5D mk.1 24-70 f2.8L focusing problem

<p> </p>

 

<p>I know this topic has been flogged to death but I am yet to remedy my situation so I thought some one could throw me some good advice?</p>

<p>I am shooting with a 5D mk.1 and have recently sold my 24-105 f4L because I felt it never lived up to it's sharp reputation. I acquired a 24-70 f2.8L to replace it. After a holiday in Australia I looked at my shots with my new lens and found out that things were even softer. Upon reading up on this I found out that many people have had the same issue with a back focusing problem particularly at the wider focal lengths. Remedy = return to Canon for recalibration?</p>

<p>In I went to the Canon service agent here in Copenhagen with both my 5D and the 24-70. After checking both they tell me that the lens is clearly in need of a recalibration and so infact is the body. The lens was of course covered by warranty but the body is sadly 25 months old. They adjusted both and stung me big time for the body adjustment.</p>

<p>Still a very low number of "keepers" in my batches of shots. Sharp focus seems to be hitting just near the ears. To satisfy my curiosity I did a focal test via a test chart, tripod, cable release and mirror lock engaged. If anything it seems that the lens seems to be front focusing slightly at some focal length and aperture combinations but nothing too dramatic. Strange?</p>

<p>Are my expectations simply too high of this lens? I believe true focus should not be too much to ask of a lens of this price?</p>

<p>My technique is as follows:</p>

<p>1) Handheld shooting in AV mode and always ensuring minimum shutter speed at focal length x 2. Will push ISO if necessary to ensure this.</p>

<p>2) Rarely shoot a portrait as open as f2.8 (with this lens) but often at f3.2 - 4.5. Should the focal plane be paper thin at such apertures (like 2 cm)?</p>

<p>3) Focus routine is NOT focus lock, recompose and shoot but rather assign custom function for focus lock to (exposure lock button) and assign focal point adjust to big wheel so I therefore select focal point, lock focus (closest eye) and shoot. Focus mode is "AI" or "one shot". I do not trust "servo" on relatively motionless subjects.</p>

<p>Anyway, the bottom line for me is that how can Canon be producing such expensive equipment (particularly L series lenses) that out of the box are simply sub standard? I sold a perfectly good 24-105 thinking it was a "bad" copy when infact it seems it was my body that needed calibration. I know that user error makes up a large proportion of "bad copy" claims and NO one is willing to accept that their tools are NOT to blame but even if it is 5% of "bad copies" leaving the factory it is simply not good enough for me. Better quality control please!</p>

<p>I started photography using a Nikon F4S and AIS lenses in the late 80's. I switched to Canon during the digital revolution due to the quatum leaps Canon were making with AF in the late 90's. Oh how I miss Nikon sometimes!</p>

<p>Btw. I do not want to here anything about fixed focal lenses v zooms. I have a few fixed lenses and they are nice BUT this brick of glass should perform better.</p>

<p>please prove me wrong?</p>

<p>thanks in advance</p>

<p>George</p>

 

 

<p>Here is a 100% crop of a sample image. Notice how sharp focus seems to be hitting my daughters hair near her ear. I have 100's of these "close enough is not good enough" shots!</p>

<p>30mm focal length shot handheld f3.2@ 1/3200</p>

<p> </p><div></div>

<p>Nothing to offer except:<br>

[[should the focal plane be paper thin at such apertures (like 2 cm)?]]</p>

<p>You can run DoF calculations here if you're curious:</p>

<p>http://www.dofmaster.com/dofjs.html</p>

 

<p>thanks Rob</p>

<p>I just ran the numbers and it seems that I have 0.57ft depth of field at this distance and f stop. That should more than enough room for error?!</p>

<p>At 30mm the child can't be more than 2ft away from the lens. So assuming that distance you will have about 0.25ft/7.5cm DOF. It would be very easy to miss the shot, especially with AF</p>

<p>Bear in mind that website is using a Circle of Confusion of 0.03mm, which is 30 micrometers. The 5D mk 1 has a pixel pitch of 8.2 micrometers, which means that Circle of Confusion will consider something "in focus" as long as it isn't smeared into a circle wider than about 4 pixels. If that's a 100% crop above, I'm inclined to think the eyes are "in focus" by that criteria. If you use a Circle of Confusion of 0.0082mm, which is one pixel, the depth of field would be considerably narrower.</p>

<p>Have you taped some newspaper to a wall and shot it with the camera on a tripod? You might have a focus problem with the camera, or it could be technique, or some of each.</p>

<p>George,</p>

<p>You have a problem, at f2.8 there is a little softness but not focus errors with mine.</p><div></div>

<p>@Ilya - I'm 100% certain she was stationery and the focal point confirmed smack bang on the left eye (as we look at her). Yes the DOF may be thin but by no means a difficult shot under those conditions.</p>

<p>@Alan - well I'll be honest with you Allan the only thing I know about the "circle of confusion" u mention is what I just read above. I can see that u r saying that my "real" DOF may be considerably less. Yes, but again, if I had a confirm on her eye and she or i haven't moved than how can it be grabbing focus on a point around 4 cm past that?</p>

<p>I will try the newspaper test tomorrow evening and let u know how it turns out.</p>

<p>But that had bad back light that reduced contrast, here is one with natural front light, so much better contrast, at f3.2.</p><div></div>

<p>thanks Scott</p>

<p>nice shot btw! have u done any post sharpening on that or has his face come up that sharp just through the camera via JPG?</p>

<p>George,</p>

<p>Absolutely no sharpening at all. I only shoot RAW, store in LR (so ACR rendering engine but all sliders unused for these examples) opened in PS to resize and get 100% crops, then just save for web at 70% or so. They sharpen up much much clearer.</p>

<p> </p>

<p>If you continue to have an issue, they cover the past repair for up to 6 months.</p>

<p>If you mean AI Focus by the term, "AI", that is not a good mode to use at all. It will interpret any camera motion as subject moving and jump into AI Servo, and will then focus on whatever moves or is contrasty behind the subject. If you are using AI Focus, weed out all bad examples using that mode, and don't use that mode in your future tests.</p>

<p>I know that the 24-70mm has a reputation for funky focusing, and I don't own one--never have, so I can't say anything about it. I use a 5D original and Tamron 28-75mm f2.8, plus other Canon lenses, including the 16-35mm f2.8L. I don't have much trouble with focus at all, and my camera has never been recalibrated since I got it. My 16-35mm has, several times, because it was bumped pretty hard and something inside it got bent. It is OK now.</p>

<p>I would not think the camera and lens would need to be calibrated together, because that would mean that other lenses used on the camera may then be 'out'. Each is supposed to be calibrated to a known good standard.</p>

<p>I would review good autofocusing technique, to rule out user error, or at least, misinformation, since your focus tests on the lens seems to indicate it isn't all that 'out'. When I used a 20D with the 16-35mm, I'd get numerous examples of back focusing, particularly on the wider focal lengths. After I figured out why, I could improve my hits.</p>

<p>A 5D is much better than a 20D when it comes to focusing, but some of the following still applies.</p>

<p>1. Don't expect miracles in low light, particularly from outer focus points.</p>

<p>2. Don't put the focus point over a subject's 'edge'. The focus point can become confused and backfocus on something contrasty.</p>

<p>3. Don't put the focus point on a smooth, non contrasty area. It will definitely cause confusion and make it backfocus.</p>

<p>4. The red squares in the viewfinder may not show exactly where the focus point is. In the 20D, and probably in cropped sensor cameras, the focus point is actually bigger than the red square, and sometimes, not centered over the red square.</p>

<p>5. With wider focal lengths, it has always been harder to accurately focus, because subjects are usually smaller in the frame, and it is both harder to see and harder for the camera to differentiate subject matter thrown at it, particularly if there is depth to the scene. I sometimes autofocus twice on a 'good' focus area (no edges, nothing smooth, etc.) and double check by focus scale on the lens. I do this less with my 5D.</p>

<p>Now, I have no idea if the following has any truth to it at all--it is just my guessing--but I think the clunkier nature of the Tamron's non Canon focusing actually helps in getting consistent focus accuracy. It isn't as smooth and quick as L lenses, but it sure is pretty consistent. As my friend and colleague, William W. says, "Canon lenses try too hard."</p>

<p>thanks Nadine for your great reply. The sample shot was actually shot via One Shot mode and I have heeded the advice of shying away from AI mode due to reading posts such as yours recently. Doing that hasn't really improved my "hit" rate.</p>

<p>The other factors you mention are indeed relevant and I have discovered them myself, particularly the one about contrast. I do rely heavily on the selection of the single focal sensor point and crop post to avoid the focus, lock, recompose dance. By saying this I have introduced point 5 of yours into play and maybe the size of the subject in the frame can / could be a factor? I will try and shoot at longer focal lengths in the future to eliminate this possibility.</p>

<p>I am also impressed that you picked up on the fact that by recalibrating the 5D body it may introduce problems with focusing with my other lenses. That is exactly what the Canon technician told me at that time. I now swear that my 100mm USM f2.8 macro is not as accurate as what it was pre the adjustment but I am hoping it is just paranoia? Unlike the 5d mk.II the mk.I does not allow for individual lens focal calibration.</p>

<p>George, stick with center point and recompose -- even with the Mk II the outer focus points have less than stellar reliability. I understand this defeats the purpose of them, but them's the breaks.</p>

<p>attached is a focus test chart shot.</p>

<p>35mm focal length f2.8 @ 1/80th, tripod mounted, cable release + mirror lock up active</p>

<p>I can see the focus is hitting between 2mm and 6mm on the closest side of the chart. Is this enough to worry about?</p><div></div>

<p>George,</p>

<p>Interesting in that your test shot shows front focus, the image of your daughter is back focused. I wouldn't be concerned from seeing the test chart results, if that result is consistent. When micro AF adjust thing my lenses I normally do it 20-30 times (I know you don't have the adjustment but you should still average out your results), also make sure that your chart is an appropriate distance away from the lens, Canon recommend at least 50 times focal length.</p>

<p>Can you show an uncropped version of your daughters image? It would be interesting to see the full framing, but with these mixed results (both front and back focusing) the truth is, it is more likely to be technique.</p>

<p>George, that shot of your daughter is from really close. Do you think you might have been inside the minimum focusing distance for your lens?</p>

<p>I agree with a previous post. I think your daughter is too close to the lens. </p>

<p>My experience with that lens is 2.8 is too soft for my taste. To me it's a F4 lens and that's why I sold it and went with the 24-105. But you have been given some great advice here and I think you will soon figure it out. Keep it in one shot focus and use the center marker. Also your sample picture could have been shot with a F4 or a 5.6 to open your DOF since you had such a high shutter speed. Since you have other lens' you can simply test them to see if they give you the same out of focus count. Good luck v/r Buffdr</p>

<p>Buffdr,</p>

<p>What do you find unacceptably soft in my second example shot at f3.2?</p>

<p>Two things. First, I want to make it clear about what you call AI. AI Focus is another choice for autofocus mode. This is different from AI Servo. I don't use AI Servo unless I have good light (not dim) and I am using it as intended, for tracking a moving subject. Otherwise, I use One Shot, and never AI Focus, for the reasons above. One day, when I may have a subject exactly like the one that is described in the AI Focus section of the camera manual (moving toddler), I might consider it. Otherwise, for most of what I do, One Shot it is. Even for a brdie walking toward me in the processional.</p>

<p>Second, when I quoted William W., I meant it half jokingly, but actually, the statement is quite true. I have found that the autofocus system 'rushes' to find it's mark, and maybe it hits, but that hit might be wrong (inaccurate). Or it may encounter an obstacle in the form of a difficult (non contrasty) surface or something and "throw it's hands in the air and give up", meaning it will find the contrastiest thing behind the subject and focus on that. Much like your example shot. You focus on the left eye. For some reason it encountered some difficulty so it lands on the contrasty hair tendrils behind. With a Tamron lens, it will front focus--land on the contrastiest thing in front of the subject.</p>

<p>And, just as another example, for group, wide angle, full length shots of people, I sometimes put the focus point on their feet (focus/recompose because I usually have good DOF). This is because there is usually nothing for it to backfocus on--no edges or depth close by. For a shot like this, if you focus on someone's face (the recommended thing), the faces are usually so small that there are plenty of edges.</p>

<p>Your focus test shot looks fine to me, and actually, as you say, seems to indicate an extremely slight front focus. And don't feel bad--I've heard of more than one professional photographer who was totally frustrated with their 24-70mm re focus and a few of them actually went Nikon when the D3/D700 came out. Or at least, sold their 24-70mm and used something else. Just to preserve your sanity, you might rent another 24-70mm and see if you get the same results, or another appropriate lens.</p>

<p>it's always fun waiting for the globe to rotate and the Americans to awake! Good to have you people back.</p>

<p>@ Dave and Jerry - yes it looks like she is close because it is a 100% crop of the full shot. Even so the 24-70mm is a "macro" rated lens and focuses to 37cm (don't know what that is in imperial measurement) but it is quite close and she was atleast 70cm away when I shot. Besides that I got a focus confirm on her left eye via one shot and focus confirm will not light if one is out of tolerance.</p>

 

<p>I know this stinks of technique (and I haven't 100% discounted it as a factor) but why can I handhold my 50mm 1.4 and 100mm 2.8 USM macro with much better results at similar focal lengths and the same technique?</p>

<p>The Canon technician who informed me about my problem when I went it for that initial adjustment told me that he was a little baffled at how the lens and body were working together. I don't know if his disply of bewilderness was simply an excuse to get me to fork out for getting my body adjusted but it struck me that he had rarely seen such a phenomenon. I am simply sick of throwing money at this thing in the hope of getting "lucky".</p>

<p>I am unable to help any more with the diagnosis but my 24-70 is a very sharp lens. It is slightly soft at the edges but i find it noticibly harper than the 24-105 it replaced. Here is a crop taken with the 24-70 F2.8 at F2.8, 1/60 ISO100 in fairly poor light. The shot is unsharpened and the process of putting it into a small JPEG has softened the image. I actually find the 24-70 to be a sharp lens - obviously not as sharp as the 105 f2.8 L IS macro but very usable at F2.8. Indeed I find that it is very close to the 50 f1.4 in sharpness at F2.8.<br>

I assume that when the lens was calibrated there were no issues with lens element alignment in the lens (e.g as a result of dropping it). You may want to compare MF and AF shots - try and use a live view body if you can borrow one. Doing this will allow to to determine if it is the AF system or the lens that is at fault. My guess is the AF system - obviously use a tripod.</p><div></div>