I call this "the beep band". The Manfrotto macro rail has an Allen head bolt in the end of the shaft, so you can put a hex key in it and wind the shaft very accurately (as long as you load up the backlash). With something like that you can play with the beep band all day.
Go ahead and try to answer that question for them. I'm interested to know what you find, but it is not the question I am trying to answer.
Excellent. The flash illuminates the AF sensor, the AF sensor says the phase difference is cancelled, the AF controller determines that no focus ring motion is required and confirms focus. That's it, we're done! Congratulations.
When I did try, you complained that my answer does not fit your theory. That was a good number of the 150 posts.
It's not that your answer didn't fit with the theory, it didn't fit with the findings. Your explanation said that the camera's PD AF only makes 1 initial Phase measurement and then it sets and confirms focus based on that single initial phase measurement.
I always said that description was oversimplified. yes, the system does take multiple looks, but something must go unchecked (or the check is not what WilbaW's theory postulates) or we could not have the types of FF and BF failures that bring up the question.
If that were true, then we could never have lens dependent FF or BF - LensAlign, lens focus calibration, and lens specific microadjust would not need to exist. It's the puzzle of how the system can fail in the ways it's been observed to fail that suggest the system has some shortcut that makes it not a perfectly closed loop feedback system.
Why not true? Couldn't there be a slight variation in the lens' image projection that throws off the calibration between the AF sensor and main sensor? So that even though the AF sensor is satisfied with the phase alignment there is still BF/FF at the main sensor?
Your model did not fit the evidence .
Of course. Erik's argument has been debunked, in great detail, many times, but he refuses to accept the evidence and let it go.
Sure. But how would this be lens copy dependent? 50mm f/1.8 copy A may need +5 and copy b may need -8 even though both will be equally sharp with contrast detect or manual focus is the same (i.e. not a lens optics alignment problem.)
When this happens, it's typically the camera that gets adjusted because all lenses behave similarly (e.g. they all FF or BF).
It fit a subset of the evidence that included the problem the OP was having. Just as your model fits a (much larger) subset of the evidence but explicitly excludes the problem the OP was having.
We're still waiting for such a model. You prefer to just handwave this issue.
I've let quite a bit go. BTW, how's that f/16 aperture simulation coming? Learn about vignetting yet? See, I do admit when I'm wrong but if you are wrong, the subject somehow just gets dropped.
Alas, "does not preclude" is very different from "equals." Until you can model how the lens copy specific differences actually can result in optical focus confirmation but varying amount of FF and BF, there is a large gap in your theory.
What if the lens is adequately aligned to the main sensor but not so well aligned to the AF sensor? Is that possible in throwing the AF? In these cases AF microadjust would relate to offsetting the point at which the AF sensor is satisfied with focus.
The AF sensor is in a fixed position in the optical path so the difference between the sensor path and the AF path should also be constant. What kind of lens misalignment would have this effect but no other significant optical problem?
Not a lens misalignment, but the seating or incorrect position of the submirror could cause this.
Yes, but it would still be constant for any given lens type, i.e. all 50mm f/1.8's would behave the same on that body w.r.t FF or BF. In fact any lens at 50mm would behave the same and all lenses would be impacted in somewhat the same way w.r.t FF or BF. This is an AF failure mode, but one that is typically resolved by fixing the body and not individually calibrating the lenses. You don't even have to send the body with the lens for many calibrations.
Only by being obviously wrong, which is not a good basis for a model.
Yes, answering that question was never the purpose of this thread.
Go for it. Best to start a new thread since that is OT for this one.
Ah, I get it! You think this thread is about answering writersbloc's issue from the other thread? No wonder you're so confused and frustrated!
You want me to make an external aperture that gives the same shutter speed as f/16? Doesn't sound too difficult. What do I do with it then?
How is my question about the effect of an f/2.8-equivalent external aperture wrong? (I'll make it easier for you. There is a measurable effect on the front-focus error - do you think it would make it larger or smaller? Why?)
You mean the stuff that I'm not modelling? Well, yeah, that's what out of scope means. Do you have any evidence within the scope of the model that I haven't considered?
Correct. "Equals" comes from the evidence you get from doing the tests. "Does not preclude" simply invalidates your excuse for not accepting the evidence.
Only in a theory that attempts to explain that, i.e. not mine.
That's not the purpose of this thread. Don't forget, this is an empirical black-box model of the control process and focus confirmation - nothing to do with how lens optics interact with AF sensors, or about answering writersbloc's question.
You'd be surprised.
