Focus discrepancies in Rebel XT and XTi, manual and auto focus

I have noticed the most puzzling and disturbing focus discrepancies in
both my Rebel XT and my XTi.

Please do the following tests with yours (and please report back!):

Set your camera on a tripod
Set up a scene with some sharp black-on-white writing in the center
Set your lens to auto-focus and press the shutter button half way down
until the lens focuses
Adjust your eyepiece dioptric to your eyesight
Shoot a picture at wide aperture (f2.8 will do)

Now, without disturbing anything, set your lens to manual focus
Move the focusing ring to dislodge the previous focus setting
Re-focus carefully on your viewing screen (do not re-adjust your
diopters!)
Shoot a picture at the same f-stop

Now upload your pictures to your computer and compare the two
side-by-side by enlarging the center to the maximum (try "actual
pixels" with Photoshop)

I've repeated these tests with every conceivable variation of lenses,
cameras, with glasses and without, with additional supplementary
diopter correction lenses, etcetera.

INVARIABLY, the picture done in auto-focus is right on, but the one
done manually shows a pronounced focus shift! You can readily see the
difference in sharpness between the two.

I sent the XTi to Canon and it was returned with the same problem,
after thorough "inspection and adjusting".

What is going on?

Do you see the same discrepancies in your own tests?

 

They didn't find a problem since this is normal.

The focusing screen in your camera is not optimized for manual focus. So,
every time you see an image that looks perfectly focused you have too much
tolerance for it not to be perfectly focused.

No need to. If you want to use this body with MF lenses you should invest
in a KatzEye screen if they make one for that model.







Rita

 

If that's the case it's a shame. I can't really believe it. Do other
brands show the same problem?
And why is not this a problem with film auto-focus Canon SLRs? Or is
it?
And why isn't there a warning anywhere in the instruction manual?
I personally like to set my lenses to manual focus, I hate to press a
"hold" button and refocus if I don't agree with the camera's chosen
point.
That renders the Canon Rebel XT and XTi near useless to me.

 

BTW, I spent two hours on the phone with various Canon technicians this
afternoon.

They seem to think the problem is with my eyes, that I need to adjust
the eyepiece diopter.

Never a word that this baffling camera behavior is "normal".

That's why I'd like to know if other users are getting the same results
from the tests.

 

I think the current trend is most people use AF and don't play around with
MF much. Yes, other brands have the same problem to more or a lesser
degree. Your camera is working within 100% of factory-designed specs. I
found the website for KatzEye.

http://www.katzeyeoptics.com/

It depends on the factory installed focusing screen. Plus, higher end
cameras give you a choice of screens.

It would confuse people since most people use AF. It's kinda like the
debates we have here about IS/VR were some people feel this feature cures
everything from the common cold to gout.

I know you don't want to hear this, but if you plan on using MF a lot you
will need to shell out $100+ and get the KatzEye.

You'll get used to AF since everybody does sooner or later.







Rita

 

:I have noticed the most puzzling and disturbing focus discrepancies in
: both my Rebel XT and my XTi.
:
: Please do the following tests with yours (and please report back!):
:
: Set your camera on a tripod
: Set up a scene with some sharp black-on-white writing in the center
: Set your lens to auto-focus and press the shutter button half way down
: until the lens focuses
: Adjust your eyepiece dioptric to your eyesight
: Shoot a picture at wide aperture (f2.8 will do)
:
: Now, without disturbing anything, set your lens to manual focus
: Move the focusing ring to dislodge the previous focus setting
: Re-focus carefully on your viewing screen (do not re-adjust your
: diopters!)
: Shoot a picture at the same f-stop
:
: Now upload your pictures to your computer and compare the two
: side-by-side by enlarging the center to the maximum (try "actual
: pixels" with Photoshop)
:
: I've repeated these tests with every conceivable variation of lenses,
: cameras, with glasses and without, with additional supplementary
: diopter correction lenses, etcetera.
:
: INVARIABLY, the picture done in auto-focus is right on, but the one
: done manually shows a pronounced focus shift! You can readily see the
: difference in sharpness between the two.
:
: I sent the XTi to Canon and it was returned with the same problem,
: after thorough "inspection and adjusting".
:
: What is going on?
:
: Do you see the same discrepancies in your own tests?
:
Listen to what Rita says here Bob. Your camera is an Auto Focus camera. If
you can't get a sharp pic on AF, it has a back focus error. This is not a
Professional level camera and you can't expect it to perform like one. If
you truly do want to manually focus this camera, get a manual focus screen
for it. You may have a problem then with detecting correct AF but it's all a
compromise on these cheap SLRs.

--

Australian Wedding Photography between Kempsy, NSW and Sunshine Coast.
http://www.photosbydouglas.com
Digital photos enlarged and printed on Canvas
http://canvas.photosbydouglas.com

 

My problem was DEFINITELY fixed with the diopter. Shooting through a
viewfinder only, I can never trust my own judgement as to whether I've
manually focused correctly or not.

I would not be a good subject for your proposed test...

dwight

 

Which brings up a question I've been meaning to ask for some time. How
accurate is the built-in diopter adjustment on a low end Canon? I have a
Rebel XT and I'm -.5 in one eye and -.75 in the other.

And if I start doing more manual focus I will definitely get a Katz Eye,
I've been thinking about it for quite some time.

Greg

 

My test removes the possibility of a diopter problem.
Notice that in my test the diopter is calibrated after the camera has
autofocussed
Using manual focus then should give the exact same point of focus,
since the eyepiece has been corrected for your eyesight already.

 

A Katz Eye will not correct an inherent focus discrepancy, should you
have one.

These are screens whose only virtue is that they're brighter and more
contrasty and thereby facilitate focusing, PROVIDED you have a
perfectly aligned system to begin with.

 

with the camera turned off, can you see the focusing squared clearly?
Watch them while you adjust the diopter correction.

 

Yes, but as others have pointed out the lack of a decent focusing screen on
current autofocus DSLRs is probably your problem with manual focus, not an
inherent problem with your Rebels.

Greg

 

And by what mechanism do you suspect the focusing error occurs?
As I have explained, in the test procedure the diopter is adjusted
after the camera has autofocussed the lens. The lens is then refocused
manually until the image, already adjusted for my eyesight, again
appears tack sharp. The second picture is then taken.
How do you explain the fact that the image focused on auto is right on
but the image focused manually shows a very perceptible focus shift?
It is quite obvious to me, and I fail to see why it isn't to everybody,
that the focusing screen brightness plays absolutely no role in the
test procedure.
Please explain to me then how the focus discrepancy occurs.
Better yet, do the test on your own Rebel XT, it won't take you 10
minutes.

PS - I have also observed that the manual focused image INVARIABLY
shows 1/3 of a stop more exposure than the autofocus image. Explain
that one also.

 

As "Rita" said:

"The focusing screen in your camera is not optimized for manual focus. So,
every time you see an image that looks perfectly focused you have too much
tolerance for it not to be perfectly focused."

I'm not going to do the test because I don't trust the combo of my eyes and
my Rebel XT's screen.

PEBSAC? Can't think of any other reason. How do you explain it?

Greg

 

Yes, after adjusting the diopter on the autofocussed image I can then
see the viewfinder perfectly with the camera turned off.
I don't see why the reverse would help any - I probably did adjust it
that way when the camera came out of the box, which, BTW, was when I
first noticed that the screen was a tad off when focussed at infinity
and what started me on this unpleasant voyage.

 

The screen could be in the wrong place, such that when the image is
focused properly on the viewfinder screen it is not focused properly
on the imaging sensor. One of the Photo magazines, either Pop or
Modern ran an article on this problem many years ago, back in the days
of film.

since what is seen in the viewfinder is not an aerial image I can't
figure how the diopter or bad eye could make a mis-focused image look
good. there's a lot of things I can't figure out, this is just
another one of them.

 

Agreed. I've always had MF problems with my D30/D60/DRebel. Amazingly, I
can live life normally like you and most of us.

I saw the light and am giving up on Sigma and Tamron permanently due to the
superior AF on Canon lenses.

 

And have a split prism and a microprism ring.

As for an "inherent focus discrepancy", how many SLRs have you
encountered that actually have one of these? On every SLR I've taken
apart you're going to to have to bend the primary structure to get an
"inherent focus discrepancy" . On a Leica the rangefinder can get
misaligned without a whole lot of difficulty, on an SLR you have to
quite a lot of damage.

 

In the procedure you are applying you should be adjusting the diopter
until the image is tack sharp, not adjusting the diopter then changing
the focus to accomodate your misadjustment of the diopter.

Of course it may be that you have the focusing screen in upside down.

If you can see 1/3 stop more exposure you've a better eye than the lot
of us I suspect.

 

I can just barely see it in the image - but I can see the shutter speed
in the camera below the screen change from 1/20 sec to 1/15 sec (I've
been shooting on "A") at the same aperture.

I'm told by a knowledgeable friend that these cameras use two different
systems to focus (and probably to determine exposure thereby) - one
system is independent of the viewing screen and uses intersecting beams
(!?) and I don't know what to focus the image; the other uses the
traditional groundglass screen and top prism to focus and possibly also
to determine exposure.