K10D focus

Just thought I'd share my little experience here...

Having used the K10D for a year now, there were situations where it just wasn't getting the shot I'd imagined. Particularly with the 12-24 I wasn't getting the clarity I'd have expected, which was a real shame.
I took a little run with the trusty old *ist-D and the SMC-FA 24-90 and got pin sharp results.

While testing my recently defungussed SMC-FA 80-320 I'd realised that the 12-24 on the K10D wasn't getting it in focus. I had to focus at around 1/2 way into the scene to get the infinity in focus. Very odd.

Then, while at London Zoo and using the SMC-F 50/1.4 I was getting back-focus while shooting at f/1.4. In the end I was focussing and then moving myself back a smidgen to try and get shots in focus.

So, I scared myself silly reading the depressing revolt site, read a few more things, and decided to run the focus test chart test ( http://www.focustestchart.com/ ) this morning. So, I setup a tripod and the chart and tested with all my lenses.

Guess what? They all back-focus No wonder I've had problems


Good news is, though, it's sorted, AFAICT. A little digital tweak was enough to sort out the issue and now I've got spot-on accuracy. DOH! I've only been having this for a year!
(I'm not going into how I did the tweak on this forum. PM me if you need any guidance though )


In addition, thought I'd share these tidbits I gleaned from this test:
* Centre focus point is within the circle of the spot meter area.
* The four corner focus points are very wide and tall. They actually seem to spread down to the middle horizon - which is not what I'd imagined. But they are bounded by the rectangular marks in the viewfinder. (Imagine them occupying the whole corner, rather than just where the LED sits)
* The two focus points to the left and right of centre are about the same height as the centre focus. Again, they don't spread out of the bounds of the box.


Now, I'm going to put the K10D through its paces again...
Matt

I posted my similar problem awhile back and LCE ended up replacing the camera. I'm amazed you could fix it yourself, good on ye.

Hi Matt

Just completed the tests as suggested in Tim Jackson's document, looks like I may have some front focus issues with the FA50 f1.4 @ f1.4 and maybe f2 also.

I would appreciate your comments/thoughts, I've uploaded the test shots here....http://www.pbase.com/mcbrian/lens_tests&page=1

Having looked at your results, I would say that you do not have a problem. If I get half a chance, I will upload a couple of pics of what the problem is when a lens truly back focuses!

Kim

Found similar issues a while back. I had uploaded a pic of the sensor arrangement on my camera. I was a tad surprised at the size and arrangement of the sensors.

It turned out that the center of focusing sensors on my K10D is not exactly in the center of the focus screen. I now focus at the lower left of the circle and my photos are spot on.

Brian,

I think you need to set up your test chart more accurately. In all cases, your chart appears to be turned in a clockwise direction from the vertical. The centre 'focussing' line must be exactly horizontal for these tests to produce any really meaningful results.

Regards,

Brian - viewing those shots I dont think you did this:

5. When shooting your test pics, you need to be close enough that the test chart somewhat more than fills the frame. Get nice and close.

I can see a lot of "non-white" in your shots (table).....this might help as well.

Brian,
Your tests look just fine.
Although there may be a way to calibrate a single lens - I don't know how to do it.

My problem was consistent, repeatable, and across-the-board back-focus.
And, Peter, I didn't have it absolutely square either - that's not the point. The point is that (1) there is a very clear target to focus on, (2) you have a means of seeing where the focus lies by looking for the OOF numbers.

For example, on my 50mm f/1.4 test I had 2mm to 20mm behind in focus, with the target just on the edge of focus. Whenever I shot at f/8 or above there wasn't usually a problem - it was only shooting wide open the problem became obvious (which is something I do rather a lot).

I'll post some shots a bit later so you can see just what it was like

NOTE: Please only PM me about this if you actually have a problem. Please verify it first before asking
Matt

Thanks for the replies guy's, I only saw a problem in the first shot with the FA50 at f1.4 but on reflection it appears that my center focus point is more to the bottom of the circle, like amoringello's. The last bunch of FA50 shots was just a quick burst at different apertures. One thing I did notice is how good the DA16-45 and DA50-200 looked straight out of the camera.

Can only go as close as the lens focuses

Right, I've read through this thread a few times - I'm being thick and missing something! What actually is causing the problem? I can see that the camera is focusing behind where you want but is this a lens problem or a sensor alignment problem or what? My paranoia knows no bounds so I've downloaded the test chart out of curiosity but I'm determined not to use it yet...

Ken

Ken,
My K10D focus was off. All my lenses backfocussed. Everything was only really in focus if I used a small aperture.
The tweak was to bring the camera's AF system more into line with reality

As I said, it was more noticeable with the 50mm at f/1.4 (and for a long time I thought it was me until I tried the *ist-D again), and the SMC-DA 12-24 creating what looked like soft images (especially at infinity) - it was back-focussing infinity so that everything was soft. Now it's all pin sharp

Does that help?
Matt

So, is that a digital tweak (hack the firmware) or a mechanical tweak (adjust the focussing mirror)?

Whichever, does it void the warranty?

Mechanical tweaks are a seriously bad idea. AFAIK there is a complex AF table that has to be initialised on a special test-bed. Trip back to Japan for that I reckon.

The digital tweak was available in v1.10 (and can be restored to what it was)
Matt

So are you saying you reverted back too firmware 1.10 or just hacked 1.30 :
Should i be asking this here :

Understood. Thanks

Ken