Autofocus expectations vs. reality

Eye controlled focus! Go here...https://photography-on-the.net …ight=eye+contro​lled+focus

John from PA wrote in post #16303294
Eye controlled focus! Go here...https://photography-on-the.net …ight=eye+contro​lled+focus

There's a fundamental flaw with ECF:

You can't look around your viewfinder to pay attention to details. As soon as you start looking at various parts of your scene, the focus point changes.

kurt765 wrote in post #16301530
I am curious what people expect out of their autofocus system. I recently read someone's comment somewhere that mentioned a "focusing on a twig issue when shooting birds" and it got me wondering.

So if the first thing that the camera arrives at in focus is a twig that is in front of a bird, then why would I be surprised at this result if I did not tell the camera where to focus? I guess I am trying to decipher what the problem could be other than the person expected the camera to magically know what was expected to be in focus and focused on something else instead. I

I suspect that the person complaining about the camera focusing on a twig probably put the little red box directly on the bird - only to have the AF select a twig instead. This usually happens when the bird we want to focus on does not have a lot of contrast in its plumage - the AF will ignore what we are telling it to do because it is not sensitive enough; it cannot recognize the small amount of contrast in the feathers of the bird. It can only recognize the greater amount of contrast that it "sees" between the twig and the twig's surroundings.

What I expect, or rather want, in an autofocus system is that it is sensitive enough to "see" whatever it is I am putting the red box on, even in very low light, and even if there is only a wee bit of contrast to "grab onto".

cdifoto wrote in post #16303316
There's a fundamental flaw with ECF:

You can't look around your viewfinder to pay attention to details. As soon as you start looking at various parts of your scene, the focus point changes.

No, it doesn't. The eye controlled focus reads which AF point you look at when you half-press the trigger button/press AF-ON, and then that point remains locked, even if you look elsewhere.

It's only when using Servo AF you'll get the effect of tracking what you look at, but that's usually a good thing, and if not, then you have to disable the eye control at that time.

I have one of the better eye control implementations in one of my EOS cameras, so I know.

apersson850 wrote in post #16303576
No, it doesn't. The eye controlled focus reads which AF point you look at when you half-press the trigger button/press AF-ON, and then that point remains locked, even if you look elsewhere.

It's only when using Servo AF you'll get the effect of tracking what you look at, but that's usually a good thing, and if not, then you have to disable the eye control at that time.

I have one of the better eye control implementations in one of my EOS cameras, so I know.

Assuming it reads the correct point. Your EOS might be far more amazing than mine but mine only has 3 points and it still gets confused. Yeah, I calibrated. It still doesn't work in practice as well as theory led me to believe. It'd be hell with these 45-pointers.

It didn't exactly go away because it was awesome.

cdifoto wrote in post #16303662
Assuming it reads the correct point. Your EOS might be far more amazing than mine but mine only has 3 points and it still gets confused. Yeah, I calibrated. It still doesn't work in practice as well as theory led me to believe. It'd be hell with these 45-pointers.

It didn't exactly go away because it was awesome.

Cdifoto you and I had the same bad experience with ECF, however I had mine with the 45 points and it was a constant battle fighting it to select the focus point I wanted. Turned it off eventually and went back to normal focus point selection.

cdifoto wrote in post #16303662
Assuming it reads the correct point. Your EOS might be far more amazing than mine but mine only has 3 points and it still gets confused.

Mine has seven AF points (it's the EOS 30), and it very rarely picks the wrong point, in spite of me using glasses. But you have to calibrate it in different light conditions, since it stores more than one calibration for each user. If you calibrate outdoors, in sunshine, that's not appropriate for indoor shooting. But you can calibrate indoors too, and it will be added to the database of calibration data for that user. After a few different calibrations, it's very precise.
Just a pity you have to load film in it to use it.

mccamli wrote in post #16301829
One legitimate issue might be that the actual focus points extend beyond the boxes in the viewfinder. Aim a little square at an eye and the cross hair may pick up the nose/ear/lips. Another legitimate issue might be that the focusing point needs decent contrast to work effectively. Aim the focusing point at someone's cheek and the camera may decide to focus on an eye/nose/ear.

These things are obvious when you know how the system works but may be a surprising/confusing to the uninitiated or intellectually challenged.

For me, the issue isn't knowing how it works (tho I don't claim to be an expert) but how it works If I don't want to focus on the nearby point of highest contrast, I'm screwed. This is novel and odd to me as a long-time MF film shooter and I don't always like it.

mccamli wrote in post #16301829
One legitimate issue might be that the actual focus points extend beyond the boxes in the viewfinder. ...

One way to find out for sure is to put a 1 inch long line on a piece of blank paper. Tape it to a wall and test your AF points. Don't forget some can only see vertical and some can only see horizontal. I did that for my T1i, because it missed focus often when using the top-center square. The 'real' AF point was about 50% in the square and 50% above it. Manufacturing tolerances and all that.

YashicaFX2 wrote in post #16305216
The 'real' AF point was about 50% in the square and 50% above it. Manufacturing tolerances and all that.

I've only really looked into the 7D's AF points and each detailed review had pretty much the same results.

I suspect it has little to do with manufacturing tolerances and much more likely that sales & marketing departments overruled the engineering & technical departments.

I am sorry, but I don't have a clue how what you posted is related to the quote you included.

YashicaFX2 wrote in post #16305418
I am sorry, but I don't have a clue how what you posted is related to the quote you included.

Just the fact that you indicated that the reason why the real AF point didn't match the indicated AF point was due to manufacturing tolerances. I suspect they don't match because the real AF points wouldn't look so appealing.

I'd much prefer cross hatches that mirrored the real thing than little uniform squares a uniform distance apart.

Some cameras are better in AF consistency and some are worse. I know I am going to get lynched for saying this but when I had two Canon 7D's the AF consistency on both of them were the not very good in the basic one shot mode center point focus. Even after numerous trips to the service center they were not up to my expectations because I found my XSi had better AF consistency. After fighting it for a year I sold the 7D and got a 60D. Yes it was a downgrade but the AF consistency was a lot better. Now I'm thinking getting a 70D because there hasn't been any complaints about the AF.

I'm using middle AF point with back button focus, so I only have myself to blame if I miss.

mccamli wrote in post #16305457
Just the fact that you indicated that the reason why the real AF point didn't match the indicated AF point was due to manufacturing tolerances. I suspect they don't match because the real AF points wouldn't look so appealing.

I'd much prefer cross hatches that mirrored the real thing than little uniform squares a uniform distance apart.

Got it. Thanks. And yes, I agree with your last statement.