This is somewhat off-topic from FOSS, but this is a pretty knowledgeable group so maybe some help me to understand. My in-body camera micro focus adjustment (MFA) requires adjustment settings for both wide view and telephoto focal lengths, while my Sigma utility asks for correction settings for various focal lengths and settings.
But if I understand the need correctly (admittedly I may not), calibration is sometimes necessary on DSLRs in order to correct small differences in optical path lengths between the autofocus sensor and the photo sensor. In that case, I would have expected that you would only need a single correction because the error is inside the camera body and not the lens and that correction would apply to all lenses.
I haven’t been able to find an explanation elsewhere so perhaps someone here could help me out.
Hi @Dave22152 ,
First, there’s a basic assumption in your logic we need to correct. Lenses aren’t perfect. Not even close. And zooms are even worse. I could go on and on about all the ways a zoom lens could never be perfect, but I hope we can agree on the basic premise that lenses (and camera bodies) aren’t perfect.
There, now that we have that out of the way, when we look at the optical path we have to account for both body and lens anomalies. Since the lens isn’t perfect, those anomalies will present themselves differently in different lenses, and for zooms at different focal lengths. You are correct that the anomalies in the body will typically present as relatively consistent within and across lenses, but we still have to account for the anomalies in the lens, too. That’s why we MFA at different focal lengths for zooms.
Anecdotal story for you: A while back I was helping a good friend MFA his lenses using FoCal. He’s a professional shooter, and all his glass is pro line Nikon. All, and I mean all, of his lenses were back focused. This was due to the body being out of spec/alignment. His lenses though, ranged from -8 to -15 MFA setting. The moral of the story is no matter how good the lenses are, nothing is perfect. MFA at different focal lengths helps us negate these imperfections the best we can.
Hi Michael, thanks for your response and welcome to the group!
My experience with my 150-600mm zoom certainly bears that out!
Yes, clearly the lens adds a component of error but I wasn’t sure how that works. If we corrected the in-body camera error then I would have to assume that the change in focal length causes the light to converge differently on the phase contrast focus sensor than the image sensor, but how so if the two optical paths are now presumable the same lengths?
A simple guess is that the various wavelengths change with focal length, but then again the focus sensor should “see” that convergence no differently then the image sensor. Obviously not since we have to calibrate for those FL differences.
Ah, I think I see what you’re asking now. Apologies if my previous answer was a little too simplistic.
In the communication between a body and the lens, the body tells the lens to focus “X” distance away. How the X is computed is different for each body, but it’s based on getting the phase contrast focus sensor readings within a certain set of parameters. You experience this as the lens hunting for focus. Once it’s in locked in, the communication between body and lens moves from hunting to discrete adjustments to focus. This is where MFA is key. It fine tunes those discrete adjustments accounting for anomalies in the body and lens.
In other words, the phase contrast sensor isn’t constantly sending microadjustments to the lens to get the absolute best reading, but instead is sending discrete focus adjustments based on what the sensor “thinks” is the best it can do. MFA helps tweak that.
Don’t know for sure but I’m guessing that having two measurements would compensate for error in the focusing mechanism. If the camera expects that commanding the focus motor to turn x degrees will move the lens by nominally y mm, but something is slightly out of tolerance so that turning x degrees only moves it by 0.98*y mm, then it could figure out that correction factor of 0.98 from two measurements.