Help! 16-55 has issues, possible decentering? (samples)

Testing for this is not easy and that's one reason manufacturers get away with their sloppy assembly and poor QC. When people who haven't properly tested for it and have no clue their lens has the problem see it in an image and they pass it off as "I must have missed focus" or some other reason.  In some cases it's a tilted element and not decentering and with a tilted element a soft side or part of the frame may still be able to be sharp but just not at the same focus distance. With tilted elements the entire focus plane may be tilted left, right, up or down.

The other issue is field curvature and all lenses have it to a greater or lesser degree. Field curvature makes it very difficult to compare one lens like the 10-24 to another like the 16-55. In reality field curvature makes your comparison invalid as you don't know the characteristics for each lens.

Focus accuracy is another factor and I never trust manual focus for lens testing. If on my Canon bodies I use liveview with contrast AF and take several series defocusing the lens each time. I'll defocus, focus, turn AF off and then shoot wide open and then at each full stop to F11 and then do it over again 3 to 5 times. Camera needs to be on a tripod with mirror up if DSLR and use either a remote release or the 2 second timer.

I've developed a test setup using an Edmunds Optics USAF1951 resolution chart lit by 4 500 watt halogens. I've also used the chart outside in pure Sunlight which works even better. I use a mirror in the exact center of the target to align the camera perfectly in the horizontal and vertical planes (this is not for those without attention to detail). The chart has to be perfectly aligned in the viewfinder for the results to be accurate.

Here's what the chart and alignment looks like when I was testing my 100-400 outside. The chart is on the side of a large electrical transformer box and held in place with magnets and I'm at 67 feet from the chart at 400mm. You have to move and do the realignment for each FL you want to test (yes, it's tedious and I got a good Sunburn that day).

The frame below is from testing the 56f1.2. Click on the original and compare the lower left quadrant to the others and you'll see this lens is far from uniform at f2.

Below is what I consider a mild level of decentering with the 100-400. This is a comparison of the bottom two corners if memory serves me.

And here's an example of a decentered Sony lens compared to the Fuji 10-24. Notice that even at f5.6 the one side of the frame on the Sony's blur plot is considerably worse than the other. The Fuji looks really good here.

As I said earlier testing for this is tedious and requires discipline and excellent technique. There are far too many things that can influence the results and give either a false positive or negative result. For that reason and after doing this for 30 years I highly suggest doing the camera flip method I outlined earlier. It's a quick, simple method to determine if you've got a clunker and it eliminates many of the variables.

Good luck.

Bob