

Absolute nonsense. Again, the effect of focus adjustment is to move the focal plane by a matter of millimeters - maybe an inch or two t that distance. But the DOF would completely cover that effect. If an object at that distance is out of focus, it is not and can never be due to any sort of focus error that adjustment can fix - it is simply a case of the camera not reading your mind about where you *wanted* it to focus. Again, completely flawed testing methodology. Any change you saw after focus adjustment is not evidence of the success or lack thereof of your adjustment - it is evidence of how poorly controlled your tests were.

Basically, you realy need to come to grips with how to control where the camera is trying to focus - merely pointing in the general direction of something is never a guarantee with any camera. Sometimes the camera guesses right, sometimes not. Such is life with autofocus.

You also need to come to grips with DOF and field curvature, as these two effects explain why your efforts at focusing and decomposing are not always successful. Again, not due to any flaw in the camera, but just simple physics.

[Williunck]

Marc, as a moderator it seems like you could be a lot more helpful and a lot less condescending. He asked for advice on how you would run the tests which you still haven't given him in 2 posts now. Tom managed to make a very helpfull post right before you explaining a more controlled test, and he wasn't negative even once. The only thing I get from your post is that you like to tell people how fail You think they are.

[stevebrot]

To add to Marc's comments on this thread, it always amazes me how often someone will expect the camera to "know" what element of a complex scene is supposed to be in perfect focus. As Marc noted, the focus "patch" is fairly large on most cameras and focus is determined by peak contrast within that "patch". It is fairly likely that the camera will miss focus on a subject with complex depth. This is particular true for oblique portraits at wider apertures. Why blame the camera when focus fell to to the trailing edge of the eyebrow rather than the leading edge?

The short answer is to do what the pros do and not expect miracles from your camera. If you want critically accurate focus, use manual focus with an aftermarket focus aid.

As for focus tests...here is my standard advice:

• Flat target with high degree of detail and high contrast (newspaper classified glued to masonite will do)
• Camera on tripod with sensor/film plane parallel to target
• Do exposure using 2s delay to lock up meter.
• Natural light
• SR OFF

Failure to attain consistently accurate focus with this setup pretty much indicates a hardware problem.


Steve


(FWIW...complaints about AF accuracy are pretty much universal across brands...)

[lammie200]

...If you want critically accurate focus, use manual focus with an aftermarket focus aid...

I believe that this may be a reason why Pentax has incorporated quick shift on at least some of its lenses. If you need to touch up AF'ing the quick shift feature would be handy. Too bad it isn't incorporated on all Pentax lenses.

...(FWIW...complaints about AF accuracy are pretty much universal across brands...)

Seems that way, doesn't it?

[Anvh]

I believe that this may be a reason why Pentax has incorporated quick shift on at least some of its lenses. If you need to touch up AF'ing the quick shift feature would be handy. Too bad it isn't incorporated on all Pentax lenses.

Only the budget lenses are missing them and the FA limiteds.

[Zafar Iqbal]

TomTextura: thank you for your input. I'd very much like to do controlled tests but it all seems so weird. The evening yesterday and today during the day have been a much better than previously (with K-r) - and I still pretty much photographed same kind of stuff. Street lights, signs etc.

Since it seemed to be a more reliable day today, I tested the thing about AF being confused. I'm am convinced it is not *the* issue, but can accept it may have sabotaged some of the photographs I've taken previously.

Todays photos:
Street lights were a hell even at a day like this.

Marc, Please see the following (preferably in larger sizes):

Focus on the tree. Micro Adjustment turned off.


Micro Adjustment @ -10


Micro Adjustment @ +10


Yes, the distance might be closer than the pics I refereed to and yes, it's not the K-5. Although I do not know this as a fact, I'm assuming they work similar or close enough. I'm sticking with my original claim.

And regarding recomposing:
Principally it's a issue, yes - but everything is again relative.

Focused and locked it (AF switch turned to MF)

Recomposed:


Recomposed back to center so you can get an idea of how my the cam was moved (or not) since 5 photos ago.


Technically an issue? Yes. Practically in my case? No.

Steve: I'll look for a flat target .

Edit: btw, I don't blindly focus a single time and shoot. I work with it. If I can't get focus on the center on something then I try aiming for the edge. If one side doesn't work then the other. Plenty of stuff can be tricky to get in focus, but doable with that method. but not all the time, I know that.

[Ahab]

I guess I would send the camera in if you are certain there is no user error. I have no trouble focusing with my K5s, except with wide angle lenses where my subject is positioned pretty close to the back ground. I find the center auto focus point is the one that struggles the most -- it is just bigger and tends to grab onto whatever is closest and most contrasty in the general area. Edge points are smaller and seem not to focus on extraneous stuff as much.

Yah, same here. You hit the nail on the head.

[Zafar Iqbal]

Just to clarify - I do not have the K-5 right now. It's in for repairs.

I have the K-r instead - is the center point as large as K-5's? Doesn't feel too different than my K-x tbh.

[normhead]

Gee I hate that term.."sick and tired". My mom always declared she was sick and tired right before she slapped me upside the head. Hence I've been avoiding this thread like the plague.