How can I properly test whether a lens is decentered, and do these corner crops indicate a problem?

Asked 10/30/2018

5 views

2 answers

0

I bought a used Samyang 16mm f/2 and tried to check for decentering by comparing corner crops. My first test involved moving the camera on the tripod, and one corner looked noticeably softer than the others. After retesting more carefully, the top-right corner still appears less sharp than the rest.

How should a lens be tested correctly for decentering or other alignment issues? If one corner remains softer than the others in a careful test, is that necessarily decentering, or could it be caused by the test setup or focus differences?

Originally by Photography Stack Exchange contributor. Source · Licensed CC BY-SA 4.0

Photography Stack Exchange contributor

7y ago

2 Answers

4

There's no way to tell using the methodology you used. An improperly aligned lens is just one of several possible explanations for the differences between each of the images. Even among improperly aligned lenses, decentering is only one of several types of improper lens alignment that could produce such results.

When you move the camera around on the tripod, you change the relative distance from the optical center of the lens to different areas of your scene. This affects how sharp a particular thing will look because if the optical distance changes (because the optical center of the lens moved in an arc created by the position of the tripod's axis of movement) and the lens is not refocused, the point of focus will move in front of or behind your object. This is why "focus and recompose" can introduce focus errors that are noticeable when shooting portraits at very narrow apertures.

The only real way to do what you are trying to do is use a flat test chart that is properly aligned to be parallel to your camera's sensor. Then compare each of the corners of the test chart from the same image or series of images taken without moving the camera or target.

If the chart is not properly aligned with respect to your camera's sensor, then the results could indicate lens tilt, lens decentering, other optical alignment issues inside the lens or it might just mean improper alignment of the test chart or a combination of both - there is no visual difference between them in the results.

The easiest way to align a tripod mounted camera with a test chart is to use a stable, non-distorted mirror such as one solidly mounted to a wall. Align the camera so that the center of the lens in the reflection is exactly in the center of the camera's viewfinder. The marking for the center AF point will help in this respect. Then tape your test chart to the mirror without moving the mirror even slightly. You're now ready to start taking test shots as long as the camera does not move in relation to the test chart. For the purposes of this exercise, pointing the camera in a different direction is moving the camera because you have moved the optical center of the lens closer to part of the chart and further from other parts of the chart.

Keep in mind that most lenses demonstrate some field curvature. The field of focus is not a perfectly flat plane, even with a theoretically perfectly manufactured lens. Lenses that are highly corrected for field curvature, such as many macro lenses, still have a field of focus that looks more like a lasagna noodle than a flat plane. They're not perfectly flat, they're just more flat than uncorrected or less corrected lenses.

A few years back, Roger Cicala, the founder of LensRentals.com, wrote an informative blog entry about How to Test a Lens.

For more about the subtleties of field of focus, I recommend reading Roger Cicala's excellent series about it:
Fun with Field of Focus Part 1
Fun with Field of Focus II: Copy-to-Copy Variation and Lens Testing

Also related: Why You Can’t Optically Test Your Lens with Autofocus

I tested my lens according to your answer and edited my post with the results.

Based on the results of that test, I'd say you are still struggling with alignment issues between the camera and test chart. The test targets all skew in the same direction. That is, the top right is the weakest in all four quadrants and the lower left is the sharpest.

If the test chart were square with the camera one would expect the outer corners of each quadrant to be worse than the inner corners. In other words, the top right would look worse in the top right corner but the lower left would look worse in the lower left corner. It's possible tilt could cause the same thing, but a decentered element probably wouldn't.

As it is, the lower left corner is the sharpest, the top left and lower right corners are about equal and just a tad softer than the loser left. The top right corner is even softer. What would have been helpful at that point is to have refocused the lens until the top right corner was brought into focus as sharply as possible and then see what the other corners (and the center) look like.

Originally by user15871. Source · Licensed CC BY-SA 4.0

user15871

7y ago

0

AI Answer

Generated from our catalog & community — verify before relying on it.

Your first test method cannot reliably show decentering. If you move the camera around on a tripod and don’t refocus, each corner is no longer at the same optical distance from the lens, so focus shifts can make one area look softer even if the lens is fine.

A proper decentering test keeps the camera fixed and square to a flat target, with the target evenly lit and as parallel to the sensor as possible. Then compare corners under the same focus conditions.

If one corner is still softer in a careful, fixed-camera test, that can suggest a lens alignment issue, but it does not prove decentering specifically. Several kinds of misalignment can cause uneven corner sharpness.

As for whether it is “acceptable,” that’s partly subjective. Some variation between corners is common, especially in wide lenses, but a clearly weaker corner after a proper test may indicate the lens is not perfectly aligned. The key point is that your original results were not sufficient to diagnose it, and the retest is more meaningful only if the camera and target alignment were controlled carefully.

UniqueBot

AI

7y ago

Your Answer