There is not one adapter that does that, there are many.
That conclusion is mostly incorrect. AF is geared towards speed and avoiding inconclusiveness, which you might refer to. The AF sensor can see when something is actually in focus. A AF lens can produce quite different focus results depending on which mode the camere/lens is in (One shot focus with focus confirmation, AI-Servo mode, or manual focus with focus confirmation, and even focus trap).
How do you account for those chips' inability to provide appropriate Best Focus Correction Values?
Again, AF behaviour and MF behaviour of the AF sensor system are not the same thing. Even AI-Servo and One shot focus with focus confirmation can deliver different results.
Focus confirmation is the same thing in all cases. Appropriate BFCVs are required for accurate focus and those chips can't provide those values.
Different in what way?
I succeeded.
But it was irrelevant to what I'm talking about so I didn't address it.No, it isn't and you know it (I know you do from past discussions). One shot focus does NOT check if focus is reached, if it told the lens to do the last step. Manual focus just checks constantly to look if it sees something in focus.
Why "are they required" if the AF sensor actually can see if something is in focus or not?
One shot focus does not check if focus is reached. AI-Servo constantly looks if something is in focus or not.
Focus trap and manual focus focus confirmation images taken with various adapted lenses with various focus confirmation chips, on Canon EOS 450D and Canon EOS 6D:I succeeded.
You remember incorrectly. There is always a final check except when both a) the shutter button is fully depressed and, b) the subject is just out of focus. That is a very rare event for most photographers, but let me say what I meant again with further clarity - focus confirmation is the same thing in all cases of normal use.
Yes, it confirms in the same way but does not halt.
Because cancelling the phase difference between the AF sensor elements and having the lens in best focus is not the same thing, that's why you need correction factors to bring the two in line with each other.
No, steps are small compared to the thinnest DOF. The best way to think of body calibration and AF MA is as an offset to the BFCVs.
It doesn't matter how the subject comes into focus (AF, MF, or change in camera-to-subject distance), confirmation is the same independent process with the same behaviour (including interrogating the lens for BFCVs, which are not available from an adapter chip).
You can't rely on getting accurate focus confirmation if you don't have good BFCVs (except by luck), and those chips can't deliver them.
It does in all cases apart from one rare one.
Yep, it's the same process as One-shot but One-shot stops when focus is confirmed.
Sure, you can get lucky, I never said you couldn't.Focus trap and manual focus focus confirmation images taken with various adapted lenses with various focus confirmation chips, on Canon EOS 450D and Canon EOS 6D:I succeeded.
That is even a contradiction (and not true in the 1st place).
It is not.
That is incorrect. unless you mean that the AF can see part of the spectrum in focus and we, looking at the whole spectrum in the resulting image, see it OOF.
Correction factors are about the not checked last step not being in line with what the camera would expect, and correction for LoCA/spherical aberrations (see above)
No, steps are small compared to the thinnest DOF. The best way to think of body calibration and AF MA is as an offset to the BFCVs.
It doesn't matter how the subject comes into focus (AF, MF, or change in camera-to-subject distance), confirmation is the same independent process with the same behaviour (including interrogating the lens for BFCVs, which are not available from an adapter chip).
You can't rely on getting accurate focus confirmation if you don't have good BFCVs (except by luck), and those chips can't deliver them.
It does in all cases apart from one rare one.
Yep, it's the same process as One-shot but One-shot stops when focus is confirmed.
Sure, you can get lucky, I never said you couldn't.Focus trap and manual focus focus confirmation images taken with various adapted lenses with various focus confirmation chips, on Canon EOS 450D and Canon EOS 6D:I succeeded.
--
Check out the unofficial Rebel Talk FAQ. Sorry it's out of date, but DPReview still won't allow us to edit our articles.
Really? It sure does on my SL1.
The AF takes several steps, the final step does not get confirmed. This is the reason why some lenses sometimes, or consistently, give totally OOF results.
Only several? Care to elaborate?
Care to elaborate? Source for this information?
Several is open to interpretation.
You can notice it yourself (with certain lenses). My EF 35mm f2's movement you feel when holding the camera/lens. In One-Shot focus mode, when I just let it focus on targets far and close by, at times you hear the beep/see the focus confirmation dot in the view finder while you feel the lens AF mechanism is still busy moving to its final position. Same with my EF 70-200mm f4 L USM, but the lens focusses a tad faster and more silently so it is a tad harder to notice.
That doesn't sound like the right thing to do, and for what it's worth, my SL1 doesn't do that. I just checked with an STM lens and a USM lens. Actually, it stops focusing a fraction of a second before I get the confirmation light. Furthermore, if you move the camera before it can focus, it focuses on the new target, which shows that the lens does not just go to the defined focal point and stop without checking.
The lens/camera combo does this:
Well, I did, quite a few times, and it never worked that way. The results were unambiguous.
You're right of course, experiments like the Squirrel Test (Busted! The Myth of Open-loop Phase-detection Autofocus) prove beyond any doubt that focus can only be confirmed in One-Shot with a half-press of the shutter button when an AF sensor sees an in-focus subject.
Please provide a reference to the source of this information.
Yes, what I described was in essence the Squirrel Test. The SL1 does not use an open loop, at least under the conditions I tested.
