EOS 5D and serious AF problem

[pascal]

Hi all 5d users,

This is my first post here (I was an addict in robgalbraith forum and love reading info in Luminous for a long time, so it's time now).

To be honest this is a copy paste from dpreview. I launched this thread 20 hours ago but have too few pro answers.

Just finished a 4 days corporate portraits shooting mixing daylight and fill flash (ste2 + one 550ex with umbrella).
camera : 5d + 135L f2 + monopode

all shots taken at f2 to create max. beautiful bokeh.

all shots taken in portrait mode (vertical grip) using upper focus point in order to help composition and sharp eyes results.

problem : many serious back focus photos !!! see one typical example in my pdf here :
http://www.pascalbarreiro.com/ftp/eos5d/

during last months, I shot many portraits with 5d and 85 f1.8 with the same upper af point and not a problem.

See “typical” portraits here http://www.pascalbarreiro.com/... .../books/09_macha_juin_06/09_macha.html

here http://www.pascalbarreiro.com/... .../photo_hd/598_photo_hd_01/index.html

and here http://www.pascalbarreiro.com/... .../mode/06_mode_ronloyd/06_mode.html

Just before, as a test, I used my brand new 135L with the 5d and ste2 + 550ex, again with the upper af point and “almost” no problem (almost because I noticed sometimes the nose was off but suspected at that time a maybe very small back focus due to calibration, nothing very dramatic to my taste)

Have to say 2 years ago, I encountered severe back focus in the same condition - upper af point - but with the “in-famous” 10d…)

Between the corporate portraits and my test the days before, the only difference was the longer distance between the subject and me (I decided, due to the beautiful building, to compose ala american portrait). IMO, this could be the only explanation why I had good results before and why I had bad ones the days after.

Near the end of the session, I finally decided to only use the center focus point (my brain started to work…) and… results were PERFECT !
Conclusion : this is not a calibration problem

Question : is it again a “typical” problem due to VERY BAD external AF point capability to analyse contrasts like the ones in the 10D ? If yes, why Canon Engineers still create those focus points In other word and without playing with words, WHAT’S THE POINT ?

Or is it a STE2 problem ? (this accessory, appart firing the 550EX is supposed to assist the camera isn’t it ?)
Waiting for your experience before writing to Canon.

Pascal

Someone in dpr wrote : Its in the specs, and in the manual, that focus tolerance for outer AF points is 1 DOF (as opposed to the precision of 1/3 DOF for the cross-type center point).
But could not find anything about 1 DOF versus 1/3 DOF.
And if true, how can I measure 1  DOF ?  
Thank you in advance for your help

[Big Bird]

Your post is a bit unusual. The upper point on the 5D, in my experience is almost useless. It is so small, if you have the least bit of contrast problem, or miss the target perfectly, the thing will hunt.
IMO, Center point is good, using all points is good, but relying on the upper point alone is a mistake, once the focus drifts, you lose too much time getting it back.
I never hit the focus button until I am sure the point is dead on.

[pascal]

Your post is a bit unusual.
The upper point on the 5D, in my experience is almost useless. It is so small, if you have the least bit of contrast problem, or miss the target perfectly, the thing will hunt.
IMO, Center point is good, using all points is good, but relying on the upper point alone is a mistake, once the focus drifts, you lose too much time getting it back.
I never hit the focus button until I am sure the point is dead on.
[a href=\"index.php?act=findpost&pid=76867\"][{POST_SNAPBACK}][/a]

Why unusual ?

The thing never hunt during the session (also due to ste-2 assist beam), that is precisely my problem. I would have preferred a hitting focus point to help me deciding focusing only on the cross type center one. But the "toy" made me understand by focusing quickly that my decision was apparently good (until I saw the result on the laptop 30 mns after !)

what do you mean by "upper point alone " ? if I understand well when you choose one focus point then it is always alone (apart AI servo with the 6 others hidden) no ?

[Big Bird]

Just having a hard time following different points, my "point" about the upper (could be lower as well), is that they are very finicky, and IMO, poor at AF, as you have to dead on or it will hunt very easily.

[gochugogi]

I often use the 5D in the vertical and lock AF with one of the outer AF points. I rarely have any problems as it nails AF 98% of the time with my 24-105 4L or 85 1.8 USM. I don't have the 135 2L. Unlike the above poster I don't find the out AF points prone to missing or inaccuracy, save in low light without AF assist. However, I normally can't hand hold in in low light anyway, so AF assist fixes it.

It seems reasonable that individual AF sensors would need calibration.

[Sheldon N]

I've experienced the same type of results with my 20D and 50mm f/1.4. My personal conclusion is that the exerior sensors are just not as precise as the central AF sensor. You can sometimes even see it happen in the viewfinder - focus using the exterior AF point, then switch over to the central AF point and focus again to see if there's any difference.

My personal believe is that in the type of situation you describe (longer lenses, shallow DOF), it's better to use the central (most accurate) focus point, than to worry about focus/recompose.