Re: [snapscan] Core testing group

From: Steve Underwood (steveu@coppice.org)
Date: Wed Nov 01 2000 - 17:07:47 PST

  • Next message: Timo Hoderlein: "Re: Agfa Snapscan 310 color problem"

    "William L. Sebok" wrote:

    > Steve Underwood <steveu@coppice.org> said
    > > There seems to be a bit war in the scanner industry right now. We are
    > > starting to see some outrageous numbers of bits quote. The best real values
    > > are about 14 per color, or 42 overall. Anything more is fiction. In practice,
    > > most scanners only send 8 bits per colour to the host. The rest are lost in
    > > gamma correction witin the scanner.
    >
    > However, having enough bits to get a decent 8 bits after gamma correction
    > is important, especially when trying to scan dark images or slides.

    Yes, I should not have said "lost in gamma correction". "Consumed in gamma
    correction" is more accurate. However, the noise floor of the CCD scanner makes
    extreme numbers of bits bogus. The have a 12 or 14 bit A/D, then interpolate up to
    16 bits, then gamma correct down to 8. Its really just a bogus set of calculations
    to be able to justify some high number of bits on the box. Scanners have for a
    long time offered ridiculous levels of interpolate pseudo-high resolution, because
    its cheap to do. Its of absolutely no use, though. High resolutions sound good,
    until you find a scan with take a gigabyte or so!

    Regards,
    Steve

    --
    Source code, list archive, and docs: http://www.mostang.com/sane/
    To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
    



    This archive was generated by hypermail 2b29 : Wed Nov 01 2000 - 17:33:01 PST