Hi Jim,
Testing at 2.000 x f takes away any effect of the difference in distance as you showed. In my view, you don’t even need 2.000 x f, even if you use your way of testing. You can make the CoC larger since no lens/sensor combination is able to see details smaller than about 1.2 x the pixel pitch (in terms of lp/mm or cycles/mm it’s even 2.4 x the pixel pitch). And then you might argue that at f/1.4 or larger apertures there will be so many uncorrected aberrations in the corners that the blur due to the difference in distances in the center and in the corners is neglectable. But ok, I understand that you want to be on the safe side. Enlarging the object distance to approx. 2.000 x f gives us i.m.o. at least three advantages:
- you test to see how well a lens does at large distances, in other words you disregard the fact that a lens might not be ideally corrected for objects at closer distances
- skewing the camera slightly makes it possible to position the target everywhere in the image while keeping almost the same distance, hence avoiding the need to refocus and the need for several targets (DOF will compensate for the relatively small difference in distance between corners and center).
- you avoid problems with regards to the parallelism of the camera sensor and the targets.
It also has disadvantages.
- It’s not very practical way of testing, unless you have a very large back yard and the weather is good (that was Leonard's valid point).
- It doesn’t say much about how well a lens performs at moderate and small distances.
Your use of Siemens-stars at 2.000 x f seems a valuable extra method to test lenses. It’s way better than the long distance shots DPR and others make, because then it’s very hard to compare meridional and sagittal structures – while it is very easy with a Siemens star. But if you really want to test more than whether a lens is not decentered or tilted or similar, you should make tests at other distances too.
From experience I know there can be large differences, e.g. with the three Nikon 50-ish standard lenses (50/1.4 G, 50/1.8G and 58/1.4G). All three display the same pattern: at 50 x f and less the corners are bad and don’t even get really good until you stop down a lot, at 100 x f the corners aren’t great until you stop down till at least f/4, and finally at 300 x f it’s a lot better and at infinity slightly better than that. The performance at around 100 x f at large apertures can be important e.g. for group pictures with many lenses or for theater photography. On the other hand I’ve tested a 85mm at 18 x f and compared that to the Nikkor 85mm f/1.4G. The lens that must not be named gave better test results at that distance than the Nikkor, even in the center. The reason I tested it at that distance though, was that the lens that must not be named turned out to emphasize almost invisible skin blemishes on a model with a nice skin in such a way, that it was impossible to remove them by lowering clarity and sharpening without getting strange looking results. So each portrait I made with that lens had to be retouched. The Nikkor 85mm f/1.4G on the other hand, works very well wide open and at f/2 and makes portraits you can use without the need of retouching with most subjects. At 100 x f there were small differences in sharpness at large apertures between the lenses and at infinity there were no differences! I also tested a Fuji lens, that was great at 101 x f, but not so great in the corners at infinity (probably because curvature of field compensates point 2 below partly but shows its ugly head at infinity). If you test a macro lens in this way, you are even drawing conclusions based on the least important use case for such a lens. All of that is to show that testing is not an easy feat, you have to define different use cases and different criteria for different lens types, hence different distances. As a rule, I test at 101 x f but also at larger distances and sometimes at smaller distances.
The problems with less sharp corners when making pictures at certain distances, have at least two different causes:
- Field curvature, one of the Seidel aberrations that lens designers have been trying to correct since ages.
- The effect that Jim described, that especially in wide angle and to an extreme part on fisheye lenses, the objects at the borders and in the corners are larger or even infinitely larger away from the camera than the object in the center of the image.
In all modern lens designs, there has been made some kind of effort to correct both phenomena in such a way that both at moderate distances and at infinity you get reasonably sharp corners. You can argue though, that with a 24 mm f/1.4 sharp corners at one feet at f/1.4, is something that almost no photographer will need. Correcting a lens for that situation, on the other hand, will likely go at the expense of other corrections or at least make the lens more expensive.
So most lenses have floating elements or moving groups that partly compensate both phenomena I mentioned and sometimes even field curvature is used to partly compensate the second cause. But such a correction always has some kind of disadvantage in a certain situation. What I’ve seen with the 14-30/4 and also with many other lenses, is that if you make a picture focused at infinity, parts of the scene that are closer and fall outside the DOF can vary in sharpness between the corners and the borders e.g.. Most of the time it’s no problem, but if it is, you can make a focus stack, and with the Nikon Z cameras you can do that with the blink of an eye.
Apart from that, I think the corners of the 14-30/4 @ 14 mm are ok, but the 14-24/2.8 is better at 14 mm. The 14-24mm f/2.8G is best at the short range, the 14-30/4 is best at the long range. Most users might not see the difference though and it’s relatively small and light. I’m doing some extra testing with the 14-30 right now. But maybe those who want the very best should wait for the Nikkor Z 14-24mm f/2.8 S. If it's as good as the Nikkor Z 24-70mm f/2.8 S, it will blow both the 14-24mm f/2.8 G and the Z 14-30mm f/4 out of the water.
