Ok, this has been driving me crazy for some time now, both with my 350D and now my 30D. Sometimes I can get nice, sharp pictures like the ones I posted in the "Wildlife" section, which where pretty good with a 70-300, and I've gotten some nice images with the rebel and a (rented) 100-400L, but more often than not, I get pictures like the one below (the only "good" lens I have is the 28-135 IS USM). In this image, I was using my center AF only, the kitten was stationary, as was I, flash fired with a shutter of 1/250. AF the bottom line of the AF box, if drawn further left and right, would go through the center of the kittens eyes (the box was almost exactly centered between the eyes). The problem, as you can see, is that the kittens nose/face/eyes are not in focus, but it appears as if the fur on the kitten's back is more in focus. To tell the truch, I don't think I've ever gotten a good image with the 28-135 (I was VERY dissapointed with at least 95% of my honeymoon pictures, all of which where taken with the 28-135, and regretably, only jpegs exist; From now on, especially with the "discovery" of LR, I'm shooting RAW only). Even landscapes where out of focus. I would think that once it gets out past 50', into the "infinite" range (of the 28-135), that the focus would be less sensitive than inside 50'. The thing that threw me for a loop with the Rebel, is a while back, I had my hands on another Rebel; I took two pictures of the same subject, at the same ISO, same sutter speed, same everything except the body, and the image from the other Rebel, while still not "razor sharp", was noticably better than the one from my Rebel. The lens has been to Canon twice; Once to CA and once to NJ. The first time they supposedly replaced the IS assembly (and introduced dust into the lens while it was there), and the second dime I think they said nothing was wrong.

I know I have a lot to learn, and I don't come even close to holding a candle to pretty much anyone here, but with an auto focus camera and lens, one would think that the focus should not be my issue, especially considering I've seen much sharper images form P&S cameras that cost 1/4 (or less!) what this camera and lens would cost new (I bought the camera from another member here, but the lens I bought new).

I'm open to any suggestions fo find out if the problem is me, the camera, lens, or some combination thereof. Would buying a 50mm 1.8 be a start?

Your lens seems to be backfocusing. The front paws of the kitten seem to be sharper hence in better focus than the target. I am not sure whether you focussed and recomposed or not but using a 28-135mm lens with a fairly small aperture, you should not have any issues with this technique like you would with a fast lens at a large aperture. The bottom line is, it looks like you have to have the AF of the lens calibrated by Canon .

What AGeoJO said.

If it's not your lens, perhaps you should experiment with different focus-area selections.

Start with center-focus selected, and if still no joy, try manual focus (assuming that your mirror is correctly aligned, which is a cr@p-shoot with Canon.)

Note that in your pic, above, the center of the image is perfectly in focus. That's the cat's neck fur.
So maybe you ARE using center-focus and need to either change that a wider grid, or use the focus-hold function (holding the shutter button halfway down then panning)

Personally, I like to shoot manual-focus, even with my Canon lenses, because having the wrong focus-area selected can bite you in the butt. I also hate to have to focus, focus lock, then re-compose, if my subject isn't centered.
-------

If all fails, then your lens is out in some way, or (worst case) your camera's mount is mis-aligned. I would send both the lens AND camera to Canon for calibration, just to be safe.

I just shot a few with the 28-90 F/4-5.6 III kit lens I have (came with my old Rebel T2 which died), and they came out a little better, but not great. The right plane appears to be in focus, just not all that well when zoomed in. That lens may just not be all that high quality, though. I just printed an 8x10 and 5x7 of one of the pictures taken with the 28-90, and they look pretty good there (despite the fact that my Laserjet 4550 is NOT a photo printer ), so I'm guessing that this particular lens wan't really designed to give great results on a 8MP DSLR.

As a note, the pic above is cropped. In the original, the bridge of kitten's nose right between the eyes was dead center of the frame, and the location of the center AF point, and the aperture was 5.6. I got focus confirmation (pressed shutter half way) then took the picture without panning. As a general rule, I always use the center AF point (I like knowing which point it's going to try to use for AF). What do you mean by "change that a wider grid"?

The problem with manual focus is the screen; IMO, it's a BEAR to try to get the focus right with these "matte" screens, but I hear what you're saying. I never had problems with my old Minolta XG-M and a split screen though (similar to the "Katz Eye Screen).

What kind of "subject" should I use to try to get the best detail and make it "easy" to tell if the focus plane is where it should be?

Cougar 281,
I just want to make sure that you understand. If you press the shutter button halfway and the lens focuses and when it thinks it reaches the target focus it will lit up and you hear a beep (if opted for that, of course). Again the lens thinks it reaches the target focus but if the AF is not properly calibrated or got out of whack for whatever reasons, it doesn't actually focus on the target. The AF confirmation light and beep do not guarantee an actual accurate focusing on the target.

I would suggest for you to do a simple AF test. Use a straight ruler or tape measure and set it at 45 degree angle from your camera setup, which should be on a tripod, take a picture of the ruler at a point of your choice at the widest opening. You can download a focusing target from the internet if you will. Take a look at the resulting image and you can easily tell whether your lens backfocus or even front focus. Take a few shots but each time focus the lens away in between shots. You can do this with your lenses. If the results of your 28-135mm lens has pretty consistently disappointed you, I bet you almost anything, the lens will probably backfocus, judging by the kitty picture.

Hope this helps,
Joshua

I haven't gotten a chance to do the AF test on the 28-135 yet, but I picked up a 50mm f1.8 at a local shop yesterday, and WOW.. I got a few SHARP images (far shaper than any I've gotten from the 28-135). Main drawback I can find is at f1.8 or so, the focus plane is VERY shallow. I'll post back when I do the AF Test with the 28-135.

Cougar281 wrote:
with an auto focus camera and lens, one would think that the focus should not be my issue, especially considering I've seen much sharper images form P&S cameras that cost 1/4 (or less!)

Hehehehe.... i totally hear you there... auto focus shouldn't be an issue if it works as advertised. Quite often, even on certain high end cameras, it fails- and often not necessarily due to user error.

What AF mode are you shooting in?
It seems to be quite dark in that kitten picture- do you find that it happens only when dark, or under any situation? Have you noticed that it happens at large apertures more often than small? Do you have the AF assist beam on?
It definitely seems to be backfocusing- when I had a 30d, it too back focused on me quite a bit. I found that out after shooting focus targets with all my lenses, and comparing them to those shot on the 40d. With Canon equipment, sometimes it's not just one element that is mis calibrated- it could be both, thus accentuating your problem. I realize its really annoying to buy such expensive equipment to realize that it doesn't function properly, so I am more than willing to help contribute to solve your problem.

My 350D front-focused quite annoyingly when i got it. Took a while (and buying a 50mm F1.8) before i diagnosed the problem as not being either me or the autofocus just missing the target. Sent it back to canon and they tweaked it under warranty. The autofocus is still crap, but now when it hits, it hits.

orangefirefish wrote:
What AF mode are you shooting in?
It seems to be quite dark in that kitten picture- do you find that it happens only when dark, or under any situation? Have you noticed that it happens at large apertures more often than small? Do you have the AF assist beam on?
It definitely seems to be backfocusing- when I had a 30d, it too back focused on me quite a bit. I found that out after shooting focus targets with all my lenses, and comparing them to those shot on the 40d. With Canon equipment, sometimes it's not just one element that is mis calibrated- it could be both, thus accentuating your problem. I realize its really annoying to buy such expensive equipment to realize that it doesn't function properly, so I am more than willing to help contribute to solve your problem.


I usually have it in one shot mode (Actually, I don't think I've ever used AI Server or AI Focus).

The poor focus shows up in any light; The AF Assist light is/was set to emit, and I was shooting with the flash (It wasn't dark out, but it wasn't bright enough to shoot w/o flash, and the kitten was in the shadows of a brush pile). I would say the problem followed the 28-135 from my Rebel to the 30D, so I don't think it's the Body.

The thing that REALLY burns me is the 28-135 was sent to Canon twice while it was under waranty, and they claim that, with the exception of replacing the IS assembly, nothing is/was wrong with it. What do you think it would cost to send the body and lens back to Canon for focus calibration?

Here are two pics I just took of my "photogenic" cat Smokie... the first with the 28-135 @ 50mm f/4.5, and the second the 50mm @ f/4.5, both shot with flash and @ 1/250.

The one from the 28-135 doesn't look too bad (at "normal" viewing size, it's actually one of the better images I've captured with this lens), but when you view it native res, the 50mm completely destroys the 28-135.




Edited by Cougar281 on Sep 07, 2008 at 05:02 PM GMT

Edited by Cougar281 on Sep 07, 2008 at 05:03 PM GMT

This could very well be a body calibration problem, but to be sure it would make
sense to send the lens with both bodies and have Canon check them all out.

My 350D consistently front focuses a little bit, but it's a small amount
so I've never bothered to send it to Canon. In general, the 350D autofocus
is poorer than the 30D, but even the 30D can mis-focus. So far the 5D has
seemed a bit better in this regard, but I make sure I shoot multiple shots
for any that are super critical that I get keepers of.

I use center-point-only for most shooting, but will occasionally switch to
an off-center point when composition warrants it. I never use auto point selection.

Is your subject moving? If you are shooting with wide apertures in one shot, and your subject shifts slightly forwards or backwards, then you could experience some issues.

In the images I've posted, the subjects where stationary at the time; Here are two from my honeymoon (as another example), taken with the Rebel and the 28-135 @ 28mm, ISO100 1/200 f/9. One's not bad, although I wouldn't call it great, but the other is horrible.

AGeoJO wrote:
I would suggest for you to do a simple AF test. Use a straight ruler or tape measure and set it at 45 degree angle from your camera setup, which should be on a tripod, take a picture of the ruler at a point of your choice at the widest opening. You can download a focusing target from the internet if you will. Take a look at the resulting image and you can easily tell whether your lens backfocus or even front focus. Take a few shots but each time focus the lens away in between shots. You can do this with your lenses. If the results of your 28-135mm lens has pretty consistently disappointed you, I bet you almost anything, the lens will probably backfocus, judging by the kitty picture.

Hope this helps,
Joshua

I finally got around to doing this. I did the test with my 30D, on a tripod, using a remote release and MLU. I took quite a few more than I'm posting, at several stops including wide open, middle and smallest, and three of each. The center AF square is directly over the "25". At 28mm, the 28-135 appears to be less of a focus issue and more just plain horribly soft at the wide open end of the aperture. at f/22 it appears better, while not "tack sharp". It looks to me like all the lenses are back focusing to varying degrees. The 50mm, while "25" is in focus, it looks like the center of the DOF is around 25.5 or 26.

Here's the album I set up on Photobucket.

Thoughts? Is it possible I have more than one backfocusing lens? The image results of the 28-135 (and the 28-90, for that matter) on the 30D pretty much match those from the Rebel XT. I would have repeated the test with the rebel, except I don't have an adapter to be able to use the 30D release with the Rebel.

One thing I don't get; Why do the images vary so much in brightness? I was using shutter priority, and letting the camera determine the f-stop. Shouldn't they all be more or less uniform in brightness?

At f/22 you'll definitely be seeing the reduced resolution due to diffraction. f/11 would be your small-aperture limit for the 30D. Smaller apertures will look progressively worse over the entire image area. The effect of diffraction is greater on larger prints or enlargements than on smaller prints.

More than once I have had multiple, simultaneous gear problems so yes, you could be having more than one problem. The problem could be with the camera or lens or both.

Could IS be part of the problem ? Are you giving IS a full second to stabilise after activating it before taking a photo ? Some IS lenses only need half a second but the 28-135 is pretty old in IS terms.

The metering mode you use will determine what emphasis is given to the brightness of different parts of the image area. In some modes a slight change of composition can result in a significant metering change. In other modes (notably centre weighted average) there will be relatively little effect.

When doing that angled ruler test I suggest that you focus on a separate target that offers just a single edge that is parallel with the camera sensor, but also have the angled ruler somewhere in the image frame to let you judge focus offset. Camera AF doesn't perform as well when there are multiple edges or when the edge is at many different distances from the camera.

- Alan