Without AA-filters, moire can appear and has to be removed manually in the RAW-converter. But WHEN it appears, no crucial information is lost during this process that would have been captured WITH an AA-filter.
Well, I really don't want to hijack this thread into being about AA filters, but I ain't buyin'
that one!
The reason for my skepticism is that at work I use a medium-format back that lacks an AA filter. When you get moires -- which you inevitably do when you photograph fabrics or printed materials with halftone dot screens -- you're supposed to remove them via the manufacturer's software, which includes about two dozen different AA filter routines.
None of them work as well as a DSLR with a properly-designed hardware AA filter. Crucial information is
always lost (which is one reason I only use this back when I absolutely have to.)
I wish I'd saved a technical whitepaper I downloaded a few years ago, when the M8 first appeared and this less-is-better notion about AA filters began circulating widely among photographers. It was put out by a maker of AA filters (obviously they'd have an incentive to lie, of course, but their argument was very logical and the technical details were plausible.)
Among the points made were that an AA filter isn't simply a "blurring" filter -- it uses birefringent crystals to cut off too-high spatial frequencies with negligible effect on lower ones, preventing aliasing from occuring in the first place. The best that software solutions can do is blur the aliasing
after it occurs -- and since the software has no way of knowing which structures are aliasing and which are actually present in the image, inevitably it has to compromise detail. Medical analogy: the AA filter is preventive medicine that keeps the aliasing "disease" from occurring, while software is an after-the-fact remedy that can only mask the symptoms. That's been my practical experience with my non-AA back, too.
Another point is that one form of aliasing is "false resolution," which causes an image to appear to have fine details that aren't actually present in the original subject. It's been speculated that this is actually what photographers are seeing when they say their camera produces "sharper, crisper" images after the AA filter has been removed.
Counterpoint: After my original post, I got a private reply from another member who wishes to remain in stealth mode, suggesting that while AA filters do a better job than software of handling moires, they do so at the expense of color purity -- they muddy up fine nuances of color.
That I can believe. I can imagine that people who say their non-AA cameras give "more detail" might well be seeing
color detail rather than
luminance detail. And that would explain why people for whom finely nuanced color is very important -- landscapists and catalog shooters come to mind -- would prefer minimal AA filtering. (Natural landscapes probably contain few potential sources of moire, and catalogs are shot under controlled conditions where you can manage the problem at the source.)
That's still not the same as saying "less is better, and none would be best of all." It's a question of which image characteristics are top priority for you. But I'll concede that for some people, ditching the AA filter might be the best answer, even if that's not for quite the reason they think it is.
And now we'd better be getting back to arguing about the desirability of various potential feature sets for the M9 vaporcam...
