Craig W. Shier <[EMAIL PROTECTED]> wrote:
> physically implementable scanner. If the sample areas
> are sufficiently samll. i.e. if they do not overlap,
> there will be no reduction in sharpness for a
> sufficiently high resolution scan. For example, if
> your lens resolves 50 lppm, a 2540 dpi scan is
> theoretically good enough to reconstruct the original
> image. (Realistically though 2 or three times that
> would improve the reconstruction.)
Craig, I like your explanation, but it doesn't take into account the
interference between the grain (or dye cloud patterns) in the film and the
scan resolution, which seems to be the main source of pain in film scanning.
When the grain is taken into account, it's the relationship between the
grain (or dye clouds, whatever) and the pixel size which is the most
important factor, not the resolution of the camera's lens system. Tony
could probably point out other limiting factors, but in my (albeit limited)
experience in film scanning, the main problem is interference between the
frequency of samples and the random variations in dye clouds that are the
most significant - and it's certainly aliasing as I understand the term. If
the patterns were regular you would get moire, but the patterns are random.
Rob
