40d Landscape/Portrait focus issue....

Hi, today i performed a few a random tests on my 40d around the garden - it has been at Canon elstree [UK] having it's err99 issue fixed, so i just wanted to take a few random shots to make sure it was working ok and also some shots to check how the focussing was as i asked them to check and calibrate if needed.

Well, the test results weren't initially good with the 24-70L - i have never used the lens before as i acquired it whilst the camera was sick, so anyway, i stuck that lens on 1st and was very disappointed with my wide open shots but i'll go into that in another thread..........

So with the disappointment of the new lens getting me down i decided to carry on taking test shots with the 85 f1.8 instead.

I took random shots around the place and all seemed to be spot on until the last few shots that i took of some rusty looking cable that was at an angle...... i figured this would be perfect and would show up any front/back focus clearly

I wrapped a small piece of white nylon tape around the cable as a marker to focus on and was flummoxed by what happened.... Portrait orientation it focussed absolutely perfectly........

PORTRAIT ORIENTATION

I can't even guess what's going on here. It will be interesting to follow this thread. Was the camera mounted on a tripod. Was there any chance you moved after focusing? Were you using one shot or AI Servo? I'm not trying to suggest any hint of a reason behind the described issue, just trying to get a feel for the set up.

Looks like the problem that haunted early samples of the EF 70-300 mm f/4-5.6 IS USM. There it was portrait orientation that caused some part in the lens to move, and ruin the image quality. Canon rebuilt the faulty ones and have now changed the design of that lens, to avoid things like this one.

Surprised to see it on a lens as well built as the EF 24-70 mm f/2.8 L USM, though.

Zonieart wrote in post #4884890
I can't even guess what's going on here. It will be interesting to follow this thread. Was the camera mounted on a tripod. Was there any chance you moved after focusing? Were you using one shot or AI Servo? I'm not trying to suggest any hint of a reason behind the described issue, just trying to get a feel for the set up.

the shots were all hand held in one shot, though i was lent against a fence for extra stability

I really do not believe there will have been any issues caused by me moving slightly [If i did] as i repeated the test several times as i couldn't figure out what was going on..... i took the shots as follows: 12 in landscape = 12 OOF 7 in portrait = 7 in near perfect focus

i took 1st shot in portrait - ok
next 5 were landscape - all OOF
next 4 portrait - ok
next 5 lanscape - OOF
2 more portrait - ok
2 more landscape - OOF

surely too much of a coincidence that only the landscape shots were oof..... like i said, i will test it again tomorrow this time with a tripod.

apersson850 wrote in post #4884929
Looks like the problem that haunted early samples of the EF 70-300 mm f/4-5.6 IS USM. There it was portrait orientation that caused some part in the lens to move, and ruin the image quality. Canon rebuilt the faulty ones and have now changed the design of that lens, to avoid things like this one.

Surprised to see it on a lens as well built as the EF 24-70 mm f/2.8 L USM, though.

It was the 85 f1.8 prime that was mounted when i experienced the issue demonstrated above.........

I had already taken the 24-70 off as i wasn't getting anything satisfied me....

Oops, sorry. I did read, but didn't comprehend.
Well, then maybe... How far away were you? Lenses move internally, or sometimes externally, when focusing. Sometimes that causes a loose part o the lens to move to some position where it shouldn't be. What distance was it focused at?

Yeah let us know what happens after a controlled test with a tripod + remote + MLU.

Seems like it is an issue though, if it's as common as you say to your shots - I'm stumped.

Also, when redoing the test with a tripod, make sure that the target has some contrast and that the contrast part is running through the center of the AF point.

If I had to hazard a guess, I'd guess that the focus sensor was much larger than the indicator in the viewfinder, much like it is on previous models. Here's the 5D's sensor area superimposed on the indicator boxes in the viewfinder (courtesy of Popular Photography Magazine):