What follows is an unscientific test of my new Canon 17-40mm f4.0 L lens. The object of this exercise was to learn how the thing works and what its basic characteristics are. Having come from the world of a point and shoot, stepping up to a 20D digital SLR camera has opened a whole new chapter of learning. An experienced hand might boggle at my embarrassing ignorance, but I really don't know how these sorts of lenses perform. Are they better (as in they take sharper photos) zoomed in or out? At f4 or f22? I hadn't a clue, so I set about to find out.

Please don't bother emailing me telling me that my test was flawed. I know it was, even if I don't really know how. I did this because I wanted to learn, not because I wanted to give any objective metrics to lens buffs who are obsessed with this sort of thing. I just wanted to get some guidelines on how to use the lens. On the other hand, I would be interested in hearing from people who feel I've drawn wrong conclusions from my test.

I went outside on a sunny day and taped a bit of newspaper to the wall, put the camera on a tripod in front of it and took photos of it at all apertures from f4 through to f22 with the lens at 17mm. I then stepped back a bit, zoomed in to 40mm, and took another set of photos. That was it - test complete. :o) I put the images into Photoshop and cropped out a 256x256 pixel portion from the centre of each image. What follows is a comparison of these images.

17mm Centre of the Image

Have a look at the image below. It shows the 100% crops from my test shots at 17mm, at, from top to bottom, f4, f5.6, f8, f11, f16 and f22. To my eye there is very little practical difference between f5.6 and f8. I'd say the sweet spot is f5.6, but even zoomed in at 200% it's hard to tell the two images apart. f4 is a little less contrasty than f5.6, but in any real photograph it would be impossible to tell.

At f11 there is a notable step down in sharpness and contrast, and this fall away is apparent all the way down to f22 in a linear fashion. That's not to say f11 is in any way poor - it's not - but there is a visible difference in the test images. Real world images are likely to be virtually indistinguishable from each other. Down at f22 there is a marked difference that would show up in real world images.

40mm Centre of the Image

There's another set of images below, this time the centre crop from the photos taken at 40mm. Again, from top to bottom, they are f4, f5.6, f8, f11, f16 and f22.

The obvious thing to notice here is that at 40mm f5.6 is has notably more contrast and sharpness than f4.0. I was surprised by this and wonder if I did something wrong in the test. If I did I don't know what. The EXIF information from the originals looks like it should. f8 is a little softer, but it's hard to tell. The contrast and sharpness drops away as the aperture closes and at f22 it is really quite soft, relatively speaking. I actually noticed this is a real world photograph before I did this test - in fact that photograph was one of the things that prompted me to do it.

Again, in the real world there probably isn't any noticeable difference between f5.6 and f11. f4 is about the same as f11. f5.6 is once again the sweet spot but to all intents and purposes real world photos taken between f4 and f11 are likely to the indistinguishable, at least to my eye.

Comparing 17mm to 40mm

Putting the images side by side isn't possible for readers of this webpage, but I could do so with the originals. There is no doubt that the lens produces sharper, more contrasty images of this subject at 17mm than 40mm. This is apparent from f4 through f22. The difference isn't huge, but it would probably show on some types of real world image. Especially now I know about it and will be looking for it!

So what did I learn?

Well, that's a Sunday afternoon gone. What did I learn from it? Well, it's hard to draw specific conclusions from such a limited, unscientific test, but it's clear to see that when it comes to zooms, even Canon's 'L' glass is a compromise. This lens appears to be at its best at the 17mm end of its range, and at f5.6. There's little real world quality difference between f4 and f11. Performance drops off at smaller apertures, but it's still thoroughly usable right down to f22.

The 40mm end of the range is not quite as sharp as the 17mm end, but it's not far short. Again, f5.6 is the sweet spot, and to all intents and purposes there's little difference between f4 and f11. The lens is clearly at it's most limited at small apertures at 40mm. I was kind of aware of that, but it's good to see it reflected in my test. It's good to be aware of these things.

I'm aware I've been spoilt here. I only own one lens, and in fact, as of this writing, it's the only one I've ever used. It also happens to one of Canon's 'L' lenses, which makes it amongst the best on the market. And I still pull out and examine its limitations! Perhaps I should punish myself by buying and testing a whole load of top quality lenses. :o)

I posted a link to this webpage on the rec.photo.digital newsgroup and asked for any feedback anyone might care to offer. A few people responded with some useful additional information and pointers.

Everyone who expressed an opinion said my test was reasonable, given what I was looking to achieve. They also said that my results were as expected. Or, at least, no one said they suggested something that would be out of the norm.

John P Sheehy said that with his 17-40mm lens it was visibily softer at f10, and that f16 was totally blurred. I don't think I'd call my results total blurring, but John's general observations follow mine. As he said, "The smaller the aperture, the deeper the range that is close to maximum focus, but maximum focus itself also softens, the smaller the aperture." That makes perfect sense. I'll strive to use f8 or larger apertures unless I really need the DOF.

Someone called Stacey offered results from similar tests he did on his 11-22 zuiko lens, and confirmed that it produced pretty much identical images from f4 through f11 but gets softer past f16 down to f22 where he wouldn't use it unless he had to. He also said that the 50-200mm zuiko pro lens performed the same at both long and short ends of its range, which is a little more than mine can claim.

So, lenses are a compromise, and the places the compromises may happen are:

• wide open (the smallest aperture number), although my 17-40mm lens is not bad in this respect;
• at one end of the focal range or the other. My 17-40mm lens is clearly a lot better at 17mm than 40mm;
• as you stop down past a certain point - about f11 in the case of my 17-40mm;
• at the edges of the frame, although my test never considered this.

So keep away from these areas. For maximum quality use the shorter focal length where possible, and keep aperture larger than f11. But don't lose track of the fact that I'm talking about extremes here. This 'L' lens is perfectly usable throughout it's performance envelope!