3.11 Lens Calibration

You may recall from section 2.6 that SLR cameras use a clever trick to determine exactly how much the focusing element of a lens needs to be moved in order to achieve perfect focus (with some caveats, which we won’t repeat here).  In order for this scheme to work properly, the camera needs to know some things about the particular lens that’s currently attached—most importantly, it needs to know the rate at which movements of the focusing element translate into focal point movements of various magnitudes.  Given knowledge of this mathematical relationship, as well as knowledge about the current position of the lens’ focusing element, the camera can determine the precise electronic signal that it needs to send to the lens to bring the lens into perfect focus (in ideal conditions). 
    Unfortunately, lenses and cameras in the real world aren’t perfect.  There are manufacturing defects that occasionally occur, and even when a lens or camera has been manufactured according to engineering specifications, individual components in either system can sometimes be
knocked out of alignment, meaning that they no longer satisfy engineering specifications (possibly due to being dropped on the ground, etc.).  In these cases, what the camera thinks it knows about the lens’ current state may not be 100% accurate (or, in engineering parlance, not within operating tolerances). 
    The result is often a camera/lens system that doesn’t focus perfectly.  You might, for example, notice that most of your bird photos seem to be slightly out of focus, while another object in the scene appears to be in better focus than the bird (even though you didn’t point the AF sensor point at that object).  In this case, we’d say that the lens or camera is out of calibration, or out of collimation.
    In order to fix this problem, you can either send the camera/lens in to the manufacturer’s service center for re-calibration, or (if the camera offers this feature) do the re-calibration yourself.  An increasing number of DSLR cameras nowadays allow the user to calibrate either the camera or individual lenses to be used with the camera.  We’ll focus (no pun intended) here on the latter problem (calibration of lenses), but the procedure outlined below is equally applicable to calibration of the camera.
    The method by which some cameras allow you to calibrate lenses is by setting a microadjust parameter for each lens.  In the case of the Canon 1D Mark III camera, the microadjust screen appears as shown in Figure 3.11.1, below.

Fig. 3.11.1: Microadjustment of focus.
This particular lens/TC combination (600mm lens
+ 2
× TC) was consistently front-focusing, so I set a
microadjust of +5 in the camera, and the focus is
now dead-on
.  Finding the right setting was a
process of trial-and-error.

This feature allows you to correct for front-focusing or back-focusing.  Front-focusing is when the camera tends to focus on a point somewhat in front of the subject (i.e., between the subject and the camera), whereas back-focusing is when the camera tends to focus on a point behind the subject (i.e., further from the camera than the actual subject). 
    In the figure above, I’m correcting a lens calibration error in which the camera tends to front-focus.  To compensate for this front-focusing tendency, I’ve dialed in a microadjustment of +5, which will force the camera to focus 5 units backward from where it normally would for this lens/TC combination (in this example I’m calibrating a 600mm lens with 2
× teleconverter attached).  Because teleconverters can impose their own front/back focusing biases, some cameras allow you to microadjust combinations of a lens and teleconverter, as in the previous figure.
    The real problem is determining precisely how much of a microadjustment is necessary in order to produce optimal autofocus behavior.  In practice, this is almost always done via trial and error.  That is, you dial in a microadjust setting, see how that affects the photos you subsequently take, and then progressively change the setting based on the effect of previous settings.  A popular way to do this is by using a focus chart, as shown below.

Fig. 3.11.2: A sample focus chart.
Right: the full focus chart, printed on 8.5"
×11" paper and
affixed to a 45-degree surface.  The red square represents
the AF sensor point (center) used to autofocus the lens.
Left: a zoomed-in view of the left ruler portion of the chart.
The most in-focus region is around 40mm, indicating a
severe front-focusing condition.

Focus charts can be readily found on the internet.  Once you find one, you download it onto your computer, print it onto standard 8"
×11" paper, and then affix it to a surface positioned at a 45-degree angle from the horizontal.  (The actual angle isn’t that important, so you can just estimate the angle by eye).  Then set your camera up on a tripod so that the lens is level with the focus chart.  Autofocus on the center of the chart, and take the photo.
    For the particular chart used in the figure above, the camera needs to be auto-focused on the black bar in the center of the chart.  Along either edge of the chart is a ruler which tells you how far out of focus the resulting image is.  What you need to do is to estimate where, on that ruler, the most in-focus point is.  If your camara and lens are perfectly calibrated, then (in ideal conditions) the reading on the ruler will tell you that the AF is out by 0mm.  As you can see in the figure above, the camera/lens combination used for this particular test was out by about 40mm, since that’s the reading on the ruler that (to my eyes) is the most in-focus.  I sent this particular camera (which was manufactured by Canon) in to Sigma’s repair center (since I suspected the Sigma lens was what was out of collimation), and they performed their own tests, concluding that the camera was indeed mis-calibrated, and in their words was
off the charts in its degree of mis-calibration.  That was a brand-new Canon camera.  After sending it to Canon, they agreed that it was mis-calibrated, attempted to re-calibrated it themselves, and returned it to me, whereupon I found that it was still mis-calibrated.  Calibration of cameras and lenses can be a real headache.
    Now, there’s a potentially serious problem with the approach described above for determining focus error in a camera/lens.  If you look closely at the zoomed-in ruler view in the figure above, you’ll see that the paper itself has texture.  And that texture was captured by the camera’s imaging sensor (otherwise you wouldn’t be able to see it in the image!).  Whether the AF sensors have enough resolution to
see that detail is another question, but if they can see that paper texture, then interpreting the results of focus experiments using focus charts like the one shown above becomes more difficult.   One useful suggestion from a fellow photographer is to print the chart onto matte paper, to reduce the amount of texture apparent on the surface of the paper.  Another suggestion is to focus on the sparser regions of the chart where there are fewer printed lines that could conceivably distract the AF sensor.  The key is to be sure that the camera’s AF circuitry has indeed focused on the part of the chart that you think it has focused on.  Any inferences about AF error must necessarily be predicated on the assumption that the camera intended to focus on the image features that you’ve isolated.  Otherwise, conclusions based on these types of tests are less than iron-clad.  It’s also a good idea to shoot with a shallow depth of field (i.e., a wide aperture), to ease interpretation of the results, though obviously autofocus is itself performed at the widest aperture possible with the lens currently attached (i.e., stopping down to your selected aperture doesn’t happen until after the camera has already achieved focus).
    Note that in the figure above we’ve superimposed the AF sensor box over the frame, to show that the center AF sensor was used to focus on the black bar in the focus chart.  Unfortunately, the AF sensor boxes that you can see in the viewfinder of your camera are only meant to indicate the relative positions of the AF sensors, and may not accurately represent the absolute sizes of the AF sensor regions in the image.  Thus, in the example above, the center AF sensor point may have included quite a sizeable region around the red square shown in the figure, and if the AF sensor can resolve details as fine as the paper texture apparent in the figure, then it may be difficult to be sure that the camera is in fact front-focusing.
    For this reason, I determine my microadjust settings in the field using birds instead of focus charts.  After photographing a bird perched on a branch that is not parallel to the camera, I look at the branch (in the image replayed on the camera’s LCD) to see how much the end of the branch closer to the camera, versus how much the end further from the camera, is in focus.  If the bird appears to be out of focus while the portion of the branch closer to the camera appears to be in focus, then I conclude that the camera is front-focusing.  In that case I change the microadjust setting by a few clicks in the appropriate direction, and then take another photo of the bird and see whether that has improved the focus.  By then the bird may have flown off, but in that case I just wait until I can find another subject.  Individual
clicks on the microadjust scale don’t seem to make an enormous difference in focus, so while fine-tuning the microadjust setting it’s typically still possible to get some usable images (i.e., usable for other things than just focus calibration).
    If your camera doesn’t provide a microadjust feature, or if the focus error is so large that the microadjust setting can’t provide a large enough correction, then you’ll need to send your camera and/or lens in to the manufacturer for re-calibration.  Some manufacturers recommend that you send in both the camera and the lens (or even all of your lenses) so that they can be jointly calibrated so as to provide optimal performance when used together.  I’ve never done this, but I’ve heard that it can result in noticeable improvement in focus accuracy.  For units no longer under warranty, this may incur a fee from the manufacturer.
    Note that the use of third-party lenses may complicate the calibration process if done by the manufacturer(s), since you’ll be dealing with two different manufacturers (one for the camera, and one for the lens).  As mentioned earlier, I once tried to resolve a focus calibration issue for a Canon camera in combination with a Sigma lens.  Canon insisted that the Sigma lens was to blame, while Sigma insisted that the Canon camera was to blame.  In the end, I decided to trade in both the camera and the lens for a camera/lens combination from the same manufacturer, with the camera being a model with an end-user microadjust feature.  This has worked flawlessly for me.