on 3/12/00 2:08 am, bjs at [EMAIL PROTECTED] wrote:

>>> The most fundamental reason that scanners lose sharpness is because they are
>>> area samplers rather than point samplers.  This is a physical necessity due
>>> to the finite size of each CCD cell.   The resulting area integration of
>>> each sample forms a physical low pass filter which softens the image.
>>> 
>>> Aliasing isn't relevant to this question.
>>> 
>> But, er, what you just described *is* aliasing.

Hate to contradict the listmom but... hell, no, it ain't.

This is my understanding of aliasing. Others will jump all over me, and
please do, but I think it shows why the above is wrong.

Aliasing occurs when a signal is sampled at a frequency too low to capture
the frequency being sampled. Done like this it appears (aliases) to be a
signal of a lower frequency (I seem to remember that the alias frequency is
the *difference* between the signal and the sampling frequency). Its effect
is likely to be *strongest* with point sampling.

The most obvious example of this is a cart wheel in a western which appears
to go backwards... you can think of this as a high frequency signal (the
rate at which the spokes pass, say, 12 o'clock) which is being sampled at a
low frequency (24 fps). Because the sampling frequency is so low the
frequency (rotation speed) of the wheel is 'aliased' to another frequency...
so that it appears to go slowly forwards, backwards, stop etc.

This is a special case I know and technically there are a few things wrong
with the example, but at least the effect is clear and familiar.

If you shot the wheel with a high speed camera (say 240 fps) you would
capture the true motion of the wheel.

A moire pattern is (I think) a visual form of aliasing. This occurs
typically when a high frequency visual signal (say a dot screen in
newsprint) is scanned at too low a resolution. As a result, rather than
showing the true frequency you get banding at another frequency which
represents the interaction between the scanning resolution and the
resolution of the dot screen.

-- 
Johnny Deadman

http://www.pinkheadedbug.com


Reply via email to