RobGalbraith.com
Go to advertiser website.
     Home
     RSS
     CF/SD/XQD
     About
     Contact
Go
Go to advertiser website.
 
An analysis of EOS-1D Mark III autofocus performance - Continued
July 2, 2007: Even more on EOS-1D Mark III autofocus
 
It has been almost two weeks since this article was first published, and the feedback continues to flow in. There are three questions that show up in the latest wave of messages: are we still experiencing problems with EOS-1D Mark III autofocus, have we done any additional testing or learned of any workarounds and has Canon given a statement about what they intend to do. This update tackles all this, plus you'll also find links to several full resolution sequences shot with the camera.
 
Q. Are you still having problems with EOS-1D Mark III autofocus?
 
Yes. In the last couple of weeks, we've been shooting with a third full production body, and its autofocus is acting up in the same way as the first and second full production bodies we used in the weeks prior to this article's original publication.
 
Q. Have you done any additional testing of the camera?
 
Yes. We've received quite a few suggestions of things that might help; of those suggestions, one that seemed worth trying was the placing of UV or IR filters on the lens, to see if somehow the collapse of long lens autofocus performance in bright light and warm temperatures might be related to a UV or IR oversensitivity on the part of the EOS-1D Mark III's CMOS autofocus sensor.
 
Unfortunately, the runner photos we shot with UV and Tiffen Hot Mirror filters on three different lenses - an EF 70-200mm f/2.8L IS, EF 300mm f/2.8L IS and EF 500mm f/4L IS - were effectively the same as those shot without either filter in place. Which is to say there were about the same number of out of focus frames in both continuous movement and erratic movement sequences. Whatever's wrong in the EOS-1D Mark III's autofocus, it doesn't seem to be caused by UV or IR light.
 
Q. I'm hearing that the camera's autofocus can be made to work much better by tweaking some of the EOS-1D Mark III's Custom Functions. Has that been your experience?
 
No.
 
But it's important to distinguish between the camera's autofocus problems that are likely the result of design flaws, and autofocus problems that are likely the result of intentional changes in how the autofocus behaves (compared to the EOS-1D Mark II N). Near the beginning of this article we describe three problems - poor initial focus, poor tracking and fidgety focus on static objects - that appear or are made much worse by the presence of full sunlight/warmer temperatures. These are all caused by what must be errors in the design of the autofocus system, and based on our testing of various Custom Function combinations, we're confident that these problems aren't helped or eliminated by Custom Function adjustments.
 
Even when shooting in conditions that don't bring about these particular problems, the EOS-1D Mark III on its default autofocus settings is still overeager to jump to a new point of focus. It's possible to calm the camera's autofocus down with Custom Function tweaks, which in turn makes it somewhat easier to track subjects moving erratically. So far, these changes seem to help:
  • Set C. FnIII-2 (AI Servo Tracking Sensitivity) to the notch in-between normal and Slow, or even Slow if you're not bothered by the sluggish feeling that the autofocus takes on when intentionally switching from a near subject to a far subject with the autofocus engaged. This, more than any other autofocus Custom Function, changes both how the autofocus feels and how it actually behaves when following a player dipsy-doodling their way down the field.

  • Set C. FnIII-8 (AF Expansion with Selected Point) to III-8-1. This will switch on the assist AF points on the left and right of the active AF point, and in conjunction with a slowed C. FnIII-2, it does make the autofocus system more forgiving when the active AF point drifts off its intended target briefly.
Changing various other settings doesn't improve the autofocus in any way we can feel or see, including these two:
  • C. FnIII-4, AI Servo Tracking Method. This setting kicks in only when the camera is configured to automatically select from multiple AF points, including when C. FnIII-8 is set to III-8-1 or III-8-2.

    On the default of C. FnIII-4-0, the camera will give priority to whatever subject is closest. For example, if III-8-1 is set, the camera will switch away from the active AF point to one of the other two AF points if it detects there is a closer subject at one of those points. Setting C. FnIII-4-1 turns off closest subject priority. Using III-8-1 again as an example, the camera will keep on tracking at the active AF point even if a closer subject pops up at one of the other two AF points.

    This is how C. FnIII-4 is supposed to work, and it is in fact how it does work, sort of, when we test it in the office, but out in the real world of tracking relay runners or a scrambling quarterback, III-4-0 does not consistently give priority to the closest subject at the available trio of autofocus points with III-8-1 dialed in. In fact, regardless of whether III-4-0 or III-4-1 are set, when shooting under non-bright/non-warm conditions and the autofocus misfocuses slightly, it's almost always a backfocus error (under bright/warm conditions long lens autofocus isn't consistent enough to fairly judge what impact this Custom Function is having).

    The only obvious difference we can see between III-4-0 and III-4-1 is when the active point drifts slightly off the subject to a distant background briefly, the camera is more likely to shift to the background with III-4-1 set. And since this is almost never a good thing, the camera's default of III-4-0 looks like the safer bet. Despite the fact that the description of what this setting is for would suggest that III-4-1 is the one to choose.

  • C. FnIII-5, Lens Drive When AF Impossible. If III-5-0 is set and the EOS-1D Mark III is unable to detect focus, the camera will force the lens to rake focus back and forth in a bid to detect subject distance. If III-5-1 is set and the camera is unable to detect focus, the autofocus system will instead stop cold until the autofocus is disengaged, then reengaged (by releasing then pressing the rear AF-ON button, for example).

    This is what this setting does, but an entry by bird photographer Art Morris in his April 1, 2007 Birds As Art newsletter suggests that might not be all that it does. In his newsletter, Morris says that with earlier 1-series Canons anyway, disabling focus search improved long lens focus stability in some situations for him, including when photographing relatively static subjects.

    His enthusiasm for this setting, albeit with different cameras and different subject matter, was enough to prompt us to give III-5-1 a whirl, in the hopes it would help settle the EOS-1D Mark III's tendency to shift focus slightly but constantly when AI Servo focusing a static subject. Unfortunately, it was a no-go. In both bright/warm conditions and under the dimmer light the camera's autofocus system seems to favour, there was no perceptible difference in the stability of focus with a long lens on humans wearing jerseys. If we hadn't made notes about which sequences were shot on III-5-0 and which were shot on III-5-1, it would have been impossible to know which was which.

    But, with III-5-1 set, the camera's autofocus system will stop completely when it can't detect focus at all, which is as annoying as you would expect it to be when shooting something like night football on a field full of shadows.
To summarize, changes to C. FnIII-2 and C. FnIII-8 definitely help settle the jumpiness of EOS-1D Mark III autofocus, but they don't solve the autofocus system's most significant and troubling problems. Changes to other settings, including C. FnIII-4 and C. FnIII-5, don't seem to improve autofocus performance, and in fact choosing a setting other than the default may make things worse in some cases.
 
Q. Has Canon given you a statement about EOS-1D Mark III autofocus yet?
 
No. We've been in touch with our Canon contacts fairly constantly, both before and after we first published this article. So far, nothing in this communication would qualify as a statement from Canon. We've seen copies of the emails that Canon customer service staff are sending out in both the U.S. and Canada to EOS-1D Mark III customers who complain about the camera's autofocus performance. These emails acknowledge the reports of problems but don't say what Canon will do in response.
 
Q. Can we see some of the sequences you've shot showing the good and the bad of EOS-1D Mark III autofocus?
 
Yes. Below you will find links to nine different sequences of full resolution pictures. The ones shot at ISO 100 or 200 were taken on warmer days and mostly in bright sunlight. They demonstrate some of the camera's most serious autofocus problems. The ones shot at higher ISOs are taken in dimmer light coupled with cooler temperatures and show the EOS-1D Mark III working okay. These photos were shot with two different EOS-1D Mark IIIs set to capture CR2s. The CR2s were processed to JPEGs in Digital Photo Professional with Sharpening set to 3, then run through a Photoshop script that superimposed the active AF point on the frame, so you can see where the point of focus should be.
 
(For more information on why these pictures were handled this way, see here.)
 
The sequences were selected not because they're wonderful photos, but because they should give you a broader understanding of the EOS-1D Mark III's autofocus traits if you don't own the camera or don't shoot the kinds of things discussed in this article. Keep in mind, however, that these photos represent only a tiny fraction of the over 22,000 frames we've now shot with production EOS-1D Mark IIIs, so they can't and don't reveal everything about EOS-1D Mark III autofocus. We do hope they give you a better idea of what we've been seeing when shooting outdoor sports with this camera specifically.
 
Also note that these pictures haven't been adjusted in any way other than sharpening, and in some cases the colour and brightness would have benefited from more appropriate Digital Photo Professional processing settings in particular. In other words, please judge focus in these pictures, but try not to focus (pun intended) too much on image quality, since you might be seeing the result of less-than-optimal processing settings rather than a camera image quality problem.
 
Click on a thumbnail to download a ZIP file containing all the pictures in that sequence.

Click to download sequence Click to download sequence
Jog: Canon EOS-1D Mark III, ISO 100, 35 Pictures, 74.7MB (Photos by Rob Galbraith/Little Guy Media) Carry: Canon EOS-1D Mark III, ISO 200, 38 Pictures, 95.2MB (Photos by Rob Galbraith/Little Guy Media) Kick: Canon EOS-1D Mark III, ISO 200, 20 Pictures, 48.1MB (Photos by Rob Galbraith/Little Guy Media)
Click to download sequence Click to download sequence Click to download sequence
Bench: Canon EOS-1D Mark III, ISO 200, 10 Pictures, 39.5MB (Photos by Rob Galbraith/Little Guy Media) Chase: Canon EOS-1D Mark III, ISO 200, 13 Pictures, 35.1MB (Photos by Rob Galbraith/Little Guy Media) Leap: Canon EOS-1D Mark III, ISO 200, 8 Pictures, 21MB (Photos by Rob Galbraith/Little Guy Media)
Click to download sequence Click to download sequence Click to download sequence
Run: Canon EOS-1D Mark III, ISO 500, 24 Pictures, 80.4MB (Photos by Rob Galbraith/Little Guy Media) Catch: Canon EOS-1D Mark III, ISO 640, 12 Pictures, 41.7MB (Photos by Rob Galbraith/Little Guy Media) Ball: Canon EOS-1D Mark III, ISO 800, 14 Pictures, 53.6MB (Photos by Rob Galbraith/Little Guy Media)

These photos are for personal viewing and printing only. They may not be republished in any form without the permission of the copyright holder. This includes the posting of these photos onto another server.
 
To have a proper look at the photos, view them in a browser like Photo Mechanic that enables you to display them rapidly at full resolution. You can also open them in Photoshop, magnifying each to 100%, but this will be a cumbersome process. Shooting information is both in the EXIF metadata and in the Caption field.
Next Page: July 20, 2007 update
Go to Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 
Go to advertiser website.
2000-2013 Little Guy Media. Not to be reproduced without written permission.