What test setup and chart do you use to test for decentering?
Also, my issue is how to determine if a corner focus is bad enough to warrant returning the lens. Maybe a more quantitive chart would help?
I have tried a lot of different methods to test for decentering, all of which work to some degree. I have settled on a method that does not use a chart. Instead, I focus on a distant object (more on this later), and then take a series of images. For each shot I recompose--
without changing the focus--so as to place the same object in each of the corners. I then compare the
relative sharpness of the
same object in all four corners. With one form of decentering the subject will be much sharper in one corner than in the opposing corner. With another form of decentering there is what looks like a parallel tilt, where the left-side corners are much sharper than the right-side corners (or vice-versa), or where the top corners are much sharper than the bottom corners. With a well-centered lens all the corners are equally sharp, or very nearly so.
A few notes on this method:
- I choose a subject that has good contrast and small details so that differences are easy to see. The size of the chosen object depends on the focal length of the lens.
- If the lens has a lot of field curvature then choice of the initial fixed focus point will need to be near the edge (or corner) of the frame, or else none of the corners will be in focus.
- If the lens has no sharp corners at any f-stop then you need to change the comparison zones closer to the center.
- It is
critically important that the focus does not change between shots.
- It is important to use a tripod (and perhaps pre-fire, depending on the shutter speed). It's best to do this on a calm day.
- If I find a problem with a lens I
repeat the test at least once, to make sure I am not seeing an artifact from a botched test. This is important.
- With zoom lenses you need to test at multiple focus lengths, since the lens might be well centered but not over the entire range.
- It is also useful to test at different f-stops. If buying a lens for landscape use you might be satisfied if the lens shows adequate corner consistency from, say, f/4.
With a really well-centered lens it is stunning how similar the corners look. That's always exciting to see! On the other hand, it can be quite stomach-turning to see how far off a lens can be. Of course we can't expect perfection. So how bad is too bad? That's subjective. After testing hundreds of lenses I have a sense of what I can reasonably expect to see. I feel that my standards are not too high. Retailers never give me a hard time when returning a lens for decentering (they know it's an issue), and even private sellers are typically very cooperative (and even apologetic) when I show them the photographic evidence.
While I haven't become less discerning over time I have become somewhat more understanding of the difficulties faced by manufacturers. It's hard to get this right. I have this learned as a result of trying to re-center old lenses that I adapt. The use of adapters can throw off a well centered lens. In that case I modify the adapter with thin shims to bring the lens-adapter set back into a centered state. These shims are SUPER thin. For my most commonly used shim it takes about 4-5 shims to equal the thickness of a piece of standard cellophane tape. One or two shims of this thickness can throw off the lens, or correct it. The tolerances are mind blowing. On the other hand, some manufacturers do a much better job than others at meeting this challenge. As Paul notes, Sigma and Leica are particularly praiseworthy. I have been extremely impressed with the modern Sigma lenses I have purchased, to the point where I am surprised when I see an issue. Obviously they are doing something right, both in the manufacturing process and with their post-assembly testing. It almost seems like they care.
www.lenstip.com
