Canon Lens Auto Focus Accuracy – Part 1

Back in May I wrote about the new Canon 70-200mm f4 L IS lens I had chosen and my reasons for making the choice I did.  After taking it out on a few non-critical sessions I haven’t been blown away by its sharpness – it’s good, but perhaps not quite as crisp as I was hoping for.  I started wondering if this was down to my unrealistic expectation or if there was something wrong with the lens.

Then one day I was out on the riverbanks in Durham and saw a heron fishing in the shallows.  I was a bit far away, but thought I’d grab a few frames just for interest.  Reviewing the shots I’d just taken zoomed in on the camera’s LCD, the heron was clearly blurred even though I was absolutely certain I had auto-focused on the body (I was too far away to pick up the eye which I would normally aim for).  What was more, the detail in the background was sharp – and it was consistent, shot after shot.  Suddenly I panicked – did my new lens have a serious back-focus problem?

Grey Heron
The shot that started it: after default sharpening in Capture One, this Heron still looked soft while background details look sharp (100% crop)

After some thought, I realised it was possible that there might be other reasons for the shots I got.  For example, was it simply that the subject was too small or too low-contrast for the focusing point and it happened to pick up the contrast in the background instead?  Was it poor technique on my part?  I eliminated some other causes of blur such as using a long shutter speed, but by now the seed of doubt had been not only planted, but had germinated, put on growth and flowered.  I had to test the lens to be sure.

Testing Process – attempt 1

Canon themselves don’t give an “official” process in any of their documentation (at least, as far as I could find).  I suspect this is because to calibrate lenses accurately you need the facilities and dedicated (read, expensive) equipment that you’ll find somewhere like… oh, I don’t know, a Canon Service Centre.  They do give some unofficial guidance (see Chuck Westfall’s comments from a forum on dpreview) but after that it’s up to you.

I found plenty of references via Google and I picked on Jeffrey Friedl’s blog which has some detailed explanations of how to do focus testing and a range of useful testing charts.  Initially, I didn’t have time to set up anything too neat, so I settled for a slightly rough & ready version of a test to establish whether I really had a back-focusing problem; it went something like this:

  1. I set my 70-200mm lens to 200mm, aperture at f4 and my camera to capture the largest RAW files.  Focus was set to single-shot and centre focus point only.  Image stabilisation was off and autofocus on (obviously!).  Mirror lock-up was enabled.
  2. I attached the chart to a hardback book and propped it at an angle.
  3. I set the camera on a solid tripod with a remote release.  The height of the tripod was set so the centre focus point at same height as middle of chart.
  4. I aligned the centre focus point with the middle bar on the test chart, trying to make sure the camera was square to the test chart.
  5. I manually set the lens to focus on infinity.  I then used the remote release to autofocus and release the shutter.  I repeated this three times to allow for variations in the accuracy of the focusing system.
  6. After shooting, I downloaded the CF card and reviewed the shots in Capture One.  I converted to black & white and enhanced contrast to make focus changes as obvious as possible but switched off sharpening so as not to interfere with the focus changes through the test chart.

Initial Results – do I need to replace my lens?


Rough focus test showing back focus
Rough focus test showing back focus (100% crop) - compare the two 8s on the left-hand scale

The example picture above shows a typical results of my test – the chart wasn’t quite square to the camera, but it was sufficient to give me the indicator I needed.  The target line is in the zone of sharp focus, but right at the front of it – to my eyes, the sharp focus region is centred around the 3 behind the line, indicating back-focus.  This satisfied me that there wasn’t a significant problem with this lens after all – the focus target was in the zone of focus, just not in the middle of it – so it wasn’t the absolute disaster I feared.

Reflecting on the heron shot again, I think there were a couple of factors which contributed to my initial concern.  Firstly, the heron itself was over-exposed.  In my eagerness to get the shot, I didn’t consider that the pale tones of the plumage were a small portion of what was otherwise a dark scene, so the bird has ended up at least a stop over-exposed.  I think this accounts for the slight “halo” which seems to surround it.  Secondly, I don’t routinely apply any in-camera sharpening (I prefer to work on the RAW files afterwards), but without it zooming into a small subject on the camera’s LCD naturally gave a softer appearance; sharpening in Capture One made a big difference.

However, I still want the best from my kit in producing images and I clearly had an issue so I decided that I needed to calibrate the lens.  Fortunately, the Canon 5D mark II features micro adjustment (Custom Function C.Fn III-8), which allows the user to set an individual calibration for each of up to 20 lenses.  This would mean some more scientific measurements and a better process… which I’ll take you through in part 2.