EF 24-70 MkII: Erratic performance

The 24-70mki is legendary. The 24-70 Mkii is better. There should be no af issues. I had a few primes that exhibit erratic focus behaviors and I sent them in for adjustment. They came back as great copies. Sharper than most and spot on af. If I were you, I would send it in for service because of the lens gets fixed, adjusted and optimized. You don't get this custom optimization buying another copy off the shelf. If the lens come back from service and you still don't like it, then return back to store.

snake0ape wrote in post #15949294
The 24-70mki is legendary. The 24-70 Mkii is better. There should be no af issues. I had a few primes that exhibit erratic focus behaviors and I sent them in for adjustment. They came back as great copies. Sharper than most and spot on af. If I were you, I would send it in for service because of the lens gets fixed, adjusted and optimized. You don't get this custom optimization buying another copy off the shelf. If the lens come back from service and you still don't like it, then return back to store.

Yeah, when it works well it's legendary. I had some good photos with it.

Reading through the dedicated thread, I found out that (at least early on...) there was copy variation.

If I got a lens that was consistent (i.e. for a certain distance and FL have no variation where the DOF is centered as you change aperture, now it's all over the place... or consistently front focus at certain FL over a variation of distance and aperture...), then I would risk sending it to Canon. There is no guarantee that it will be returned within the grace period (30 days from the placement of the order...) so I could be stuck with a calibrated but still problematic lens in a non-MFA camera...

Heck... my 18-200 is more consistent than this...

OP - I would exchange it for another one since you still have time.

I shipped it back for an exchange... Fingers crossed!...

Lets see. 3 days of use and another fellow with extensive trade show testing. LMAO!

MakisM1 wrote in post #15949013
It's not the target or the method.

I would try with a target I consider better, but if yo uare sure, then of course replace the lens.

I don't think my own will get any better even if I request an exchange. Its just that it was not cheap which led me to expect better, more in line with the superb optics of the 70-200 f/2.8L which opened my eyes to what is possible.

Lowner wrote in post #15952251
I don't think my own will get any better even if I request an exchange. Its just that it was not cheap which led me to expect better, more in line with the superb optics of the 70-200 f/2.8L which opened my eyes to what is possible.

I feel the same about mine. Some days i think its fantastic, but other days i wonder where the money actually went.
I do like it, and prefer it over the Tamron version, but i really dont get the better than prime rep it gets.

That's why I started this thread. The behavior of the lens is erratic.

I started with this lens at IMG_7584 and finished with IMG_7886. 3 hundred photos, over two hundred of which under bench testing conditions (literally... my kitchen bench... :rolleyes. Hardly trade show experience...

I have a PhD in Engineering and just about 30 years of experience, most of it spent in research testing. Nothing to do with optics and cameras, although my PhD thesis was in non-linear wave diffraction (water waves, not light waves which are easier to deal with)... What I am trying to say is that I have enough testing experience to be able to judge the efficacy of the target/method and the variability of the results.

I have done 2 series of tests with the same test matrix I discussed in my original post. The first one was with Live View, the second one with viewfinder targeting/AF.

In my experience, when shooting handheld, I achieve better results with AI Servo, however, for tripod mounted camera shooting, I think I would increase the chances of inducing BBF camera vibrations than I would gain from AI Servo 'accuracy'. So One-Shot it was...

A priori, the Live View photos displayed as much or more variation. I attributed this to the fact that manually focusing until confirmation was not accurate enough (when you hear the beep you don't stop instantly). So midway I stopped MF and went to Live View AF... Just as bad... That's why I went to VF/AF.

It is difficult to describe the whole spreadsheet, but basically there was no monotonic increase (or decrease) with FL or distance. Since I used only three apertures, I just noticed the variability with no further comment.

My take is that this is a very complex lens (13 groups/18 elements), so if one or more elements are out of tolerance, the variations are pushed all over the place, as these elements interact with the others. Of course, my EF 70-200 MkII is even more complex (19 groups/23 elements) and it nails it every time...

The clincher for me was that at near infinity (20+ feet) it front focused, making infinity soft... I don't have MFA in my super Rebel and I did not want to get stuck with a lens which Canon might have problems adjusting.

Had the lens displayed a less complex behavior, say front focusing progressively as the distance decreased and no other variability, I would probably keep the lens and take my chances with Canon service.

Thank you all for your interest, I will keep you informed.

EDIT: One last thought... if the erratic lens behavior is due to 'delicate' (rather than 'robust') design, this may be the reason for not having IS. Adding another optical group might be too much to take...

You have answered your own question many times on this thread. I can add that the erratic af behavior is not typical of the lens. I did ( and others) suggest that you take a picture of a textured wall. The reason being is that we can see not just the center but also the edges of the picture. What if we saw that everything was in focus except the top edge? I wish you gave us an opportunity to see a picture rather than a front focus test that tells us nothing more.
The focusing issues you have described suggests that one of the lens is out of alignment. If you had sent in the lens to Canon, my bet is that you will get this response. " a lens element was misaligned preventing the lens from properly focusing". Canon would have fixed this , calibrated it , and tested it for the sharpest focus . For me, I would have sent it in. I like the fact that someone calibrated the lens. You won't get optimized callibration exchanging for another one at the store.
If you want to know more about misaligned lens, google decentered lens as a start. There are tests for that too. Try some of those kinds of test. They are made for problem lens. Your test is not made for problem lens.

MakisM1 wrote in post #15947369
Question is:

Is this common, or should I try the lottery again?

MakisM1 wrote in post #15947797
Q: Is it normally erratic, or I just got a quirky copy?

How can I answer my question and furthermore 'many times'?

I answered many times that my lens is erratic, but I can't answer the question' is this common'?

Keep in mind that I have over 50 real life shots (plus all the test shots where I can see the rest of the room), where I can judge whether the lens is sharper on one side or the other, de-centered etc. The answer is NO and further discussion on this becomes irrelevant.

In the distance, for most (but not all) focal lengths the lens was within tolerance. Good sharpness, excellent contrast and rendering, a close second to my EF 70-200 MkII. It just refused to focus (or even mis-focus) consistently...

MakisM1 wrote in post #15952899
It just refused to focus (or even mis-focus) consistently...

Which is why posting a shot of a wall is totally pointless.

Maybe posting 100 of them might convince people

dave_bass5 wrote in post #15952919
Which is why posting a shot of a wall is totally pointless.

Maybe posting 100 of them might convince people

At any specific FL, aperture, distance, the (mis)focus is more or less repeatable...

The problem is it can front focus at near MFD, backfocus at 3 ft, front focus at 5 ft and back focus again at 135 in... If you then look at the same FL but varying the aperture as well, you get more variations that don't seem to follow some pattern.

Anyway, I really want this lens to work, so I hope for better results with the next iteration. Then, I might send it to Canon...

Well good luck. For what its worth i find mine works better on my 5DMKIII than on my 60D, but both seem to produce good images most of the time, just not always outstanding.

MakisM1 wrote in post #15952742
That's why I started this thread. The behavior of the lens is erratic.

I started with this lens at IMG_7584 and finished with IMG_7886. 3 hundred photos, over two hundred of which under bench testing conditions (literally... my kitchen bench... :rolleyes. Hardly trade show experience...

I have a PhD in Engineering and just about 30 years of experience, most of it spent in research testing. Nothing to do with optics and cameras, although my PhD thesis was in non-linear wave diffraction (water waves, not light waves which are easier to deal with)... What I am trying to say is that I have enough testing experience to be able to judge the efficacy of the target/method and the variability of the results.

I have done 2 series of tests with the same test matrix I discussed in my original post. The first one was with Live View, the second one with viewfinder targeting/AF.

In my experience, when shooting handheld, I achieve better results with AI Servo, however, for tripod mounted camera shooting, I think I would increase the chances of inducing BBF camera vibrations than I would gain from AI Servo 'accuracy'. So One-Shot it was...

A priori, the Live View photos displayed as much or more variation. I attributed this to the fact that manually focusing until confirmation was not accurate enough (when you hear the beep you don't stop instantly). So midway I stopped MF and went to Live View AF... Just as bad... That's why I went to VF/AF.

It is difficult to describe the whole spreadsheet, but basically there was no monotonic increase (or decrease) with FL or distance. Since I used only three apertures, I just noticed the variability with no further comment.

My take is that this is a very complex lens (13 groups/18 elements), so if one or more elements are out of tolerance, the variations are pushed all over the place, as these elements interact with the others. Of course, my EF 70-200 MkII is even more complex (19 groups/23 elements) and it nails it every time...

The clincher for me was that at near infinity (20+ feet) it front focused, making infinity soft... I don't have MFA in my super Rebel and I did not want to get stuck with a lens which Canon might have problems adjusting.

Had the lens displayed a less complex behavior, say front focusing progressively as the distanced decreased and no other variability, I would probably keep the lens and take my chances with Canon service.

Thank you all for your interest, I will keep you informed.

EDIT: One last thought... if the erratic lens behavior is due to 'delicate' (rather than 'robust') design, this may be the reason for not having IS. Adding another optical group might be too much to take...

I do think you're putting the chicken before the egg. 100 real world shots and 200 test shots seems backwards to me.

before I do test my cameras, I put them through the ringer with a lot of shots over different days. If you shoot as you normally do and have little issues, then no need to run test shots. Also, it might be a good idea to invest in Focal or something, to automate the process and take out human errors in testing procedures.

good luck with your findings.