AF Problems

GeneralBenson · Jan 2, 2009

So I guess I've ben having auto-focus inconsistency problems. I'm not quite sure what is going on, but I'm beginning to think that it is with the camera, not with my lenses. I started noticing it when I bought my Tamron 28-75 f/2.8. That's only because I heard it has issues, so that's when I started paying attention. It could have been happening before that, but I pretty much didn't know good focus if I saw it. Then I started noticing my Pentax 14mm f/2.8 what semi consistently back focusing. Then I just got a Pentax 50mm f/1.4, which is amazingly sharp, when the focus is on. I believe the kit lens does the same thing, although I pretty much never use it. So the issue seems to be that all my lenses are randomly having focussing problems, and it seems to be consistently to the back. And I don't mean like, oh it's really dark kind of auto focus problems, or I took the picture before AF locked on kind of problems. Also, it doesn't seem to take like on shot in, then one shot out; but more like it'll be fine for a while, then it'll have issues for a day or for 100 shots or whatever. The other day I was doing an engagement shoot for some friends, and the middle third of the shoot it was just giving me a bunch of problems. As in, they're standing against a wall, and I have the AF point on one of their eyes, and the wall behind them is sharp. The camera was on selected spot metering of whatever it's called, not matrix of anything. Or using the 50mm wide open for a close up of the ring, and the finger behind it is in focus. But earlier and later, everything was spot on. So I'm wondering if it could be some kind of issue with the camera that is causing AF inconsistency issues. Sorry for the exceptionally long paragraph, but that flowed out of my head like a rant. Thanks!

Garbz · Jan 3, 2009

Grab your 50mm and shoot sideways against the wall at f/1.4. Draw a vertical line on the wall first to give the camera something to focus on and do some tests.

It is not unheard of that cameras backfocus, but unless the problem is reproduceable consistently any place you send the camera too will assume it was human error and send it right back again.

davebmck · Jan 3, 2009

Here is a site that explans how to measure back focus and has a chart you can download for the test. Nikon D70 Focus Chart

GeneralBenson · Jan 3, 2009

So, I printed up the focus chart from the webpage dave posted and did some test shots with all my lenses. The results were interesting. THe 50mm is seemingly spot on, so I guess that was user error on my part. I know that the DOF at 1.4 is so shallow that I could have just moved half inch between focusing and shooting. Also, I was using rathe slow shutter speeds that day, so it could have been slight shake. But the Tamron is what showed problems. At 75mm, f/2.8 it was slightly back, but nothing worth worrying about I don't think, but at 28mm it was way to the back. The 2mm mark in front of the focus line was already slightly out, but it was in focus all the way back to the 20mm mark on the backside. At 50mm it was basically you would expect; somewhere between good at 75 and bad at 28. I was refocusing each time I zoomed, so it wasn't that. So I would assume that is something to be concerned about. Is it unusual for a lense to be ok on one end, and messed up on the other end? I haven't tested the 14mm yet, but we'll see how that does later.

davebmck · Jan 3, 2009

The lens should focus the same throughout it's range. I would contact the manufacturer and provide them with the results you have, or return it to the place you bought it from.

GeneralBenson · Jan 3, 2009

Ok, new results. I tested my 14mm and my 18-55 kit lens. The kit lens is fine good at 55, but very back focused at 18 and the 14 is horrendously back focused. So what we have then is a 50mm which is fine, and 28-75 which is fine at 75 and bad at 28, a 18-55 which is fine at 55 and very bad at 18, and a 14 which is very bad. So, it would seem to me that the camera is having a progressively worse problem focusing as the focal length gets wider. So what now? Send the camera in or something?

davebmck · Jan 3, 2009

Where did you buy the camera. I would start there.

GeneralBenson · Jan 4, 2009

Got it straight from Pentax. Guess they'll be seeing it for a visit.

JerryPH · Jan 4, 2009

The "bad" results are strangely consistent, something that would almost never happen from lens to lens, but could happen if it was a camera related issue.

Either the camera is off or you were not careful in setting up the test as precisely as you should have. The camera MUST be at exactly 45 degrees. It has to be at a distance that fills the sensor with the chart as much as possible. If any of these things are off, your results will be incorrect

The method that gave me the most accurate results was to:
- set the lens to it's widest aperture
- if not a prime, set it to a known specific focal length,
- then set the distance of the chart so that it fills the screen properly
- most likely you will need to adjust the placement of the paper to retain that crucial 45 degree angle a couple of times until it is as close to perfect as you can get. The more accurate you are here, the more accurate your results will be.
- focus MANUALLY so that the center line of the chart is the one that is in focus, alternately, manually select a focus point dead center and have the camera focus on that, if your camera has this ability.
- then finally take the pic
- view the results on a large monitor at 100% to see the results in the best possible manner.

One weekend I plan to do the focus chart test on D700 with all my FX lenses and adjust the camera to optimize the focus for each lens. If your camera has this adjustability, it may be a good thing to do, if you want to extract the best out of your equipment.

GeneralBenson · Jan 4, 2009

Thanks Jerry, I'll run all the tests again to see if there was any error. But the results do point to what I have experienced. ANd were totally consisten from lens to lens, the wider the focal length, the worse it got, which is what I've been experiencing in the field. I hope my camera has adjustability.

andrew99 · Jan 5, 2009

JerryPH said:
The "bad" results are strangely consistent, something that would almost never happen from lens to lens, but could happen if it was a camera related issue.

Either the camera is off or you were not careful in setting up the test as precisely as you should have. The camera MUST be at exactly 45 degrees. It has to be at a distance that fills the sensor with the chart as much as possible. If any of these things are off, your results will be incorrect

The method that gave me the most accurate results was to:
- set the lens to it's widest aperture
- if not a prime, set it to a known specific focal length,
- then set the distance of the chart so that it fills the screen properly
- most likely you will need to adjust the placement of the paper to retain that crucial 45 degree angle a couple of times until it is as close to perfect as you can get. The more accurate you are here, the more accurate your results will be.
- focus MANUALLY so that the center line of the chart is the one that is in focus, alternately, manually select a focus point dead center and have the camera focus on that, if your camera has this ability.
- then finally take the pic
- view the results on a large monitor at 100% to see the results in the best possible manner.

One weekend I plan to do the focus chart test on D700 with all my FX lenses and adjust the camera to optimize the focus for each lens. If your camera has this adjustability, it may be a good thing to do, if you want to extract the best out of your equipment.

Jerry, why focus manually? The purpose of the test is to test the autofocus system. Is it to have a benchmark to compare it too? Even so, you're adding the potential for human error there which I don't see any benefit too. Or did you mean just manually select the autofocus point?

Edit: Nobody mentioned using a trippod, but that is essential to doing this test.. Just thought I'd mention the obvious in case anyone is trying this hand-held.