Very inconsistent focus, AI servo, 35mm 1.4L on 5dII

[nick_rigopoulos1]

<p>Well this is a new one. I bought a 35mm 1.4L a couple of days ago, and have been shooting anything and everything with it to get a feel for it. After getting a sneaking suspicion something wasn't quite 'right' with it, I set it up on a tripod and did an informal focus test. Focusing on the small text on a pill bottle from various distances, I've determined the autofocus is wildly inconsistent from wide open to at least f2.8. Any smaller and it seems like DoF makes up for it. <br>

In Single Shot focus mode, focus is 9/10 consistent, but still not quite as good as I can get manually focusing via live view zoom. (is this normal, btw?) In AI Servo mode, focus is off 7/10 shots wide open. In between shots I would focus away or on my hand to 'reset' the test. This was using no AF area point expansion, no micro adjustment, and just the center AF point.<br>

I did similar tests with my 135mm f/2L and 50mm 1.4 and found no problems. Unfortunately I do not have another Canon body to test with to see if it's actually a problem with the lens or the body/lens combo.<br>

Now, I've read many reports of examples of the 35mm 1.4L backfocusing or frontfocusing, but I've never heard of anything quite like this.</p>

<p>If anyone has any advice, it would be greatly appreciated.</p>

<p>-Nick</p>

[Brad_]

Do you really need AI servo mode?

 

My 35 f/1.4 nails focus every time in normal AF mode.

[bobatkins]

<p>I'm not sure that I'd expect AI servo to be as good as one shot AF on a stationary target. AI servo has an algorithm designed and optimized for moving subject tracking.</p>

<p>It's also posisble that AI servo has a harder time with shorter focal length lenses. Not sure about that one.</p>

[tom_harvey3]

<p>As Bob said, AI servo is probably NOT the mode to be using...use one-shot.</p>

<p>If focus is still off, then try doing a microfocus adjustment using LiveView. Remember that the sensor-to-subject distance should be at least 40 X the lens focal length.</p>

[igord]

<p>I had problems with regular AF while using 5mk2 with 35/1.4 even @f5.6<br>

Lot of out of focus photographs. Eventually started to focus manually as it was commercial<br>

shoot.</p>

 

[zml]

<p>FWIW: my 35/1.4 was very inconsistent (Servo or not) on several cameras esp. when using off-center AF points but was adjusted by Canon and the problem seem to have gone away. Should you decide to send it to Canon for adjustments, they might ask you to send the camera and the lens: make sure (as in attaching a big sticker to the camera...) that you want the lens adjusted to focus correctly with that camera body, not the camera (unless you feel that the camera needs adjustments as well...)</p>

[steve_dunn2]

<p>In addition to what others have suggested, keep in mind that the 35/1.4 offers greater DOF than either of the other lenses you tested. Remember that AF is done with the lens wide open, and is supposed to acquire the subject to within ±DOF (or 1/3 of DOF in the case of a high-precision AF point, if your unnamed camera has one), and since the 35 has greater DOF wide open, you should expect to see more variation in AF results.</p>

 

<p>That is probably not the entire explanation, since you imply that sometimes AF is so far off that you have to stop down to f/2.8 for the subject to be sharp. But it may be a contributing factor.</p>

[zml]

<p><em>> and since the 35 has greater DOF wide open</em></p>

<p>Could you elaborate..?</p>

[daniel flather]

<p>Did you do this test before or after you took the pills?</p>

[nick_rigopoulos1]

<p>Michael,</p>

<p>I think he means the 35 has greater DOF wide open than say the 135 has wide open, simply because it is a wider focal length.</p>

<p>Steve,</p>

<p>I mentioned the camera in the title, it's a 5D mk II.</p>

[zml]

<p><em>> I think he means the 35 has greater DOF wide open than say the 135 has wide open, simply because it is a wider focal length. </em></p>

<p>All things (such as f/stop) being equal, where it comes to DOF it is (nearly) all in the subject magnification and near-far object relationships, not in focal length. If you preserve the subject magnification ratio, you'll get nearly the same DOF regardless of the focal length at the same f/stop (if fractions of a mm don't matter to you...) Of course you'll need to decrease the distance to the subject to preserve the magnification ratio with a wide(er) angle lens and (even at the same magnification rato) you'll run into near-far object size relationships issues ("telephoto compression", "wide angle effect") but the DOF per se will be nearly identical be it a 300 or a 35 mm lens. And of course if you use both, 300 and 35 mm, lenses from the same distance to the subject, the latter will have a much greater DOF because of much smaller magnification ratio.</p>

[igord]

<p>Yes, that's always the first basic lesson of photography ;) but people seem not to remember it ;)</p>

[nick_rigopoulos1]

<p>That's a very long, detailed, well-thought out version of...pretty much what I said in a few words.</p>

[steve_dunn2]

<p>Nick, my apologies; I re-read the body of your message a few times to see if I'd overlooked what body you were using, but didn't re-read the subject. Your body's centre AF point is indeed capable of high-precision operations when used with a lens that's f/2.8 or faster, so with all three of those lenses, it should achieve focus to within 1/3 of DOF.</p>

 

<p>Yes, excluding macro, if you move in closer with a wide-angle lens to achieve the same magnification, you get pretty much the same DOF. If that's the procedure you followed when comparing AF with the three lenses, then DOF shouldn't figure into it. If you swapped lenses but didn't adjust the camera-to-subject distance (or didn't adjust it enough to cancel out the change in focal length), then DOF might come into play - but with the high-precision AF point, even if the camera focuses at the outer limits of its specs, that should still produce a pretty sharp result, so it sounds like this is not what's happening.</p>

[nick_rigopoulos1]

<p>Ok, now I'm really confused.</p>

<p>I set up a much more thorough testing environment on a variety of subjects, all in good lighting. All of my AF lenses, the 35 1.4, the 135 2.0, and the 50 1.4, each of them showed a certain degree of AF inconsistency. None matched the precision of live view zoom manual focus. <br>

Am I overreacting? I was under the impression, perhaps mistakenly, that using the center AF point, AF performance should be consistent, repeatable, and damn near perfect in optimal conditions...am I wrong? Is this normal to have some degree of variability?<br>

If not, the only explanation (unless someone reeeaaally hates me and cursed all my lenses) would be a problem with the body. I unfortunately do not possess another Canon body.</p>

[zml]

<p>Nick, a few points:<br>

- If there is a consistent bias in AF errors with a particular lens/body, try MFA.<br>

- If you shoot at f/4 or smaller, miniscule random AF errors are not that important (covered by DOF in some/most cases.)<br>

- Remember that "AF points" are not literal points but rather - quite large in relation to the VF size - areas so if you try to focus on a single eyelash (even with spot AF should your camera had such feature) the selectd AF point covers a lot of ground especially if the subject is 3D and/or turned at an angle to the camera.<br>

-If you ever wondered why event shooters let it rip at 3, 5 or even 10 fps most of the time, you have your answer: in addition to increase the odds of capturing peak action and a "pleasing" scene, chances are that at least 1 frame from the series will be in perfect focus (regardeless of the equipment and focus mode used...everything and everybody misfocuses at times...) </p>

<p>And as for tests, after making sure that your equipment is not defective or in need of gross adjustments, in which case cart it off to Canon, run your AF tests in real life shooting conditions and use MFA and changes in your shooting technique to get better focus; there is more to AF, esp. with wide open fast optics, than just pointing at the desired object and pressing the shutter release.</p>