Common wisdom is that scanning at the highest optical resolution and
then downsampling via a good program provides better results.
But, I would suggest the obvious, which is try both. There are a few
reasons why I suggest testing (and they are not to be snide). Different
scanners deal with lower than optical res scans differently. Some
always use software to downsample rather than just select spaced sensors
on the CCD, some only use actual sensors data with certain resolutions
(see below for more on this). Further, the type of image alters the
result. Line drawings or images with high contrast edges and fine detail
usually do better to be scanned at native optical resolution and
downsampled in something like Photoshop. But even that needs to be
qualified. As nice as bicubic downsampling is (and it is for most color
images), for some images "closest neighbor" creates a more accurate
result.
Some images might look better with the scan directly from the scanner
(especially if you scan at a resolution which is an even divisor of the
scanner's native (optical) resolution.) Of course, this depends upon
the scanner firmware/software but a 4000 dpi optical scanner should be
able to produce a 500 dpi scan without using any interpolation, simply
by reading every 8th sensor on the CCD. Asking for a 450 or 550 dpi
scan might force the scanner to do an interpolative scan which might not
be as good as a downsampled scan in photoshop.
In your case, I honestly believe doing the empirical tests will provide
you with a much better answer than I can.
Art
David wrote:
>
> If you want to put a picture in your web with 500 pixels wide,
> what is better? Scan at full 4000 dpi and resample to adjust
> your image at this size (letting PS discard pixels) or scan
> at this size. I suppose the first choice but I am not very
> secure. ;-(
>
> Best regards,
>
> Dave
>
> ------------------------------------------------------------
> Unofficial Olympus web page
>
> Unofficial Olympus Gallery
>
> ------------------------------------------------------------
