For your convenience, these curves are simple scans of pages from the Canon book: EF Lens Work III - The Eyes of EOS. The book contains much useful information and I would urge any EOS camera owner to purchase one. It really helps enormously in your lens-purchasing decisions. Personally, after much lens-chart testing on a Canon EOS D30 and 1Ds, I have come to the conclusion that any Canon lens (with or without tele-extender) that gives >70% contrast at 30 lp/mm on these charts will fully satisfy my needs. It is very difficult to find a lens with <100mm focal length that meets such a criterion well off-axis and near wide open. The long Canon lenses that meet this and that I own are indeed capable of giving pixel-sharp stars, as to be expected from the charts, my 600mm even with the addition of either the 1.4x or 2x tele-extender. I own the 100mm/2.8macro USM (not usable with Canon tele-extenders), the 200mm/2.8L II, the 400mm/5.6L and the 600mm/4.0L IS. The 200mm/2.8L II is usable on starfields wide open even with the 1.4x extender. Of course, stars will sharpen up perceptibly with closing down the aperture slightly, but we are also eternally short of photons... I do not find the 200mm/2.8L II as convincing when used with the 2x. Wide open, this combo is also obviously less sharp than the 400mm/5.6L, even in daylight testing. The shorter Canon lenses I own(ed), 28/2.8, 28/1.8, 50/1.8, 50/1.4, 100mm/2.8macro nonUSM, do not deliver pixel-sharp starfields, at f2.8, as to be expected from the MTF data. The 50mm lenses give satisfactory starfields at f4 and the 28mm ones at f5.6. For good starfields on full 35mm format, targetted at large prints, you have to close them down another f-stop. Using a 1.6x crop factor DSLR such as the 400D or the 30D opens up the field since you need to check the MTF curves only as far out as 13mm off axis. For a full 35mm format camera like the 1DsII you need to check out the MTF all the way to 22mm off-axis. Most Canon lenses show >70% contrast centrally at f8 at 30 lp/mm, according to the MTF data. So do be careful with anecdotal comments regarding some lens being termed super-sharp. Do those comments apply to f8, was the subject placed centrally? While I am confident that most people will be satisfied by a 70% cutoff, others may be satisfied with a bit lower, say, 60%, especially for non-astro. It all depends on their own yardsticks. Let us just note that whether one uses 70%, 65%, or 60% as a cut-off is often tempered with what is actually available. This becomes even more an issue if you wish to include zoom lenses and/or the 1.4x and 2x tele-extenders. Study the MTF data. They can tell you a lot, reliably. By mental interpolation you can also hazard a guess as to how far you may have to stop down. If you have to stop down to f8 then, frankly, that lens is quite useless for astrophotography, but it can still deliver great images in landscape work. If you are unfamiliar with the MTF and how it relates to imaging then do read this reference:
and then come back here to check up the curves for your prospective purchases.
Added in 2007: It has finally dawned on Canon that it would be a good idea to make their book EF Lenswork III available for download online. It is available here:
For other stuff peruse my primary website: http://www.samirkharusi.net/