How to Check Your Lens for Decentering
A quick and easy decentering test
By beholder3 in Articles and Tips on Jul 20, 2017
When we invest our money in high-quality camera lenses, we expect them to be able to capture sharp, clear photos. One of the production quality aspects of lenses is how well they are "centered"; a decentered lens can lead to inconsistent sharpness and should be exchanged or repaired.
This article describes a quick, easy, and reliable way to check for lens decentering all by yourself.
What Exactly Is Decentering?
Decentering describes a poor alignment of lens elements on the optical axis. All DSLR lenses actually contain multiple individual glass elements and groups which need to be assembled with high precision.
Below is a greatly simplified illustration of the problem, assuming a single-element lens:
What is the Effect?
Decentering causes the photos you take with the lens to be uneven with regards to sharpness, especially near edges and corners, and this usually does not improve by stopping down the lens.
A typical effect of decentering is that your images will seem soft on one side but not the other, or perhaps one corner is significantly softer than the rest.
When a full-frame lens is used on APS-C cameras, decentering may not be apparent since the camera does not see the corners of the frame.
How To Test for Decentering
We'll be using test photos to test for decentering.
- First you need to find a nicely textured subject (with enough detail to judge the sharpness) that's 30m to 100m away from the camera.
- Place your camera on a tripod, so the position of the camera is fixed
- Take 5 test photos, with your subject in the center and in each of the corners, respectively
Let's take a look at the procedure in more detail.
This setup allows you to gain some solid depth of field with all lenses. This large DoF is needed if you use the first image with AF (see below), because you swing around the focal plane in the other 4 shots and the subject therefore might get out of the DoF area. You should avoid using very distant subjects as heat or air pollution can blur parts of the image inconsistently.
Start with the subject in the center of the viewfinder and focus on it. While you might be able to get away with autofocus, but it's best done via manual focus and live view at maximum magnification. I suggest you do this using the widest possible aperture as this will highlight any differences.
Then take a shot and check if the central subject came out nicely sharp; re-adjust if needed.
Before you go on to the next four shots now you need to turn off autofocus to maintain the focus plane where it is now.
Then swivel the camera on the tripod so the subject is in one of the extreme corners of the image and take a shot. Repeat this three more times, with the subject sitting in another corner each time.
- A bell tower of a church works wonders for this type of test
- As with any test, it makes sense to use mirror up, IR remote, or cable release to avoid camera shake when using a tripod
- Always do this in good daylight to allow the use of fast shutter speeds and low ISOs
- Redoing it once or twice to confirm findings is a good idea
- You can do this whole procedure without a tripod. You don't even need a PC as you can inspect the images via the camera in 100% view mode and usually you'll already see any optical weaknesses. If you want to keep things as simple as possible, the whole test can be completed in about two minutes.
Now you take a look at the 4 shots where the subject was in a corner. You will get one of the following results:
- The subject is as sharp in the 4 corners as it was in the middle of the image.
Perfect lens. Not decentered.
- All four shots are softer in the corner but quite equally so. This can be both the result of a lens-typical sharpness drop-off in the corners or from field curvature.
The lens is fine, it is not decentered.
- There are major visible differences between the four shots with the subject in the corner. Perhaps two corners are different than the two others, or only one is.
The lens is decentered.
Samples of the 4 Corner Shots
These are 100% crops of the corners of the Pentax K-1's 36-megapixel sensor. If you care you might want to fix metering between shots providing an even exposure (I didn't) and compensate for vignetting. Generally image brightness is not what we are looking for here, though.
Decentered Sigma 35mm Art Lens
Good copy of the Sigma 35mm Art Lens
If your lens is significantly decentered you have a few options:
- Return or exchange it
- Send it in for repairs (warranty does cover decentering)
- Keep using the lens and crop/sharpen the corners/edges, or scale down
I personally have the impression that not all lenses can easily be centered through service, and this is also quite prone being an imperfect solution. In some cases it seems lenses are not physically designed to allow a lot of service adjustments and in some cases the service technicians seem to put in little effort or lack the proper tools. I therefore do this decentering test immediately after receiving a lens and send it back straight away if I find significant decentering.
Generally decentering seems to be worse with wide angle lenses, so it gets worse the shorter the focal length is. Thus I recommend you minimally check your lenses that are shorter than 50mm, especially ultra-wides below 24mm.
When looking at various lensrentals.com test results I further am under the impression that practically no lens is ever 100% free from decentering, so try not to be oversensitive here or you'll never be happy again. After all, this whole topic is only relevant if these aberrations shows up frequently in your normal images. All makers have a different view about what is "acceptable" in terms of tolerance, and they will usually settle a few grades below "invisible". Thankfully, first-party lenses usually follow the strictest standards, so you'll likely be doing more of these tests with your third-party lenses.
For help with this procedure or if you'd like others to take a look at your lens tests, please post in the lens forum.