manual focus problems on k10d

Glad to hear your camera's Ok, Peter.

A thought strikes me... with digital it's now incredibly easy to set up these quick tests, and just as easy to disseminate the results.

A few years ago, when we were all using film, how many would have bothered to look into this kind of thing?

I've now done the test more "rigorously", tripod, chart is flat, careful focusing and the results are all exactly the same as what I've been saying. Except the test shots with the IST DS, which are all perfectly sharp where they're supposed to be.

Here's four shots

http://www.flickr.com/photos/10737481@N06/

taken with 90mm f2.5 tamron, 50mm pentax f1.2 and 16-45mm f4 on autofocus)...the autofocus in all shots are the worst ones, I manually turned the lens out of focus before each photo and let it firmly settle on the focal point via AF perfectly sharp in the viewfinder. In all shots I also repositioned the tripod a couple times, took them from various angles, etc... The ist ds shots taken with the same settings are all great, focus right where it should be!

It's not a fault in the testing or my eyesight. The results are consistently out of focus on the k10d and in focus on the ist ds.

With respect, I'd urge you to look again at the shots you have posted as they are all askew. Different angles, magnifications, some twisted off axis, some not.

But if in the end you are unhappy then return the k10D for examination by Pentax whilst it's still under warranty. I would check with them what they would do if they disagreed and said the camera was OK. Would they charge?

The real question has remained unanswered - are your images sharp? That is, your real pictures. Because if they are the error is not significant except for the photography of test charts.

they were just as small so this is deleted

Try pinning the chart down flat and making sure it is parallell to the camera. Nothing should be moved between shots.

Certainly if you're getting consistently different results with two cameras and that is the only variable, it suggests that at least one of them is duff.

Forgive a dumb question, but are you using the spot focusing setting?

It's not on 'Sel' with something other than the centre spot selected, is it?

And you're definitely using the green hexagon, not the appearance of the red dot, to tell you it thinks it's in focus?

Other than these possibilities, it's sounding more and more like there's something amiss.

I'd disagree with this, John. If there's a problem, it will be worse wide open, which he'll often not be. So some shots will come out Ok. But on another example, opened up, it's not sharp - is it subject movement, or the focusing problem? Then one's sharp - maybe as a one off, subject movement made it so, and the uncertainty eats into your confidence in the equipment.

It would drive me completely nuts if I had this situation. I was hugely relieved to get a result for my camera that convinced me that it's set up Ok, to a precision higher than I'm able to measure.

The thing about the test charts, is that they give you confidence in the machinery's behaviour under controlled conditions - so you trust it when the conditions aren't so controlled.

If his chart was even close to straight, and the camera said he was focused properly on the central marker in his example focus_IGP7303xx, something is well wrong. That one is at least (root 2) cm (ie 14mm) wrong in a meter - hard to imagine that much experimental error.

What I don't understand is not noticing this before. Unless this sort of problem can "develop" (can it?).

In answer to your questions Chris (not at all dumb as I don't know what you mean by the "green hexagon".

Yes, I'm using the spot focusing setting

not on 'Sel'

I only ever use the red dot as a bit of a guide (i don't find it precise).

Anyway, the clincher is that now that I'm aware of this, I can easily see that all my shots today are slightly off. And that the ist ds is always spot on using the same techniques, tripod or not, fast shutter speeds etc.

thanks again for your help - dave
Other than these possibilities, it's sounding more and more like there's something amiss.

Check out the manual, p136. Also p134, section 2. Also p131, section 2 (at the top).

It's the only reliable way of knowing when the camera thinks it's in focus. You'll also get it at the bottom of the viewfinder when it's in AF and it thinks it's focused.

Well, I've done this test half a dozen times in the past. I've run this on pages of text and real objects. obviously wit tripod, etc...
Maybe my methods were flawed in some way?

So, today I downloaded the test chart, set everything up just right, and wouldn't you know my camera looks fine. I've tried a few different lenses. I get an almost perfect 50/50 to 1/3 - 2/3 split of front to back focus. (Perhaps my methods today are flawed?? )

I'm not sure why so many of my photos appear to be in focus just ahead of my intended focus point. But I think I need to figure that out because my camera appears to be just fine.

I guess I run these again some time later to see if the results stay consistent.


Anyway, thanks David for the link to the chart.
I'm not sure I feel good about the results just yet, but at least now I have something consistent to check against for a later date.

[my emphasis]

This is a good point. Getting the result that confirms the expected result doesn't, in itself, make it any more likely to be right or wrong, just more believable. That's been the problem faced down the centuries by scientists responsible for overturning the accepted wisdom of the day.

Make sure that SR is turned off when using a tripod to make these tests.

It occured to me that in his test shots, more of the image above (behind) the centre line is in focus than there is in front (below) but the actual centre line does appear to be in focus, as far as I can see anyway. Has hyperfocal distance got something to do with this?. It wouldn't explain the differences between two camera bodies though.

John has already mentioned that the test chart is not consistently straight and parallel to the lens axis which renders some of the shots less than conclusive.

I'm not sure it does, in itself.

Although the calibration on the chart won't correspond to reality (it's only right when it's straight and at 45 degrees to the vertical), I think a marker behind the middle will still be the same distance behind, as the corresponding marker is in front, provided the chart is flat.

But it is faintly suggestive of potentially poor technique in other areas.

However, it's his last example, focus_IGP7303xx, that is most noteworthy.

If the camera is telling him that the middle marker is the bit in focus, I'm struggling to see how there isn't a problem.

I'm very interested to see how this all turns out..

I found the original post title to be quite surpising. While it is somewhat common to hear of auto-focussing errors, I have never heard of front or back focussing happening on manual focus. In my opinion , the diopter issue is really irrelavent with your described problem. You will notice that if you set the lens to sharp focus that no matter where you set the diopter setting after that, the sharpest focussing screen image (while definitely not as sharp as before) will still be found at the same sharpest lens setting. The diopter setting does not change the lens focus setting, only the appearance of the focussing screen. The sensor will still record the sharpest resolution of the lens setting and not duplicate what appears on the focussing screen. (I hope that came out right.)

Dave, by your wording above you do not seem clear about the red dot and the green hexagon and never did respond to ChrisA's suggestion to check the manual. The red dot's only purpose is to show where the segment area the autofocus has picked and it is not an indicator of it being in focus. If you have autofocus set on the centre spot it will always only flash on that spot. Only the green hexagon indicates when the lens has properly focussed. Do you think that may be the problem?

Assuming you are doing that right and are doing all your tests correctly then I can see no other reason for the picture to be out of focus using manual focus other than either; poor focusing technique, which you ruled out with the *ist Ds, or that the distance between the lens and the sensor is fractionally different than that of the lens and the focussing screen. "Can this problem develop?" Probably, if the camera was dropped or heavily bumped.

Anyway, those are my thoughts.

Depth of field will render some sharpness to either side of the plane focused on and is generally well understood.

There is also some leeway in a similar way at the point the lens focuses the image on - the point where the film or CCD would be optimally placed. This is depth of focus.

The mirror must also be correctly placed to ensure that the distances are precise between the observed and the exposed images. This is why cleaning the mirror is definitely off the agenda as it is so easy to put it out of register.

Normally all these things are fine, or at least within acceptable limits, but if out of adjustment will prevent accurate focusing.

For normal subjects the test is less demanding than for test charts, even fairly close up, taking a portrait as an example we wouldn't mormally use f1.4 but might use f8 to ensure the whole face is in focus. Maybe under studio flash we would use f11 or f16. DOF will cover most things in this case.

The posint about SR is a good one. If switched on whilst on a tripod it could cause shake problems which would negate the validity of the test. One problem has been lack of consistency with the chart. It's absolutely got to be pinned down flat and in exactly the same place every time.