Upload & Sell: Off
| p.4 #11 · p.4 #11 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes |
Hello snapsy. Congratulations on your method! I don't have a camera with which I can test it, but I know from having done a lot of similar testing that it will work as advertised.
I wanted to share a few ideas that I haven't seen mentioned in this thread. PD AF is proven to be a closed-loop process, that ends when the AF sensor sees an in-focus subject. I'm not sure whether you are aware of the Best Focus Correction Value (Canon term). Check out Busted! The Myth of Open-loop Phase-detection Autofocus for a discussion which covers both. (You're welcome to PM me on DPReview if you like.)
The thing you're playing with is what Doug Kerr and I call the Acceptance Zone - the range over which a lens in MF will confirm focus using PD. I've played with that a lot by changing the camera-to-subject distance, and as you found you can also determine it with AF microadjustment.
It gets interesting when the closed-loop process is trying to put the focus within the AZ. In all of my testing (mainly with Canon), PD AF clusters the focus at the boundaries of the AZ, not randomly within it as you might expect. For a lens like my copy of the notoriously unreliable Canon 50/1.8, the AZ is about 3mm deep near the MFD, which can mean that one end is consistently spot-on and the other is outside the f/2.8 DOF (4mm). On some bodies, which end you get depends on the initial condition (focus in-front of or behind the subject), and you usually get the end on the other side, as if focus is confirmed on leaving the AZ.
PD AF is more variable than CD, but if you are able to control for the end-of-the-AZ-effect (for instance, if you have lens+body combination which always finds the same end from the same initial gross misfocus), then you can prove that even with a lens like the 50/1.8, the focus mechanism is capable of tremendous accuracy, like better than 5% of the thinnest DOF, which is 3% of the AZ, and 0.02% of the MFD.
So I'm convinced that the focus process within the lens is accurate enough that it's irrelevant to overall focus accuracy. Therefore, what matters most is how the AF sensor sees the world through a particular lens, and how the AF logic in the camera is able to use that information to determine when the two images align.
Your method uses the middle of the AZ, which will be fine as long as the AZ is significantly narrower than the DOF. Let me know if any of that doesn't make sense, or helps you understand what you're doing.