Focus accuracy

krypticide · 11-05-2009, 11:08 AM

So I've noticed that I'm having focus accuracy issues on my K20D. I keep thinking my lenses aren't sharp, because the majority of my images seem to have missed focus. however, maybe one in ten images come out properly focused, and then I realize, "Hey, this DA* 16-50mm is actually sharp!"

It's not like I'm straining the AF system. This problem happens even with a simple portrait session in good lighting. My AF is micro-adjusted using a focus test sheet, so in theory I should have eliminated any lens BF/FF.

Any ideas on what's going on? I typically use center-weighted focus, aim at the face, and recompose. I see focus errors even at f/5.6, so it shouldn't be issues with focal point shift due to recomposing.

Is the K-7 AF more accurate?

*Wheatfield* · 11-05-2009, 11:44 AM

Originally Posted by krypticide

So I've noticed that I'm having focus accuracy issues on my K20D. I keep thinking my lenses aren't sharp, because the majority of my images seem to have missed focus. however, maybe one in ten images come out properly focused, and then I realize, "Hey, this DA* 16-50mm is actually sharp!"

It's not like I'm straining the AF system. This problem happens even with a simple portrait session in good lighting. My AF is micro-adjusted using a focus test sheet, so in theory I should have eliminated any lens BF/FF.

Any ideas on what's going on? I typically use center-weighted focus, aim at the face, and recompose. I see focus errors even at f/5.6, so it shouldn't be issues with focal point shift due to recomposing.

Is the K-7 AF more accurate?

In theory being the key. I've found that the goofy little focus charts are a bit less than useful. If I find that a lens is missing focus, I use a real world object to check focus rather than a chart that I have to be within a few feet of.
It seems to work better for me.
The K-7 isn't much different from the K20, sample variation will have as much to do with your AF adjustments as anything else.

**Marc Sabatella** · 11-05-2009, 02:04 PM

Can you post some samples?

schmik · 11-05-2009, 02:17 PM

I think that using the focus point other than the center one, to avoid recompossing is a better idea. If it misses at f/5.6 then there is something serious going on. It may be worth setting up some sort of focus test at your most used distance... rather than very short distance used by focus charts.

Are you sure it's focus that is the issue. The DOF at 50mm and 3m to subject distance is 80cm.

mike

the swede · 11-05-2009, 02:21 PM

I'll have to agree with Wheatfield on that! I currently have my DA*50-135 att Pentax for repair. I noticed on real world pictures that there was BF on it. Testing it on a chart was acctually harder to see the slight BF.

I tested the lens by pointing it at an house 40 meters from me, and used the smallest f number and fully zoomed in. I could then see clearly on the image exactly where the depth of field was. And it was clearly somewhere infront of the house. I even managed to manually focus the house and het a better picture.

So, a slight bf of 4-5 mm on the chart will show itself much greater in a real world image.

In my case the in camera af adjustment didnt help. There was also an missalignment in the lens shifting the focus slightly to the right making the center even more soft.

Oh...that reminds me. they've had my lens for 2 months now. Must be a hard nut to crack.....

krypticide · 11-05-2009, 06:21 PM

Here are some (100%) examples. I'm wondering if shake is an issue. I usually remember to wait for SR to activate, but it's possible that I might make a mistake sometimes. However, all of these shots are relatively short shutter speeds (near the flash sync speed).

#1: good focus, very sharp

Click on the image to view it in a larger size

#2: front focused? hard to tell, maybe shake?

Click on the image to view it in a larger size

#3: weird issue, from sharp ear on the left to less sharp face on the right. doesn't seem to be DOF issue, shouldn't be shake, so what the heck?

Click on the image to view it in a larger size

Imageshack album link is here, since those images seem soft when viewed on the forum.

rawr · 11-05-2009, 07:06 PM

Originally Posted by krypticide

Any ideas on what's going on? I typically use center-weighted focus, aim at the face, and recompose. I see focus errors even at f/5.6, so it shouldn't be issues with focal point shift due to recomposing.

What I see in the 2nd image is simple camera shake, perhaps also combined with some subject movement. 1/160 isn't that fast a shutter speed. Maybe you should both use a tripod and step up the shutter speed.

That may also be the issue in image #3 - shake and subject movement.

Also what is the distance to subject of these portraits?

At 50mm you don't have a lot of DOF to work with, especially close up:

Depth of Field Table

rawr · 11-05-2009, 07:12 PM

Also sometimes manual focus + DOF preview (+ tripod + high shutter speed) is the best solution, especially if you have subjects that aren't that dynamic.

**Marc Sabatella** · 11-06-2009, 11:00 AM

Not sure why you'd say DOF isn't the issue with #3 - it seems pretty obvious to me that's exactly what's going on. You've got the ear in focus, but DOF is shallow enough at that aperture, focal length, and subejct distance that the faces are beyond the back edge. Combine that with the fact that the subjects look to be in motion, and I think this picture looks about exactly as one should expect.

#2 looks more problematic - subjects appear to be still, but nothing appears in focus. And it doesn't look like camera shake to me. Actually, are you perhaps closer than the minimum focus distance? EXIF info would help. Personally, I'd write that off as a fluke for now but try to do some more controlled testing.

krypticide · 11-15-2009, 12:55 AM

Thanks for all the advice. I've come to the conclusion that it was indeed FF. I had adjusted FF/BF using a focus chart, but real world testing shows that the focus chart was giving me bad results.

Moral of the story is, don't use a focus chart, test with real pictures. My next shoot should yield better results...

**Marc Sabatella** · 11-15-2009, 10:07 AM

I disagree. The moral is, be careful how you test. Real world test subjects very often lead people to be seriously mislead, as the camera might be simply choosing a different focus poin. AF test charts are designed to reduce/eliminate that possibility, but it's still possible to use them incorrectly. And they do have the disadvantage that they can usually only be used from close focus distance. But the lessons you learn from using AF test charts is essential if you want to perform a test using real world subjects that isn't completely invalidated by the camera choosing a different target than you intended. And this sort of experience also carries over in focusing in the real world for real - eg, not just for tests.

Billgscott · 11-15-2009, 11:18 AM

Check this out on Ned Bunnells blog.

NED BUNNELL: Caution: Photography can increase anxiety & stress

It throws another light on the subject.

philbaum · 11-15-2009, 12:19 PM

Originally Posted by Marc Sabatella

I disagree. The moral is, be careful how you test. Real world test subjects very often lead people to be seriously mislead, as the camera might be simply choosing a different focus poin. AF test charts are designed to reduce/eliminate that possibility, but it's still possible to use them incorrectly. And they do have the disadvantage that they can usually only be used from close focus distance. But the lessons you learn from using AF test charts is essential if you want to perform a test using real world subjects that isn't completely invalidated by the camera choosing a different target than you intended. And this sort of experience also carries over in focusing in the real world for real - eg, not just for tests.

Can't disagree with "be careful how you test"

I can understand when one is working in a carefully set up repair lab, that a close focused test chart can be very accurate. Too many tiimes, i've been reading "but be careful to keep the camera lens perpendicular to the test chart" which is starting to make me think that people have experienced problems with their setups. The first time i did a test chart and submitted it online, sure enough, i got the comment back that hadn't aligned it properly. Also, it seems like one might have a potential front focus problem of doing the focus test indoors with tungsten lighting, then going outdoors in sunlight to take real pictures (unless one has the K7 of course)

The problem is, that a small alignment errors up close can get magnified many times at normal image distances.

I was stopped on a highway for an extended time yesterday, and so took out my K20 with my latest ebay lens purchase, and did a auto focus on the window edge of the car in front of me, some 25' away. I did trial and error AF camera adjustment, plus and minus until i had something that was close based on my katzeye focus screen (done at f2.8). Finding a focus edge on a car is not the most appropriate arrangement, but i still wonder if at distance setups may not be more useful than these up close indoor focus tests.