Tests for Backfocus with the EF
After I had completed my usual series of focus and resolutions tests, several contributors to Fred Miranda Forums indicated that copies of the EF 50mm f/1.2L were backfocusing when tested using a Jackson 2004 FocusTestChart. I went back and ran additional tests to be sure I wasn't missing something that would impact photographic performance under conditions I use. Tests with the FocusTestChart were initially run with my lens at a working distance near 18 inches (46cm). I was able to duplicate backfocus reported by others using an EOS-1D Mark II and an EOS-1Ds Mark II. Results with the FocusTestChart at 46cm are shown below with autofocus and manual focus showing that the lens backfocuses at f/2 and f/4 when autofocused, but not when manual focused. I ran additional tests at 46cm using a USAF 1951 resolution pattern in a similar manner to other resolution tests in the review. I found that higher resolution was obtained by manually focusing the lens at a 46 cm working distance than by using autofocus. I then repeated the autofocus and manual focus tests at 129 cm (51 inches) and at 258 cm (104 inches). No effect of backfocus was identified at these greater working distances (see following graph using an EOS-1D Mark II at 129 and 258 cm working distance.)
|Lens resolution tests were run at the three working distances using a USAF 1951 target using a variation of the method specified in the methods section of this review. A Canon EOS-1D Mark II set at ISO 100, high jpg setting on tripod with cable release was used. For autofocus tests, the central focus point was centered on the target. Three exposures were taken at each aperture setting with defocus of the lens between each autofocus and shutter release. For manual focus, the lens was set to manual focus, and focus was adjusted while looking through the viewfinder with the aid of a Canon 2.5x magnifying angle finder. Three series were taked at each aperture. The sharpest image out of three for each test point was taken to calculate resolution. At 129 mm, there was more variability (greater numbers of slightly out-of-focus images) using manual focus than autofocus. Only data at f/2 is shown in the graph.|
I then checked out two other lenses with an EOS-1D Mark II and the Jackson 2004 FocusTestChart. My EF 50mm f/1.4 and EF 85mm f/1.8 lenses gave front focus and back focus results in the tests, respectively, when tested at the closest possible working distance. No backfocus was detected at longer working distances. I have never encountered autofocus accuracy problems with my EF 50mm f/1.4 or my EF 85mm f/1.8 lenses that I have used to make thousands of images.
I concluded that the Jackson 2004 FocusTestChart detected real focus problems at the very short working distance (46cm). These misfocus problems are of questionable significance because I almost never try to produce wide aperture images at the shortest possible working distance to produce tack-sharp images. With the EF 50mm f/1.2L focus function was normal at longer working distances (129 and 258 cm). The backfocus detected by the Jackson 2004 FocusTestChart is not relevant to most of the photographic applications I use lenses for in the real world.
Test with Jackson 2004 FocusTestChart at 45 Degrees to Target - 46cm working distance shows backfocus
Test with USAF 1951 Test Chart at 90 Degrees to Target - 46 cm working distance shows decreased resolution with autofocus due to backfocusing