1D Mark III Focus Shifts

If you follow the top of the digital SLR market, you no doubt are aware of the controversy surrounding the 1D Mark III autofocus problem. The issue with the Canon 1D Mark III autofocus is well understood as a bright light, fast motion problem. In his latest on the subject, Rob Galbraith opines that for slowly moving subjects in any light, the system is reliable in his experience. In general, I agree except for one significant environment: lit presentation stages. Both Pınar Ozger, who often shoots with me, and I have experienced the dreaded focus shift described by Rob shooting slow moving subjects at both O’Reilly Conferences and Apple’s WWDC.

If you haven’t seen them yet, check out the examples of the problem on Rob’s site, such as the ones of the baseball pitcher at the bottom of the first page of his August 1st update. These images show how the problem manifests.

To show you a random example of the focus hunt in my own work, here’s an example of the problem in two frames I shot of Sam Ramji of Microsoft at the O’Reilly Open Source Convention last month:

1d3focusshift.jpg

Two frames. Bam bam. Right after each other. These are crops of about 10% of the image area scaled for web view, but otherwise unprocessed. Same focus point located right on Sam’s forehead/nose area. Distance to subject was constant through the sequence as Sam was standing in place and I was parked shooting off of a monopod. Two entirely different focus solutions. The left shot is crap. The right one is is acceptable enough, though not great. The point here, however, isn’t to show great photographs, but to show how much drift the focus system can put in between two consecutive frames. I see this time and time again on both of my mark III bodies.

To describe this environment to those that haven’t been to either a recent O’Reilly conference or WWDC, the stages are lit with simple high front stage wash. Sometimes there’s high back lighting, sometimes not. Almost always, these stages are lit at a level where I shoot at 1/125-1/250th @ f/2.8-f/4 at ISO 1600. That’s 6-8 EV, which is well within the Mark III autofocus system’s rated working range of -1 to 18EV.

It’s a confidence killer. It really is. Sometimes, I’ll shot sequences where every frame is out of kilter. Other times, I get a streak of nothing but perfectly focused shots. I make up for it in volume, but I hate having to do that. Pınar and I have tried different shooting modes and haven’t found rhyme nor reason to either tickling it or working around it.

Maybe one day, I’ll try to get more rigorous about sorting things out. I can say pretty safely, however, that I’ve only really run into this in the lit stage presentation environments. Out in the world, the camera works fine. At least it seems to. I'm never sure, however.

Related Posts:

This is one of 187 blog posts on duncandavidson.com. If you care to read more, two posts I recommend are Dear Speakers, a set of thoughts for public speakers that I pulled together in March, 2009 and Tilting at the Windmill, One Last Time, a call to Flickr to include important EXIF and ITPC metadata in the photographs they provide to the public.

3 Comments

I haven't followed the problem in great detail. However, looking through the custom functions on my 1DsM3 there are quite a few regarding auto-focus. Such as tracking sensitivity and tracking method. Do any of those make a difference in this situation?

If yes, I assume Canon would have published guidance on that. So lack thereof means it may not matter.

user-pic

Indeed, there are a lot of custom functions. I have tweaked with them some and haven't seen any improvements. The issue is that the problem manifests itself most in the environments where I don't have a lot of time to deal with the situation which means that I've not been anywhere near methodical. I just punch at things to see if they can help or not.

But no, no help from them when I've used them. And w/o published guidance...

user-pic

I found this on another site. Any news following the firmware update?

“March 5, 2008
Updated Information

After the AF mirror Adjustment, including the updated firmware version 1.1.3, the EOS-1D Mark III AF function performed better in our tests than all previous EOS camera models.

Canon strives at all times to continuously improve our products in order to provide our customers with the best performance and value.

We are aware that some customers have raised questions about the performance of the EOS-1D Mark III AF system under certain conditions. We will continue to investigate, and look for opportunities to improve, the performance of the AF system to ensure the satisfaction of all of our customers”

user-pic

Leave a comment