[klh]

I am continuing to have a problem with back focus with my K20D and Tamron 70-200 f/2.8 that I can't figure out.

At long ranges, 200mm, f2.8, I am seeing back focus. The focus was on the coin toss. I can improve the situation a lot by dialing in +10 on the focus adjustment.

File Date

View Picture EXIF

But in testing at close range, 200mm, f2.8, the focus looks good, maybe even slight front focused.

File Date

View Picture EXIF

Do I have a problem with the lens, the camera, or the user?

---

Last edited by klh; 08-30-2010 at 05:14 PM.

[JeffJS]

Looking at the graphics on the field, I'm wondering how much of this is A, a motion blur problem or B, too wide of an aperture (f4) for the shot or both. I don't know the lens so I don't know what I'd expect out of it at that distance, at f4 let alone f2.8.

I do note that in looking at the left side of your chart, it does look somewhat front focused. I also note that the chart is not completely horizontal with the camera (the black to white transition in the circle is tilted to the left).

[Marc Sabatella]

You might have wanted the camera to focus on the coin, but it cannot read you mind. It imply *chose* to focus on the background. That's not backfocus; it's just life with AF cameras until the day when they implement mind0reading technology. The way people have traditionally worked around this is to aim for something where there isn't a background - like the players' feet. This is assuming you're selecting a focus point yourself and not letting the camera pick - that would be the ultimate in expecting the camera to read you mind. but even when selecting a focus point (or using center point mode), the focus "point" is actually just a rough guide that "focus somewhere near here", not "focus on this exact spot". And as far as the camera is concerned, when focusing on small objects, the background appears "near" the object (after all, they are physically adjacent in the image).

[rparmar]

I would be manually focussing on where I want the lens to focus. But maybe that's just me.

[klh]

You might have wanted the camera to focus on the coin, but it cannot read you mind. It imply *chose* to focus on the background. That's not backfocus; it's just life with AF cameras until the day when they implement mind0reading technology. The way people have traditionally worked around this is to aim for something where there isn't a background - like the players' feet. This is assuming you're selecting a focus point yourself and not letting the camera pick - that would be the ultimate in expecting the camera to read you mind. but even when selecting a focus point (or using center point mode), the focus "point" is actually just a rough guide that "focus somewhere near here", not "focus on this exact spot". And as far as the camera is concerned, when focusing on small objects, the background appears "near" the object (after all, they are physically adjacent in the image).

Thanks Marc. I hope you're not seriously thinking I was trying to focus on the coin .

I was focusing on the players at midfield with only the center point AF active, but it seems that the focal point is more on the sidelines. I've got numerous examples of this (the player is blurry, but the background is sharp), but showed this one because it seems like the camera couldn't miss as the subjects were large and motionless.

My problem, in a nutshell, is that for close up objects, the focus sharp, but for distant objects, the camera seems to back focus. I don't understand how that could be, but it is. I realize it's likely my fault, which is why I haven't returned anything for warranty service, but I haven't figured it out.

I'll keep working with it and try to post a few more examples.

[Marc Sabatella]

Well, you mentioned the coin toss, not the players, so yes, I did kind of think that. But even if you aimed for a player's head, the background would still have been in range. Still seems likely that something like this is what happened - there's basically no way the focus system could be off by that much.

[wlachan]

Though popular, but avoid doing AF test/calibration with close-up distance focus chart, do it at near infinity target.

[woolcotm]

Was there ever any conclusion to this? I am experiencing exactly the same issue. AF is spot on at shorter distances which is why I hadn't noticed it (cause I use it mainly for portraits) but it back focusses more and more the further away the subject is. This weekend I took quite a few shots at an athletics meet and even the static shots (subjects standing still) show quite bad back focussing. Using a K5 BTW.

[Marc Sabatella]

My post above *is* the correct conclusion. Cameras don't read minds. Sure, calibration can occasionally improve focus by a millimeter or two here or there, but that's not the issue here. It's just the much more mundane age-old problem of not being able to completely control exactly where the camera even tries to focus. If you see focus off by more than a couple of millimeters, it is not backfocus - just insufficiently controlled technique in communicating to the camera where it should try to focus. The further away your subject is, the smaller it is rendered, and hence the greater the difficulty in convincing the camera to focus on the subejct rather than the background in that same general area of the picture.

[woolcotm]

OK. Well in this instance I do have the problem. For instance, I focussed on the side of a high jump mat (from the front) and I had the mat and bar full width of the frame (so big target with good contrast - stripes on the side). Shooting at f5.6 I found the focus starting at the edge of the mat (and only just) and ending behind (a couple of meters behind i would guess). Yet when I callibrate the lens using a test chart at a distance of say 1.5m, the AF is spot on. Never had a problem like this with my Pentax 55-300 which is what I have tended to use for these events in the past.

[klh]

Was there ever any conclusion to this? I am experiencing exactly the same issue. AF is spot on at shorter distances which is why I hadn't noticed it (cause I use it mainly for portraits) but it back focusses more and more the further away the subject is. This weekend I took quite a few shots at an athletics meet and even the static shots (subjects standing still) show quite bad back focussing. Using a K5 BTW.

I never updated the posting, so let me share a few things I learned along the way. First, I should thank Marc, and encourage you to pay attention to what he is saying here. In general, the problem was the user (me). I don't see it much anymore because of the following:

1) I purchased a magnification eyepiece and a split prism focusing screen for my K20D, and I shimmed the screen. While I did that to use my manual lenses, it also helped me to better see where the camera AF system was really focused. That, in turn, helped me to adjust my technique to operate within my camera's limitations. Incidentally, I still use the magnifier on my K-5, but I have not yet invested in a split prism.

2) For shots with distant targets (like my example), I typically use f/5.6-8 now to give myself a little more leeway. This is also helped (for indoor shots) by the great high-ISO capabilities of the K-5.

3) I switched to using the AF button to focus (in AF/C mode), then letting off when I am sure I have the target. I also pre-focus a lot more.

4) I practiced with the AF system on a plain background with a small high contrast target to understand the size of the focus points. I normally use center focus only, and was suprised to find that the center cross was larger than I expected, leading to it sometimes choosing the wrong target.

5) I read up on how the focus logic works, and started looking for high contrast items in my frame that could confuse it.

6) For tricky shots, I found out that prefocusing using zoomed LiveView and manual focusing never fails. This was very helpful for indoor diving events (low light, fast shutter, moving subject).

[woolcotm]

I get all of the above and I'm sure liveview would sort it out (different AF mechanism) but thats not very practical shooting hand-held with a large/heavy lens. Anyway, here are two shots taken at roughly the same time which demostrate my point (both are straight Lightroom conversions with all sharpening turned off):

The first is a quick snaphot of my son poking his head out of a tent. Didn't quite get the composition right but the focus is spot on (150mm f5.0):

http://www.npl-photo.com/Test/IMGP1444.jpg

The second is one of about 20 high jump shots. In each of them I focussed on the side of the mat using the blue/yellow lines to get a big AF target and then recomposed (160mm f5.0).

http://www.npl-photo.com/Test/IMGP1440.jpg

They all came out the same. The blue storage box behind the mat is sharper than the front of the mat (and I was using AF-S - not AF-C).

Finally, a shot that didn't require recomposing (200mm f5.0):

http://www.npl-photo.com/Test/IMGP1355.jpg

Check out the grass around their feet. All the sharpness is immediately behind them.

Out of 159 shots, only 20 or so had the correct focus (all under a few meters away). The rest were further away and all back focussed.

[woolcotm]

One thing I did discover, the camera was still in spot-metering mode from a previous session which explains the exposure being off. Everything else was setup OK though (I have re-checked the settings and menus). I would also add that I didn't spot the back focussing till it was too late because it was too sunny to see the screen properly

[Marc Sabatella]

OK. Well in this instance I do have the problem. For instance, I focussed on the side of a high jump mat (from the front) and I had the mat and bar full width of the frame (so big target with good contrast - stripes on the side). Shooting at f5.6 I found the focus starting at the edge of the mat (and only just) and ending behind (a couple of meters behind i would guess).

Post the image (with Exif intact), and I'm sure we'll be able to point out why the camera chose to focus where it did. it's hard to understand from your description what was happening - which specific focus point you chose, what specific object was behind that focus point, how much space in the viewfinder was taken up by that object and whether it completely filled the range of the chosen AF point.

But again, if you are finding focus behind your intended subject by *meters* as opposed to *millimeters*, there is absolutely no possible way that is a camera defect. The camera would clearly have to have *chosen* to focus there, presumably because you didn't control the use of the AF point as well as you think you did. Again, no possible camera or lens defect could possibly make the it focus meters the place where it tried to focus. beyond any possible shadow of a doubt, the camera *chose* to focus there. so your next task is to understand *why*, and then learn to control the process better.

EDITED: now I see images. In the high jump image. it is quite apparent the camera chose to focus on the wall behind the jumper. It's perfectly in focus. So assuming you aimed the camera such that the red dot in the viewfinder was directly over one of the yellow lines, you must have aimed high enough that the background was within range of the sensor. You absolutely must understand the AF sensors are much larger than the red dots. So just because the red dot is directly over the yellow line does not mean that is necessarily where focus will be.

BTW, that assumes you know you aren't suppose to let go of the shutter between focusing and recomposing. That of course would also cause what you are seeing.

[klh]

In the high jump image. it is quite apparent the camera chose to focus on the wall behind the jumper. It's perfectly in focus. So assuming you aimed the camera such that the red dot in the viewfinder was directly over one of the yellow lines, you must have aimed high enough that the background was within range of the sensor. You absolutely must understand the AF sensors are much larger than the red dots. So just because the red dot is directly over the yellow line does not mean that is necessarily where focus will be.

BTW, that assumes you know you aren't suppose to let go of the shutter between focusing and recomposing. That of course would also cause what you are seeing.

If I can add another possibility: If you were using the auto-AF, the camera may have decided to focus with the sensor on the wall. When I focus and recompose, I make sure I am only using center point AF.