My 24-70 2.8 L has been to Canon service three times for the same problem: poor focusing/muddied images at the wider angles. I don't think its an out of focus issue since none of my tests show front or back focusing. It looks more like low resolution or a dirty lens. I get the same results on my 5D and 1D IIn bodies. On my last repair I sent Canon sample files showing the issue and the repair ticket says they replaced the collar and adjusted F/B focus.
** UPDATED EXAMPLES**
I'd like to get a second opinion on what I am seeing before I send it back to Canon again. Here are three shots at 100% mag at f/4 on a tripod with the zoom settings of 24mm and 35mm compared between a 17-40 and 24-70.
17-40 @ 24mm f/4 100% crop:
24-70 @ 24mm f/4 100% crop:
17-40 @ 35mm f/4 100% crop:
24-70 @ 35mm f/4 100% crop:
Edited by cheapscot on Oct 28, 2007 at 04:17 PM GMT (Reason: updated examples)
I'm just curious here, but what metering mode are you using? The presence of leaves in the background my be fooling your meter into believing the scene is lighter than it really is, so the face turns out darker. Because at 50mm there is more of the face in the photo perhaps the meter is reading more accurately. Jsut a thought.
Yes - they are unprocessed 100% crops. Nope, not really horrible (I'll let my daughter know! :-) as sharpening makes them usable. This is probably not the best example. What I have found is that group shots at the wider angles are unclear, especially when compared to my 17-40 L f/4 at the same focal lengths.
As for the metering, yes it could be better. I have other examples that are metered identically that seem to show a problem to me.
Then again I may just be seeing things that aren't there. All I know is I get better results with my 17-40 and have not figured out how to tell the Canon repair folks what to fix, since I don't think its a focus (or exposure) issue.
This strikes me a bit as inconclusive.
(1) Try to get closer to your subject, at typical head/shoulder portrait distances (don't worry about spatial distortions...it's just testing).
(2) Have better light on your subject, and do not shoot your subject against much brighter background.
A sharp image properly exposed should show skin pores and fine facial hair on 100% crops, straight out of camera.
Thanks everyone for your comments. I realized that the previous examples I provided did not do the problem justice, so I took more shots showing what my 17-40 looks like compared to my 24-70 using the same focal length and exposure.
Even with the 24-70 stopped done 1 f-stop and the 17-40 wide open, the 24-70 images are really bad in comparison.
The 24-70 is perfect for wedding work so I prefer to get it fixed.
This is an issue to all 24-70s I tested, save one. 24-35 (appx half the range) has issues achieving full focus from distances of 10 feet and beyond. Yep, its acceptably sharp, and with some USM, can be useable. I ended up accepting the issue and using the parfocal method to get critically sharp images from that range when shooting from a distance.
Until I tested a friend's 24-70 and finally found one out of 15+ that I have tested in the years. My buddy from VA agreed to lend me his lens as I drove mine to Canon NJ and told them I wasnt crazy. I have had the lens calibrated 3x beforehand (plus the back and forth with my testing after they initially sent it back to me) and all the times, they said same issue, in tolerance. I said get my wide range the same as this one.
They did.
I test mine every now and then, and I get sharp images from the wide range everytime I get it right. Funny thing, though, when shooting formals, it took a while for me to not shoot parfocal and trust that the lens will get it right.
Thanks for the story. Now I just need to borrow yours for calibration! :-)
Maybe Canon will take my 17-40 to use for the calibration - but then it might not come back the same.
For what it's worth to you, I agree that there's an issue.
In an environment where no one, least of all the repair tech, seem to take the time to actually read and understand painstakingly prepared problem documentation you might try including something else...
I would suggest including a focus accuracy test chart taken in proper conditions with the two lenses. That way your sample will show that the subject is "clearest" at the target focus distance but remains unacceptably sharp compared to other lenses.
Perhaps then they will stop adjusting your front/back focus and get to the issue of sharpness at the wide end of the zoom.
I wouldn't expect my wide open 17-40 to beat my 28-70 stopped down to f/4 like that, let alone the supposedly superior 24-70.
I brought pics of charts, soup cans, trees, both with pics shot at 24, then same shot with focus acquired using 70, then back to 24, the ones shot at 24 were still marginal at best. It wasnt till I had another lens that it was done right. It didnt matter what body I used, 1.6/1.3/FF (both 5d and 1ds), focus was not sharpest. I knew how much manual adjust needed to get it there, and it will if I did. The AF on the wide side with distance to subject, just couldnt get it. Didnt matter the aperture either, the level of focus was just not there.
I was there so many times, they had the lead tech come out and talk to me and ask me specifically what was happening. After 3 tries going outside, checking, giving it back at the center, it finally caught sharp focus. There's apparently a range of 7 steps per focal length that can be adjusted either way, and we got specific on how many, front of back, was necessary. Truth or not, it worked for my lens.
I played with another 2 month old 24-70 (build date 2007) on a 5d this past weekend and it had the same issue. Far focus, acceptably sharp. Get to 70, refocus, gets sharper, back up and take shot, critically sharp. Even the one at the booth at the expo, same thing.
calemon: Thanks for the tip about using a focus test chart!
njjuliano: Wow that's really sad - especially for a $1k lens. My 17-40 had back focus issues that took 3 trips to fix. On the final repair I got the name and number of a guy who promised to personally see to it. I think I'll try getting him on this one too.
I had not tried the parfocal trick before - it does make a difference, but still worse than the 17-40.
I wouldn't necessarily rely on your 24-70L being parfocal.
Mine certainly isn't. I definitely have to refocus as I zoom.
Now, perhaps mine should be but has drifted out of calibration.
At any rate, do a few tests to see if yours is parfocal. It may not be.
sorry this thread perked my ears, but i'm not really following the tests to reproduce it, or test my lens for the same thing. could someone explain again how to recreate the test?
I know my 24-70 shot sharp either 24 or 70. With my Mark III I ran the micro adjustment on the 24-70. I set up a test chart at 45 degree angle and shot numerous shots. If I adjusted at 24 it was terribly out at 70 and vice versa. From the 24 to the 70 it is a total of 15 steps...+3 to -12....so I set it at -6....
I thought I had read that zooms are like that....but it sounds like they shouldn't be that far out. My 70-200 2.8L IS is similar but not as bad.
I was going to send them in with my Mark III when it goes in for repair. Should I expect it to be good all the way through? I do expect it, but not sure it is realistic though.
Is there any chance of Canon wrecking your lens when you send it back? (e.g. scratching it, getting dust into it, dropping it etc.) I have heard of some horror stories where people pay to get their sensor cleaned by Canon and it comes back with even more crap on it
trumpet_guy wrote:
I wouldn't necessarily rely on your 24-70L being parfocal.
Mine certainly isn't. I definitely have to refocus as I zoom.
Now, perhaps mine should be but has drifted out of calibration.
At any rate, do a few tests to see if yours is parfocal. It may not be.
The 24-70 is designed to be a parfocal lens. From quick google:
).