Previous page Next page

Focusing accuracy

There has certainly been a lot of talk about AF issues with the EOS 10D, our own tests showed that while our test EOS 10D did slightly front focus* it was minimal and certainly within acceptable tolerances for auto focus. Since then many people have worked on auto focus tests and charts. My opinion is that (a) there were a very small percentage of cameras and or lenses which were miscalibrated (these should simply have been returned to Canon for testing), (b) there were quite a lot of tests carried out without a proper understanding of AF, where the AF sensors are and what is acceptable. The diagram below was produced from one posted by a Canon employee on another digital camera forum. It shows (gray) the actual size of the AF sensors compared to the indications on the viewfinder focusing screen (red). Thus when testing AF with small subjects it would be relatively easy to 'catch' the bottom of the sensor with detail slightly closer or slightly further away.

* Front / back focusing are where the camera indicates an auto focus lock but the actual focus position produced is slightly in front or slightly behind the actual ideal focus position. This can be caused by errors in the camera or a problem with the lens.

Our test

AF Test Target

For this test we are using a variation of various 45 degree tests seen elsewhere. Ours uses a simple rule print with a focusing target placed at 45 degrees from the rule and facing the camera directly. This target was designed to be easy for the camera to pick out as well as being large enough (3 x 2 cm) not to allow the AF sensors to pick up background detail. The camera is placed at exactly 45 degrees from the chart (facing the AF target) at a distance of 0.5 m (from the AF target to the focal plane).

The chart is then shot at both wide angle and telephoto at a large aperture (to keep the DOF as small as possible) with center point AF selected. It is then possible to examine the image and see whether the AF target is in focus, it is also possible to see the DOF on this chart. Each small line on the ruler is 1 mm, larger line is 10 mm. Each crop is produced by reducing the image by 50% (to make more of the image visible).

Canon EF 18 - 55 mm F3.5 - F5.6

At 18 mm (F3.5): Good
At 55 mm (F5.6): Good (DOF more to back)

Canon EF 28 - 70 mm F2.8 L USM

At 28 mm (F2.8): Good
At 70 mm (F2.8): Back focus

Canon EF 50 mm F1.4 USM

At 50 mm (F2.8): Good (DOF more to front)

Overall results were as expected, while there is a little variation in the position of DOF (slightly in front or slightly behind) the actual target was always sharp (apart from our 28 - 70 mm F2.8 L which appears to have a problem at 70 mm).

Other lens tests (on the EF-S 18 - 55 mm)

We don't normally perform tests such as distortion, macro focus and fringing in our digital SLR reviews, this is because we are reviewing the SLR body, lens choices are up to the individual. However Canon are promoting the EOS 300D Kit which includes the EF-S 18 - 55 mm lens, this Kit has everything you need (apart from a Compact Flash card) to begin shooting right away and is likely to be the most popular lens for first time Canon SLR buyers. Thus I believe it is important to see how the combination of the EOS 300D and the EF-S 18 - 55 mm performs in our normal prosumer lens tests. Note that due to problems with our test equipment we were not able to complete the vignetting / light falloff test, we hope to add it later (to be honest the 18 - 55 mm lens doesn't appear to suffer light fall off).

Macro Focus

The EF-S 18 - 55 mm lens is actually a fairly respectable macro lens at telephoto providing a fairly close coverage (by D-SLR zoom lens standards) of 67 x 45 mm. The macro tests below are using our new macro focus test chart and measurement system; each line on the grid is 10 mm, taken at shortest subject distance in macro mode.

Wide angle (18 mm) - 208 x 138 mm coverage
15 px/mm (375 px/in)
Distortion: High
Corner softness: Medium
Telephoto (55 mm) - 67 x 45 mm coverage
45 px/mm (1155 px/in)
Distortion: None
Corner softness: Medium

Barrel and Pincushion Distortion

The EF-S 18 - 55 mm lens exhibited 1.0% barrel distortion at full wide angle, this was to be expected considering the focal length at this zoom position and the low cost of the lens. At telephoto we managed to measure some pincushion distortion but it's unlikely to be visible in most everyday shots. Distortion was better than we would expect of a wide angle 3x zoom prosumer digital camera lens.

Wide angle (18 mm)
1.0% Barrel Distortion

Telephoto (55 mm)
0.3% Pincushion Distortion

Purple Fringing (Chromatic Aberrations)

The EF-S 18 - 55 mm lens did exhibit some chromatic aberrations near the edge of the frame at full wide angle (18 mm). While not as objectionable as seen on some consumer digital cameras it could still be visible as a five to six pixel wide band along the edge of dark detail against an overexposed sky.

Fringing visible around contrast, 18 mm Our standard chromatic aberration test shot

Vignetting / Lens Shading

The EF-S 18 - 55 mm lens did exhibit some visible lens shading at full wide angle and maximum aperture (18 mm / F3.5) however at smaller apertures or slightly more zoom this drops to levels which would not be visible in everyday images. Our vignetting measurement is made by taking the average luminance value of the corners of the image and comparing it to the average luminance at the center of the image, any difference greater tha 20% may be visible in everyday shots.

Wide angle (18 mm) Telephoto (55 mm)
Wide angle, F3.5 (wide open)
24% average fall off
Telephoto, F5.6 (wide open)
13% average fall off
Wide angle, F8
12% average fall off
Telephoto, F8
7% average fall off
Previous page Next page
I own it
I want it
I had it
Discuss in the forums