An idea for helping the investigation. There are event procedures dealing with the position of the focus lens.
This is a good idea. An enhance test script that reports more about what the camera "thinks" it is doing might be valuable.
The big test suite we did turns out to have had two major limitations. First of all, the test was done at distances from 1/2 meter to just over 1 meter. And it depended on the camera's report of the actual subject distance - which we compared to the requested subject distance. We learned a lot from that about which cameras would respond to what function calls. But we did not ask anyone to check ( or calibrate ) the resulting images. And we did not check that the camera would focus at infinity.
So I have a few working theories at the moment.
- I suspect that the SD values reported by most cameras are poorly calibrated guesses.
- When a camera is in auto focus mode, the focus algorithms adjust the lens mecha to get the sharpest image. The control loops work with lens mechanical position - not calibrated subject distance.
- For cameras without MF capability, using CHDK to set a focus position gets you a lucky guess of how close the result will be compare to the requested distance
- There is considerable variation between cameras of the same model and firmware release.
Having an enhanced test script like you suggest might help prove / disprove some of my theories. I'll add it to my "to do" list.
