Note: the best way to navigate this set of pages is to use the “Articles” section in the right-hand panel. It has links to all the pages and their titles.
For the last few years, I have followed Internet discussions related to lens screening. By screening, I mean the process that a photographer employs to decide if a recently-purchased lens is worth keeping or not. For the purpose of this set of pages, I’m concerned with identifying a lens that is improperly constructed, not one that is poorly designed — that would involve skills that I don’t have and jumping into a morass of value judgments from which I wish to stay far, far away.
In my observations and interactions with non-technically-inclined photographers screening their own lenses, I have reached several — unverifiable, I admit — conclusions:
- Most people screening lenses are not testing what they think they’re testing
- Most new lenses that are returned as defective are perfectly good.
Maybe I’m wrong about this. But I don’t think so. I think most of the testing that people do has setup or operational issues that yield false positives. There are methods to cross-check to weed out these false positives, but people almost never take the trouble to perform them.
[Added 12/9/19. I recently came across a particularly egregious example of bad lens testing. The person was using the Gletscherbruch test, and reported an extraordinary number of bad Sony lenses, which he blamed on Sony’s total lack of quality control. He was conducting the test with a target at least a kilometer away in an urban environment over heat sources, through a window, using a filter on the lens, handheld with IBIS on, at an unstated shutter speed, using different post processing for the various corner captures, and not checking his results by inverting the camera. Is there any wonder he thought he found so many bad lenses?]
For the past several years I have been experimenting with procedures for lens screening. Using an idea provided by Prof Hank at DPR, I posted a simple decentering test two years ago. Most people were able to perform it. Some weren’t, and each time someone failed, I modified the instructions to make them clearer. It’s now a pretty good test, but the key problem with it is that most of the lens construction problems that people encounter are not related to decentering, even though many folks say a lens which produces sharp images in the center of the frame and soft ones someplace (or every place) else is ‘decentered.”
I’ve spent much of the last month or two developing a more complete lens screening test that gets around many of the problems with existing tests. Here’s an exposition on what I think those problems are. The test that I’ve come up with needs a target which can be purchased for less than $30 or printed by anyone with a large-format printer. You’ll find files that you can print here. You’ll also need a stand for the target, and tape to make sure the target doesn’t blow off the stand. All the other equipment strictly necessary is likely to be found in your equipment locker. Here is a complete list. The test requires no precision alignment. It is compatible with just about any raw processing workflow.
Want to get started?
While it’s not required, I recommend starting by reading about the ideas behind the test. You can find that here. There are three pages that walk you through performing the test, one describing how to capture the images, another telling you how to process them, and one describing how to interpret the images. You can also look at some examples.
I don’t have enough test results that show bad lenses. If you run these tests for yourself and find one of your lenses to be improperly constructed, I’d be interested in seeing your raw files. Please contact me and I’ll figure out how to make that happen.