042/366: Distortion

For the 42nd installment of this photo-a-day thing, it seems appropriate to try to do some SCIENCE! to get an Answer. So, here’s a composite of a bunch of images I took yesterday in order to investigate something:

Graph paper shot with several different lenses, to look for distortion of the images.
Graph paper shot with several different lenses, to look for distortion of the images.

OK, this needs some explanation…

So, I do a lot of shooting with moderately wide-angle lenses (either a 10-18mm zoom or the 24mm fixed “pancake” lens), because SteelyKid and The Pip tend to want to be right on top of me a lot of the time, and it’s hard to get good pictures with them fully in the frame otherwise.

Occasionally, images shot with these lenses look a little odd, which is not all that surprising, really, because it’s a tricky optics problem to get a really wide field of view without some fish-eye distortion. It’s tough to quantify that, though, so yesterday I spent a little time trying to make an experimental check of how much these lenses distort things.

The composite above is spliced together in GIMP from the top just-over-half of a bunch of individual shots of a sheet of graph paper, shot with a bunch of my camera lenses. From top to bottom, these are the 10-18mm zoomed all the way out, the 10-18mm zoomed all the way in, the 24mm pancake, the 50mm fixed lens, and the 55-250mm telephoto zoomed all the way out. (The paper doesn’t fill the frame in that one because it wouldn’t focus any closer than where it is… I didn’t test the 18-55mm lens I got with my original camera, because I was getting tired of swapping lenses and repositioning the graph paper.) These were all taken using a tripod pointing more or less straight down; the bits of yellow are Post-It notes I used to label which lens was which.

What’s the upshot of this? Well, I’m really impressed with Canon’s optical engineers. These are all really good, in terms of producing images of straight lines that actually look like straight lines. I mean, there’s some visible bowing of the lines at the edges of the frame in the widest of the wide-angle lenses:

The 10-18mm lens zoomed all the way out.
The 10-18mm lens zoomed all the way out.

But that same lens zoomed in all the way is way better:

The 10-18mm lens zoomed all the way in.
The 10-18mm lens zoomed all the way in.

(Admittedly, the paper doesn’t get all the way to the edge in this one, so there might be more distortion in the corners that it’s not really picking up…)

And here’s the 24mm, which is the other wide-field lens I’ve been using a lot:

The fixed 24mm lens.
The fixed 24mm lens.

Again, a little bit of bowing of the lines in the top right corner of the image, but otherwise really good.

So, you know, kudos to Canon. This also makes me want to get my hands on a real fish-eye lens to see what more significant distortion looks like, but I may be approaching “admit you have a problem” levels with my lens acquisitions…

4 thoughts on “042/366: Distortion

  1. Welcome to the next level of camera geekitude! Hope you enjoy it as much as the rest of us. (For some irrational value of ‘enjoy.’)

    If you’re curious about where lens analysis leads to, check out the Lensrentals blog. I don’t think he’s spent time on the lenses you’re looking at — he tends to work on the higher-end, full-frame lenses — but for example here’s a ‘brief’ look at the Sigma 24mm f/1.4 I’m currently coveting.

  2. Are you looking at RAW files or JPG output from the camera? Lens and camera manufacturers know a great deal about lens distortion and have lens profiles based on optical models and lens calibration testing. The lenses might actually have much more distortion than you see in your images but the camera (for JPG output) or your processing software (for RAW files) corrects for it using the lens profile metadata.

  3. I’m looking at the .jpg files only, because I have disk space issues as it is, and haven’t felt like figuring out how to work with RAW files. I may well give that a shot at some point during this year, but the thought of having to wrangle the installation of yet another software package is a big activation energy barrier.

    That does suggest another set of photos, though, assuming I can figure out a good way to mount a lens separate from the camera…

  4. Since your camera is a moderately priced DSLR, you probably have fairly good lenses. I have a pocket sized ultra-zoom camera that almost certainly has a fair bit of lens distortion since it goes from a fairly wide angle to a telephoto with the optics in a tiny package. The pictures still tend to have straight lines thanks to the digital image processor correcting things. This means that the pixel per area ratio varies with the lens setting and position, but it lets a camera costing less than a moderately serious lens take surprisingly good photos.

Comments are closed.