EOS M5 front-focus problem

In addition to all of what's been mentioned above, there is the one bugaboo that always haunts any autofocus system:

How does the AF system know exactly where the photographer wants the focus to be?

And, unless the AF "point" is incredibly small, it can never be extremely precise.  The AF "point" or "box" or whatever must, necessarily cover some distance or area.  Further, there's no really good way to achieve extremely precise positioning of the AF area.  Yes, the "touch and drag" is pretty nifty on the M5, but for me, at least, it's still slower and more cumbersome by far than using manual focus.

It often seems that the manufacturers struggle endlessly to achieve autofocus systems that can do better and better, and the users struggle to learn how to use these AF systems to their fullest.  But the user interface never comes close to the ease of use and accuracy that we had 45 years ago with a big, bright, viewfinder and manual focus.

I can turn the focus ring of a manual lens to get exactly what I want far easier and faster than I can position an AF box and then press a button, or fumble with various AF modes and hope what I've selected will magically choose the part of the scene I want to be in focus.  The tracking is cute, and works OK in some situations, but often, it gets lost.

More people should try manually focusing, especially if your camera has focus peaking.  But another problem is that often, lenses optimized for autofocus have a terrible "feel" and response when used in manual focus mode.

So people who have never used old-fashioned manual lenses don't realize what they're missing.  They get a bad impression of manual focus because the lenses they've tried it with are not well suited to it.

I'm really enjoying using my old FD and FL lenses adapted to the M5 because of the focus peaking in the M5 and the eye-level viewfinder.  But when I use the new lenses with the STM AF motors, their manual "focus-by-wire" has time lag and seems to ignore small inputs (too much hysteresis in the reading of the MF rings).  So they're not nearly as satisfying in MF mode as lenses with a mechanical manual focus linkage.  So lens choice affects how well MF can work.

The problem I'm finding with some of my old FD lenses is that they're really not as good as modern lenses.  As nnowak pointed out in another discussion of manual focus, film hid a lot of flaws that our fancy new digital cameras reveal to us easily.  We were not as discriminating back then.

Still, the "feel", ease of use, and satisfaction, for me, is often far better when I shoot in MF mode, especially with lenses that have good, smooth, mechanically-coupled manual focus.

For tracking, etc., I do like AF, of course, but for critical focus in most static scenes and even some fast-action situations, I often find it frustrating, slow, uncomfortable, non-intuitive, etc.  For me, AF is often gimmicky and unreliable.  Then again, I'm not a sports fan or sports shooter these days.

I did shoot plenty of indoor basketball, and other sports, though when I was a newspaper photographer back in the '70s, and it was all done with manual focus, TRI-X, and sometimes "push processing".  Grain?  What's that?  The halftone hides it! 

What today's sports photographers capture is amazing, though.  And good AF is responsible for a lot of it.  So I can't deny the utility of AF in some situations.  But I'm almost never shooting in those situations!

I will admit, though, that I've use servo AF with macro lenses to some advantage at times, shooting bugs on plants, in the wind, etc.  But I've never really felt that it had a huge advantage over MF and good timing.  It's fun to have these modes at our disposal, but the user interfaces still seem clunky with a lot of it.

I think the real future, for me, is to see further enhancements of focus peaking and simply use MF when I want very precise focusing, but use autofocus for situations where that is appropriate.

One improvement to current focus peaking displays might be for the lens itself to "know" about any focus shift that occurs when stopping it down, and to communicate that information to the body so that, when manually focusing, you could have the advantages of full-aperture focusing, but corrected to display the peaking highlights where the center of focus will actually be when the shot is taken.  Then again, as you stop down, the DOF increases to hide that shift, so I'm not sure how much of an issue it is, in practical use.

Another improvement, for me, would be to have the focus peaking be "more critical" yet give an easily visible display of the sharp-focus locations.  It would be nice to have some user adjustments for all of that.

But I hate to complain, actually.  I have found MF with the M5 with its focus peaking to be a joy.  So I think they're on the right track with that.

Maybe when they come up with a sensor that has four microlenses and four photosites "per pixel" so that you get virtual cross-type AF "sensors" everywhere (QPAF?) things will be even better.  Both for AF and for focus peaking.

You can really see how the DPAF only "sees" vertical features when you use focus peaking!  But it's still very neat for what it is.

Autofocus will always be a compromise.  It's handy for its speed, tracking, etc., but it still fails to read my mind.  Even with a contrast-detect "touch up" at the end of each AF cycle, it will do no good if it doesn't know exactly what, in the scene, I want to be the point of sharpest focus AND have the ability to home-in on just those tiny points.

I think we can pixel peep and see how the AF has missed the focus in our photos, but for me, I can never be sure if the AF "missed" or if I just didn't communicate the location I wanted to be sharp to the AF system.

Although I never tried one, Canon had eye-controlled AF in some models.  That might not have been successful, but at least they were trying to achieve a better interface between the photographer and the AF system.