[Mister Guy]

I asked this on a different board and didn't really come up with a satisfactory answer. I'm trying to figure out if the problem I've been having is lens specific or if it's my camera.

Can anyone who has a better understand of what causes back focusing problems explain to me if it's possible for my K110D to correctly indicate focus on my two manual lenses, but seem to skew backwards on my two auto-focus lenses? I just did a quick test, and it seems like I can do a better job by hand than my autofocus does. Actually, after typing that sentence it occurred to me to compare the same lens in manual and automatic focusing modes.

Not sure it'll really come across when resized for the forums, top is automatic, bottom manually focused:






I got a little paranoid when I was going through pictures from a recent shoot and was just convinced I COULDN'T have screwed up the eyes that much on my own. I realized in several of the pictures that I was SURE I had recomposed properly that the eyes weren't clear but the necklace around her neck was tack sharp, so I went and grabbed the backfocus test chart from Nikon D70 Focus Chart and did the test as carefully as I could with multiple lenses and realized the results were inconclusive. Then I realized it seems like my old primes were focusing fine but my two zooms weren't and I was paranoid I was doing the test wrong. So I compared manual and automatic on the same lens, and found that if I used the focus indicator manually focusing my lens, it was right, and autofocus wasn't, which boggles my mind a little bit. I also can't tell if it's JUST the Sigma lens messing up or if it's that the kit lens doesn't have a narrow enough depth of field wide open to be as noticeable. In order to try and conclusively test if I was messing up the execution of the test, I also tested with a different chart I found with the round light/dark oval in the middle with a level camera with the chart set up on a wedge. The results were even more noticeable there but I don't have those images handy to show.

So that brings me to the question: How is it possible for the camera to "know" what focus is right, and then set the automatic lens incorrectly? I don't understand what is really CAUSING it, and what I need to try and get fixed.

[BCtoad]

That will all go away with a K20D......

[Mister Guy]

That will all go away with a K20D......

To be honest, debating talking myself into a K20D has been the reason I haven't pursued it more, but since I've somewhat resigned myself to stalling around to find out what's coming next, I figured I'd ask whether it's my camera or if I should just be sending the lens off repairs.

[Fritz]

What lens were you using, and what f stop? I can't seem to read the exif from your posted pictures.

With my k100d it back focused with fast lenses shot at max aperture in bad light. I adjusted the camera a little, but it seems just to be an issue with the above combination of bad light and wide open shooting. Focus charts will drive you to insanity.

[audiobomber]

With my k100d it back focused with fast lenses shot at max aperture in bad light. I adjusted the camera a little, but it seems just to be an issue with the above combination of bad light and wide open shooting. Focus charts will drive you to insanity.

This is pretty much what I found, except only under Tungsten light. All apertures and all lenses were fine at any aperture in daylight. With apertures of 1.4 to 2.8 under bright or dim tungsten light, I measured front focussing. These results were consistent across three primes and two cameras (K100DS and K20D).

[Fritz]

This is pretty much what I found, except only under Tungsten light. All apertures and all lenses were fine at any aperture in daylight. With apertures of 1.4 to 2.8 under bright or dim tungsten light, I measured front focussing. These results were consistent across three primes and two cameras (K100DS and K20D).

Yes, exactly my experience too. Thanks for clarifying. Additionally, if you adjust the camera to focus on the chart correctly under tungsten light and wide open aperture, it will not focus correctly in bright sunlight. Fun, fun. The workable solution for me has been just to touch up the focus a little manually if needed. I've only had my k20d for a short time, but from what I can tell so far the focusing is better then my k100d.

[Mister Guy]

That may be, but it still drives me nuts that if I'm personally turning the lens to focus using the green hexagon, the camera knows how to focus. Yet, the same lens, same light, same aperture misses doing the same thing automatically.

[PentaxPoke]

Note that when AF is working, the lens is at maximum aperture until the exposure is taken (unless you are using certain older style manual lenses.) Therefore, when your exposure is set to wider apertures, you are actually seeing closer to what the AF sensor "saw" when the AF was done. I think that when people notice more BF or FF at wider apertures, it is because the DOF is shallower, so you are more clearly seeing the focus that the AF sensor achieved.

Note that the AF is not always consistent from shot to shot. Do you get the same results if you defocus, and shoot multiple times? Also, there is a rather wide tolerance to where the camera can be pointed and still achieve focus on that chart.

I am not sure exactly why the camera back focuses, but I think that AF lenses can be calibrated by the manufacturer. I had it done on my only lens that was front focusing, and it came back perfect.

[Marc Sabatella]

So that brings me to the question: How is it possible for the camera to "know" what focus is right, and then set the automatic lens incorrectly? I don't understand what is really CAUSING it, and what I need to try and get fixed.

A physical cause could be this: the phase detect AF system knows when something is in focus, and it also knows, for something that is out of focus, how far out it is, and in which direction. And in that case, it tells the lens to go focus itself (by turning the screw with most lenses). Sure, it checks again when the lens is done to see if it got it right, but it seems this second check might not be too picky. So if it didn't completely nail the amount by which the screw had to be turned, you could be a little off.

Whether that's going on here is hard to say. I'd assume you are correct that the kit lens doesn't render a shallow enough DOF to expose any focus issues that are the fault of the camera in your case. And it's tough to tell from those small pictures and the information you've given what might be going on - like you say, it seems inconclusive so far.

[sluggo913]

I had the same issue with my K10d and brand spanking new Tamron 70-200 f2.8 Di. How could I have a problem with back focus it must be the lens, I sent it back 4 weeks later I get a replacement marginaly better. What is going on? My other lenses work fine, but upon closer inspection hints of back focus showed up in other shots, I down loaded the debug program and spent some time adjusting and retesting and on and on in the end I had to adjust the camera +150um to get it dead on. The lens was adjusted on a tripod at 45 degrees under daylight balanced lighting at f2.8 and 70mm now it focuses fine each and every time now I just need to see how it affected the other lenses. Hope this helps and good luck.

[Mister Guy]

A physical cause could be this: the phase detect AF system knows when something is in focus, and it also knows, for something that is out of focus, how far out it is, and in which direction. And in that case, it tells the lens to go focus itself (by turning the screw with most lenses). Sure, it checks again when the lens is done to see if it got it right, but it seems this second check might not be too picky. So if it didn't completely nail the amount by which the screw had to be turned, you could be a little off.

Whether that's going on here is ahrd to say. I'd assume you are correct that the kit lens doesn't render a shallow enough DOF to expose any focus issues that are the fault of the camera in your case. And it's tough to tell from those small pictures and the information you've given what might be oging on - like you say, it seems inconclusive so far.

auto_50mm on Flickr - Photo Sharing!
man_50mm on Flickr - Photo Sharing!


And here's a "real" example of the situations when it really annoys me:

IMGP1703 on Flickr - Photo Sharing!

That last one managed to lose it's EXIF at some point, but it should be somewhere in the 5 foot distance, f/6 range, unsure on focal length. I can look up the original later, it's on a DVD somewhere.

[Marc Sabatella]

IMGP1703 on Flickr - Photo Sharing!

That last one managed to lose it's EXIF at some point, but it should be somewhere in the 5 foot distance, f/6 range, unsure on focal length.

OK, but think about it - the larger test pictures you posted showed a difference of different of only a few millimeters in terms of where the focus zone zone, and that was wide open. Shooting at f/6, the DOF should have completely obliterated a focusing error that small. If the focus is not where you wanted it here, that would seem to be a simple matter of the camera having chosen to focus somewhere other than where you wanted. Even if you were careful to use the right focus "point", those aren't points at all - they are fairly large rectangles. You could easily have a focus point right on her eye but have the camera notice something else nearby and choose that instead.

The larger test pictures you posted, BTW, do help, but that's still just one test sample. Really, you'd want to look at several tests each at a variety of focal lengths (for zooms), in different types of light, etc, before drawing any conclusions. It's way too easy to get fooled by one aberrant result.

[Mister Guy]

Well that entirely depends on the mechanics of it, which is what I've been trying to figure out. The problems I'm experience with that particular lens seem to be proportional. What's a few millimeters off at a foot seems to be a few inches off at three or four feet. I would buy into user error if swapping lenses didn't fix it.