Scientific Inconsistent Focus with 10D

[Michael Leuzinger]

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with crisp markings) on a table in a diagonal stagger. Took many shots with many settings. In EVERY case, the marker BEHIND the one I was aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and 50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the D30. Well guess what... now ALL of them focused accurately. Now even the 10D and 50/1.4 combo focused just fine. The only difference was the lighting. To make things worse, the lighting was better (brighter) at work.

So, same equipment, same day, same test... and there was serious back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send it to Irvine. But now what? Can anyone explain what might be going on here?

 

[FUCK DPREVIEW!!]

I've done two fairly well-controlled front/back focus tests...

I see... ...have you taken any "real" pictures with your 10D yet? If yes, did you experience any "focus issues"?

Andi

 

[Adam2]

You raised a very important point: lighting.

In other discussions, people have mentioned the effect of the type of lighting on the AF. Exactly, what this is, I don't know. But the repeated mantra is "take it outside". Sure enough, in daylight, the AF problems seemed to disappear.

One of the significant improvements touted on the 10D is its improved AF. While I have noticed that the Servo mode is indeed better (for eg. I was much happier yesterday taking some indoor, ambient light, action shots of swimming), the jury is still out on the accuracy of the AF.

If indeed the temperature of the light significantly affects the AF, then a serious problem exists. I hope that Canon will come clean on this issue, one way or another.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Noel Carboni]

This is probably a stupid question, but are you using one focus point? Surely you're not letting the camera choose what to focus on.

It is possible that the focus sensor and the little square(s) in the viewfinder are not 100% perfectly sized/aligned. My D30 used to focus "around" the focus point, especially if I was shooting into, say, a tree with a lot of twigs at different distances.

I use the center focus point only, and reframe after locking focus. FWIW, all my 10D images are coming out perfectly focused.

-Noel

 

[davidhood]

Using AI servo cleared up a lot of my focus problems.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Mike Leuzinger]

Good question Andi.

Yes, theses tests definitely help explain the issues I've seen with "real" pics.

I was one of the people who swore the camera had a "break-in" period. Pics the first couple of days were horrible, but then much better after that. Looking back, I think two things were at play...
1) luck

2) that I used the 50 1.4 more the first couple of days, and the 70-200 2.8L more the next couple of days.

So far usage with the 70-200 2.8L seems fairly consistent... but usage with the 50 1.4 is hopelessly inconsistent.

The 50 1.4 has been very consistent with my D30.

I've done two fairly well-controlled front/back focus tests...

I see... ...have you taken any "real" pictures with your 10D yet?
If yes, did you experience any "focus issues"?

Andi

 

[Mike Leuzinger]

Yeah. I'm going to be away for a couple of days, but when I get back I'll try some tests with various lighting temps (Tungsten, Flourescent, Halogen, Sunlight) etc. and report back. I definitely need to get this figured out by my Munich trip in May and hopefully by my San Fran trip on April 12th. I was hoping to use the 10D instead of the D30 for both trips since the extra resolution will help greatly with architecture/landscape photos!

Lighting wise, here's how the 2 tests compared:

Test 1: Bright Flourescent lighting
Result: Consistently back focused

Test 2: Dim Tungsten lighting
Result: Focus consistently accurate

I'll try an outdoors test tomorrow.

In other discussions, people have mentioned the effect of the type
of lighting on the AF. Exactly, what this is, I don't know. But
the repeated mantra is "take it outside". Sure enough, in
daylight, the AF problems seemed to disappear.

One of the significant improvements touted on the 10D is its
improved AF. While I have noticed that the Servo mode is indeed
better (for eg. I was much happier yesterday taking some indoor,
ambient light, action shots of swimming), the jury is still out on
the accuracy of the AF.

If indeed the temperature of the light significantly affects the
AF, then a serious problem exists. I hope that Canon will come
clean on this issue, one way or another.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Mike Leuzinger]

Hi Noel,

Yes, I have it set to only use the center focusing point and set to One Time AF.

You could be on to something with the viewfinder not being aligned, but if that's the case, it must be fairly way off. I've tried different tests with the diagonal pieces close and far from each other and it's the same result. Also, it back focuses even if the only other objects are vertical (higher) OR horizontal (to the right), so it would have to be out of alignment fairly significantly to the upper right.

I'll try the tests again with the farther objects to the left and bottom which should verify if it's a focusing screen alignment issue.

This is probably a stupid question, but are you using one focus
point? Surely you're not letting the camera choose what to focus
on.

It is possible that the focus sensor and the little square(s) in
the viewfinder are not 100% perfectly sized/aligned. My D30 used
to focus "around" the focus point, especially if I was shooting
into, say, a tree with a lot of twigs at different distances.

I use the center focus point only, and reframe after locking focus.
FWIW, all my 10D images are coming out perfectly focused.

-Noel

 

[David Waterman]

Did you notice any shots being out of focus after focus had locked (beep!) when viewed through the view finder?

Max.

Yes, I have it set to only use the center focusing point and set to
One Time AF.

You could be on to something with the viewfinder not being aligned,
but if that's the case, it must be fairly way off. I've tried
different tests with the diagonal pieces close and far from each
other and it's the same result. Also, it back focuses even if the
only other objects are vertical (higher) OR horizontal (to the
right), so it would have to be out of alignment fairly
significantly to the upper right.

I'll try the tests again with the farther objects to the left and
bottom which should verify if it's a focusing screen alignment
issue.

This is probably a stupid question, but are you using one focus
point? Surely you're not letting the camera choose what to focus
on.

It is possible that the focus sensor and the little square(s) in
the viewfinder are not 100% perfectly sized/aligned. My D30 used
to focus "around" the focus point, especially if I was shooting
into, say, a tree with a lot of twigs at different distances.

I use the center focus point only, and reframe after locking focus.
FWIW, all my 10D images are coming out perfectly focused.

-Noel

 

[Mike Leuzinger]

A) I'm surprised you've had better results with AI servo

B) Pretty sure you can't use AI Servo if you want to hold focus while recomping... a necessity for most of my still shots.

I'll give it a try though just to compare. ;-)

BTW, I'll be trying out AI Servo a lot when I get a chance to hit the track! ;-)

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Mike Leuzinger]

The focus had locked (beep) in ALL of the test shots. The camera definitely thinks it has a focus lock, it just happens to be focusing farther back than what it thinks it is.

But I can live with back focusing... after all, that's a clear cut case of sending the camera in for repair. What is driving me nuts is that it only does this some of the time even in controlled tests (possibly due to the lighting temp as mentioned by someone ealier in this thread)... we'll see.

Max.

Yes, I have it set to only use the center focusing point and set to
One Time AF.

You could be on to something with the viewfinder not being aligned,
but if that's the case, it must be fairly way off. I've tried
different tests with the diagonal pieces close and far from each
other and it's the same result. Also, it back focuses even if the
only other objects are vertical (higher) OR horizontal (to the
right), so it would have to be out of alignment fairly
significantly to the upper right.

I'll try the tests again with the farther objects to the left and
bottom which should verify if it's a focusing screen alignment
issue.

This is probably a stupid question, but are you using one focus
point? Surely you're not letting the camera choose what to focus
on.

It is possible that the focus sensor and the little square(s) in
the viewfinder are not 100% perfectly sized/aligned. My D30 used
to focus "around" the focus point, especially if I was shooting
into, say, a tree with a lot of twigs at different distances.

I use the center focus point only, and reframe after locking focus.
FWIW, all my 10D images are coming out perfectly focused.

-Noel

 

[Derek Hawkins]

It's the nature of a poorly adjusted 10D (or any Canon DSLR for that matter). Put the camera on a tripod and note the movement in the len's distance display window while half pressing the SRB several times. You will see that it doesn't focus at the same location every time even though the subject/camera hasn't moved. Typically there is "slop" of about .5mm in the focus mechanism. Even a properly adjusted camera cannot achieve exact focus at either extreme of the movement (this would be impossible) but the difference should be minimal and within the specification for the camera. However, a poorly adjusted camera can easily swing from front to back focusing depending on where the mechanism settles. This explains why one day or for a series of shots all seems to be well and the next it's OOF. And several "corollaries" of Murphy's Law dictate that things will happen in such a way as to confuse the daylights out of you. Other than having Canon service it you can do the following;

1. Take several shots of subjects (completely releasing the SRB between shots)
2. Stop down the aperture as much as light or the composition allows

The above applies to just about all AF cameras.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[David Waterman]

The focus had locked (beep) in ALL of the test shots. The camera
definitely thinks it has a focus lock, it just happens to be
focusing farther back than what it thinks it is.

OK, that's one thing - but did the shot look in focus through the view finder.

That's one thing I noticed was that they all looked like they were in focus, but they weren't after the shot was taken.

Max.

 

[Mike Leuzinger]

While I'm doing my testing, I'm very interested...

Has anyone else noticed color temp (not brightness) affecting focus?

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Mike Leuzinger]

Yes, they looked in focus as well... although to be honest, I'm not entirely sure if I would have noticed that the farther away object looked just as in focus or possibly even in better focus through the relatively small focusing screen.

The focus had locked (beep) in ALL of the test shots. The camera
definitely thinks it has a focus lock, it just happens to be
focusing farther back than what it thinks it is.

OK, that's one thing - but did the shot look in focus through the
view finder.

That's one thing I noticed was that they all looked like they
were in focus, but they weren't after the shot was taken.

Max.

 

[Mike Leuzinger]

Good point, but it is shocking how inconstent my 10D is when compared to my D30.

I'll do more testing to see if your "slop" theory plays out.

Really wish Chuck Westfall was still around for issues such as this... I don't trust the average front-line Canon service rep (I don't trust ANY industry/vendors front-line service reps... not that some aren't great... but I've worked the biz... and I know how many newbies there are. ;-)

1. Take several shots of subjects (completely releasing the SRB
between shots)
2. Stop down the aperture as much as light or the composition allows

The above applies to just about all AF cameras.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[Derek Hawkins]

that sometimes a 50mm f/1.4 @ f/8 would be totally OOF. It was so bad that I could tell from the image in the viewfinder that it was OOF before even taking the shot. Once in a while it would be dead on. Corrected it myself using the Service Manual CD. I gather the CD for the 10D isn't available yet.

I'll do more testing to see if your "slop" theory plays out.

Really wish Chuck Westfall was still around for issues such as
this... I don't trust the average front-line Canon service rep (I
don't trust ANY industry/vendors front-line service reps... not
that some aren't great... but I've worked the biz... and I know how
many newbies there are. ;-)

1. Take several shots of subjects (completely releasing the SRB
between shots)
2. Stop down the aperture as much as light or the composition allows

The above applies to just about all AF cameras.

Okay... I admit it, my head hurts.

I've done two fairly well-controlled front/back focus tests with
the 10D and the 50/1.4.

Did the first test at work today. Set up 4 dry erase markers (with
crisp markings) on a table in a diagonal stagger. Took many shots
with many settings. In EVERY case, the marker BEHIND the one I was
aiming at was the one in focus.

Then I came home, and decided to try the same test with the 10D and
50/1.4, 70-200 2.8L, and Sigma 15-30, AND all 3 lenses with the
D30. Well guess what... now ALL of them focused accurately. Now
even the 10D and 50/1.4 combo focused just fine. The only
difference was the lighting. To make things worse, the lighting was
better (brighter) at work.

So, same equipment, same day, same test... and there was serious
back focusing in brighter light but accurate focus in dimmer light.

I would rather it would always back focus so I would know to send
it to Irvine. But now what? Can anyone explain what might be going
on here?

 

[doctor digi]

Okay... I admit it, my head hurts.

Has anyone ever thought about the possibility that these lenses, being mechanical and being controlled by a motor, will NOT give reproducable results?

Personally I think AF is just inaccurate if you are trying to get it down to a small percentage of the total distance to the subject. There is plenty of "play" - both mechanically and in the sensors.

I'm willing to be proved wrong, but from what I've seen with my 10D and from all these tests, it would appear that people are wasting their time trying to get it spot-on.

dd

 

[doctor digi]

I should also add that this is why the "reset" appears to work for some people. They test the camera, apply the "reset" and hey, magically it focusses correctly when they retest it. I imagine it would have done so anyway without resetting the camera.

Likewise, I'm willing to bet some people have found the reset makes their camera "worse"... perhaps not posting that up on here though!

dd

 

[Derek Hawkins]

a properly adjusted 10D should give reproducible results with any lens IMO. I have two D60s capable of better than that and if anything the 10D can equal or better that. Having to stop down more than two stops just to correct AF issues could mean losing control over your composition. All indications are that Canon has been sloppy in this area due to the lack of proper feedback. Their response would probably be similar to yours and no doubt they would recommend a 1D or 1Ds as a solution.

Okay... I admit it, my head hurts.

Has anyone ever thought about the possibility that these lenses,
being mechanical and being controlled by a motor, will NOT give
reproducable results?

Personally I think AF is just inaccurate if you are trying to get
it down to a small percentage of the total distance to the subject.
There is plenty of "play" - both mechanically and in the sensors.

I'm willing to be proved wrong, but from what I've seen with my 10D
and from all these tests, it would appear that people are wasting
their time trying to get it spot-on.

dd