Lens Quality (?)

[BJL]

If I understand you correctly, the IS lens performs fine with real subject matter but not quite as well as the older lenses when photographing high contrast USAF test patterns.

If you photographic tastes involve a significant amount of very finely detailed high contrast monochrome subjects like the test patterns or text at a great distance, I would be worried; otherwise the success with real subjects should be what counts, and you have probably just seen evidence of why high constrast test patterns can be somewhat unreliable in evaluating photographic tools. (The test patterns probably makes sense for the USAF, which likely has it reasons for wanting to photograph text from a great distance!)

I believe that there are also low contrast test patterns: 1.6 to 1. These might be harder to work with, but they probably measure something of more relevance to image sharpness with normal subjects.

[DaveT]

I recently purchased a copy of the Canon 20-200 f2.8L IS.  I compared this informally at infinity using a variable contrast scene with the non-IS version, as well as with the older 80-200 f2.8. These images were recorded with using a Canon 1Ds body set at f-8 on a tripod.  All the shots had similar contrast and resolution at the edges and center both at 100 and 200mm.  On the high contrast USAF test targets it was a different story. Here, at f 5.6 and a 10:1 reprodution ratio, The IS was about 1/2 to 1 group behind the others in the center, and about 2 groups off at the edge.  The edge contrast was also markedly lower that with the older two lenses that were both very even across the field. So my question is, should I return this lens, or is what I'm seeing the result of field curvature that was sacrificed in the lens design for the advantage of IS?