DotTune: AF tune technique without taking photos

1.) get a good high contrast target suitable for PDAF and CDAF (look out here comes Lenord S.) ...

2.) open LV and focus using CDAF (of course); magnify and observe the result simply to get some idea of a benchmark visually sharp result.

3.) close LV, manually defocus, activate PDAF, open LV, magnify, and observe/compare sharpness

4.) trim AF fine tune thru a range of values and simply discover the value that gives the sharpest result observable in live view

step 2 might not even be necessary

Also, I must be a dodo but I can't see the distinction between "input" and "output" or why a PDAF algorithm can't in fact have both.  ie "how far to go" and "is/is not good once arrived"; reiterate to a tolerance value or x number of times; if within tolerance "don't move".  And I believe I read somewhere that each AF point (which is just a kind of split prism for the purpose of thinking about it) has a bias value individually assigned in some look-up table resident in each camera.  AF fine tune just biases each of those values and those values make it possible to, in fact, actually have the PDAF module in the first place as there would be no way to manufacture to that precision (prohibitively expensive, physically impossible).