Originally posted by falconeye I think the K-7 has no widely observable problem because its less light-sensitive AF module gives up (in the vast majority of cases) before the colorimetric sensor does.
Are we sure the K-7 does not have a problem? I thought I've seen a few posts from users saying they have noticed front focusing on the K-7 as well?
Perhaps the K-7 has a milder version of the problem?
I quite like your theory (even though I would still call it a hypothesis) - it does seem to fit observed behaviour reasonably well.
My own limited testing indicates my camera performs similar to yours. I am using different lenses, which may account for me getting slightly better results than you (I start to get a focus shift around EV1). I don't think that's significantly different.
However, my results are not consistent, which is why I was interested to know whether your results are also inconsistent.
In exactly the same conditions, over a sequence of shots - sometimes the camera will correctly focus, and sometimes it will not. It may depend on the starting focus position of the lens - focusing from the front tends to create front focus, focusing from the back sometimes create a correctly focused picture.
I don't know whether you have noticed this as well?
The other thing that remains an issue for me is the wide variance in the reported severity of the problem.
That's why I asked you for an interpretation of your chart. It seems from the chart that you have not been able to observe any focus shift above EV2 - in your testing conditions.
That seems very reasonable to me, and completely consistent with your theory.
However, the degree of focus shift you are reporting, as well as the conditions (below EV2) will mean the problem should not be noticeable under "normal" shooting conditions.
But we seem to have reports of users who claim quite severe focusing problems at fairly bright conditions.
You have hypothesised that the problem can occur at high EVs - I accept that, but I am not certain it explains the issues that are being reported.
It is a possibility that some of these reports are "user error", but unlikely that every single report can be dismissed like that. Therefore there is the possibility that the problem appears to be different in intensity across cameras.
Do you agree? If so, how does your theory account for the apparent differences in magnitude? Does that mean that each camera needs to be individually calibrated?
I am only asking because it seems to me that if Pentax issues a fix that's based on recalibrating the AF correction based on the sensor reading, then it may not work on all units?
I am also wondering (if the units are individually calibrated) that AF fine adjustment may throw the correction off. This may explain the post on dpreview that claims resetting the camera back to factory settings reduces the front focusing.
Anyway, sorry for the long post. I enjoy talking to you, and you have a good analytical mind.