RE: filmscanners: ReSize, ReSample or ReScan ?

["Shough, Dean" <dean.shough@lmco.com>]

> Don't buy into this "magic resolution" claim.  Just because a particular
> pattern (that has nothing to do with the variety of real images you will
> be
> printing) prints "better" at a particular DPI does not mean there is a
> universal "magic resolution" for every/other image(s).
> 
> The image that was "posted" only has black and white lines.  Again, hardly
> indicative of a normal photographic image.  Personally, I believe this is
> a
> very flawed test, and any conclusions drawn from it are only valid for
> THAT
> very image, and are erroneous for any real photographic image.
>

I believe that in practice you are correct.  I created a similar test file
using a real image (a small piece of the 47 MB  PhotoDisc TIFF test file
from ftp://ftp.photodisc.com/Tech/Target/ ) and detected very little
difference between the various images.  It was only when the image had
regular, repeating patterns in it that it became obvious that the print
driver was doing some sort of simple sampling.  I am not surprised that the
effect is not visible with real images as most real images do not have much
content at the higher frequencies.  


The line test I posted shows the sampling effect most clearly.  A zone plate
(grayscale, concentric circles with various frequencies) test shows the
effects of sampling, but it is not nearly as obvious.  The real image only
showed the effects where the original image contains high contrast, high
spatial frequency, repeating patterns.  Even then it took very close
examination to detect.

I expect the effect will appear completely different  in any printer that
does not use the regular halftoning used in the printers I tested.  I am
curious how it appears in the Epsons with their error diffusion halftoning.