A thought why exif "focus position (2)" is biased for true focus distance

[weltyj]

In a recent adventure, I was playing with exiftool, setting focus position 2 for a completely manual lens (Samyang 12mm NCS CS).

Being the curious sort, I at one point was looking at the thin lens equation for distance of lens to sensor vs distance to subject:

    1/s + 1/D = 1/fl
where s is distance of lens to sensor, D is distance of lens to subject, fl is focal length of lens,

it dawned on me how sensitive D is for very small changes in s, when s is just a little more than the focal length.

For example, a lens with fl=50mm
s(mm)    D(m)
50.002  1250
50.012   208
50.022   113
50.032    78
50.042    60

Think about that sensitivity with regards to the machining of parts to create the lens, the placement of the optical elements inside the lens body, the machining on the mating ring of the lens to the camera body, the placement of the sensor in the camera body.

It seems within the realm of possibility for actual "s" values to be off 0.01mm or more from one camera copy to the next just due to slop in manufacturing/assembly tolerances, and how they could add up in a worse case scenario.   I assume the thin lens sensitivity problem is similar for lenses with multiple optical elements.

So it seems for any particular lens/camera combo, getting true focus distance given a "focus position 2" in the exif is going to be biased, maybe quite significantly.

Just another reason why any attempt at determining actual focus distance automatically with our "commodity" lenses should be viewed as a very rough estimate.

Jeff