[snapscan] High resolution scanning

From: Oliver Schwartz (Oliver.Schwartz@gmx.de)
Date: Tue Apr 10 2001 - 14:25:54 PDT

  • Next message: Karsten Festag: "Re: [dev] New and updated backends"

    Hi,

    is there anybody out there who's successfully using the snapscan backend with
    resolutions of 1200 dpi and above? It doesn't work for me, and if I'm not
    mistaken for no one else either.

    I looked into this and found that

    1) x-resolution can only be increased up to the optical resolution. It's
    possible to request higher resolution from the scanner, but it will only
    return the image data with maximum resolution padded with junk. On my
    scanner, which has an optical resolution of 600dpi, this means that for a
    scan in 1200 dpi I get the scanned data on the left half of the image and
    junk in the right half of the image. However, it's possible that there are
    some scanners out there that do an interpolation in their firmware - anybody
    seen this?

    2) Performing a scan on a wide scan area with high resolution will lead to a
    hanging backend, because the size of one image line becomes larger than twice
    the size of the SCSI buffer. Some computations inside the backend calculate
    bufferSize/lineSize, which will evaluate to zero - the rest is left to your
    imagination.

    The best solution is probably to use different resolutions in x and y
    dimension. The x resolution may than be increased up to the optical
    resolution, which avoids problem (2). If a frontend requests a higher
    resolution the backend can use some decent interpolation algorithm.

    This will probably take some time to implement. Until then, I suggest to
    disable the higher resolutions to avoid problems. If however there are
    scanners with firmware interpolation those resolution settings can be left
    untouched.

    Oliver

    --
    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 : Tue Apr 10 2001 - 15:03:26 PDT