At 1:44 PM +0000 3/14/03, olympus-digest wrote:
>Date: Thu, 13 Mar 2003 15:53:54 -0600
>From: "Bill Pearce" <bspearce@xxxxxxxxxxxxx>
>Subject: [OM] A conspiracy?
>
>"Even if digital is close to or surpasses film you would think there would
>be some concern about the glass. When did that go away? Or do I just not get
>it? :-)"
>
>An interesting point is that, several years ago, about the same time we were
>being "educated" about the way light exits a lens, and what is needed for
>digital, someone (camera manufacturer, chip manufacturer, or something like
>that) was presenting the theory that if a lens is "too sharp" it makes for
>bad digital photos. I remember this theory was received with a surprisingly
>little raising of eyebrows.
>
>That was then, in the 4-6 megapixel age, and now were on the cusp of 14 (if
>Kodak can fill orders), so I don't know if the theory still holds.
The theory does still hold. Nor was it nonsense back then.
The issue is that if the resolution of the lens much exceeds the spacing
between pixels (of the same color), it will allow moire beats (seen as color
fringes) to happen whenever the pitch of some pattern in the subject happens to
be more or less the same as and aligned with the pixel pattern of the CCD.
In mathematical terms, what's happening is that the CCD "undersamples" the
pattern in the subject, causing "aliasing" (those beats), and the only solution
is to "low-pass filter" (blur) the image hitting the CCD chip. The necessary
blurring can be done in a number of ways.
One way is by not using a lens that is too good for the CCD. This is likely
the standard way used in cameras with non-removable lenses, and has the
considerable advantage of making the lens less expensive to make.
Another way is by coating the CCD chip surface with some kind of anti-aliasing
(diffusing) layer that causes the sharp image formed on the surface to bleed
slightly. This is reportedly that standard way to solve the problem with
lenses intended for film.
I would assume that some cameras use both blurring methods, carefully balanced.
One wants just enough blurring to make the fringes insignificant, which is
ultimately a question of physiological optics. The human eye is fairly
sensitive to color fringes and ripples, though I don't know how much so in
numerical terms. (I'll look.)
As CCDs grow in resolution (megapixels), less and less blurring will be
required, so the effect will eventually just go away. That said, the optical
system (lens and optional anti-aliasing filter) will always be designed to have
less resolution than the same-color pixel spacing, to supress the fringes.
It's just that as CCDs get better, this will be less and less of a limitation.
Film, with its random grain and granularity patters, is not vulnerable to morie
beats. The regular pattern of pixels in the CCD is required to generate the
effect.
Joe Gwinn
< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >
|