On August 1st, I bought a Canon EF 50mm F1.2 L USM Lens. I’d been thinking of getting this lens for a long time, though had held off because I heard that it back focused when trying to focus on things close to the lens. For much of my current work though, if I’m working close up, I often use LiveView and manual focus anyway, so I thought I’d go for it. After all, I usually put claims of back-focusing down to bad technique. I’ve not had a single lens that back focuses myself, and I’ve been through some glass.
Anyway, I’d been pretty busy over the last 10 days, but as I was coming towards the end of my two weeks grace period to take the lens back if there were problems, I figured I’d do a few tests. I set up my tripod and grabbed the Focus Test Chart that Tim Jackson kindly made available a few years ago, and shot a few tests. (UPDATE: I removed the old broken link. Try Jeffrey Friedl’s excellent chart instead.) To cut to the chase, I took the lens back and got a refund.
I have to admit I thought it was going to be a bit sharper wide open than it is. My 85mm F1.2 is much sharper wide open. Still, this is not the reason for my disappointment. Read on to see what happened…
First, here’s an animation of a cropped and of course re-sized image, showing most of the test chart page. This is really to let you see the big picture (pardon the pun). To see the 100% crop, you’ll need to scroll down a little more. First though, note that I used the 5D Mark II’s Autofocus Micro-adjustment feature to set the camera to bring the focus forward by 2 steps, and that was enough to make it possible to auto-focus on the black line that runs through the middle of the Test Chart. From this point, there was no more back-focusing, which is a good start, I’d hoped. Note, the 2 steps in the micro-adjustment does not necessarily equal 2mm, though it seemed to in my case today. Apparently the distance that a step adjust the focus depends on the maximum aperture of the lens (according to the 5D2 manual).
Now that you have an idea of what the entire chart looks like, take a look at a 100% crop of the right side of the chart (below). As we reel through apertures 1.2, 1.4, 2.0, 2.8, 4.0, 5.6 and 8.0, although it’s very soft to start with, you can see that the focus starts with the 2mm text both above and below the line of small text pretty much the same. They are both blurred, but to roughly the same degree. That’s because I did the micro-adjustment, and I used LiveView to ensure that the focus was set on the thick black line on the chart before starting to shoot. As I said, I didn’t start back-focused.
I didn’t adjust the focus again at all as I shot through apertures 1.2, 1.4, 2.0, 2.8, 4.0, 5.6 and 8.0. I actually shot F11 and F16 as well, there wasn’t much point in including these apertures here. Take a look, and see what you think of these results before reading on below.
As you can see, despite the fact that I focused as closely as possible on the thick black line on the chart, which is the same distance from the camera as the text that says “This text should be perfectly in focus” in this crop, the text starts to sharpen up from the back “2mm” text. Then the depth-of-field continues to increase backwards, and doesn’t really come forward to sharpen up the front “2mm” until we get to around F8.
Here’s what’s happening; as the aperture closes, the point of focus is slipping slowly backwards. By the time the lens starts to sharpen up at F2.0, the focus is already back as far as the back 2mm text, and because the focus continues to move backwards, it takes an aperture of F8 to give us enough depth-of-field to encompass the front 2mm text.
Of course, the depth-of-field should be distributed by around 1/3 in front and 2/3 behind the point on which we set the focus. But that’s not what we’re seeing here.
Because of this “dynamic” shift backwards, the only way to manually correct the focus while shooting is to hit the depth-of-field preview button near the lens mount on the camera body, and focus with the button depressed. I tested this at F2.0, F4.0 and F5.6, and it is possible to focus correctly, but when you look through the lens or check with live view without the DoF preview button pressed, and therefore the lens automatically set to wide open at F1.2, the focus runs off.
I was prepared to live with having to manually focus this lens when shooting things close to the camera, but I am not prepared to do all of my focus adjustments with the depth-of-field preview button pressed. That to me is just bad design.
I do just want to reiterate, that this problem only seems to happen when focusing at around 50cms, as in my tests. If you do not intend to work this closely, the lens is probably fine. There are many people getting great results with this lens at longer shooting distances. I might buy another copy at some point, and just forget about using it close up.
I am not prepared to put up with this at the moment though. I threw my test shots onto my laptop and took it to show the folks at the camera shop. Even though I’ve heard that Canon advises manual focus adjustment at close distances with this lens (though I can’t find a link to this statement anywhere), I was hoping that the the people at the camera store would tell me that the focus should not dynamically shift backwards, and suggest that I might have a dud lens. It would have been the my first dud lens from Canon, but I was hoping that would be their response. They didn’t say that though, and they couldn’t say that replacing the lens would make it any better, so I asked for my money back.
Boy am I glad that I checked this lens before my two week grace period ran out. I have to admit, I’m sad about this. I wanted to love this lens. I like the focal length, and I wanted that insane bokeh for some work that’s coming up, as well as some of my flowerscapes, but I just couldn’t bring myself to try another copy today. Maybe I’ll buy another copy later. For now, I’ve got a little space in my lens cabinet that looks awfully sad.
Share this post with your friends!
You can also use the social icons below...