Flatbed Scanners

Discussion in 'Scanners' started by JSF, Aug 29, 2006.

  1. My experience with the regular holder with the 8000 is that if the film
    curls, it's way larger than +/- 0.5 mm and there's no way to keep it flat.
    If the film lies flat, maybe. But if one really wants corner to corner sharp
    scans, one has to go to fairly extreme lengths to keep the film truly flat.
    Mine's 2GB at 3 GHz. Only slightly better. If work keeps up at its current
    pace, I'll be able to buy something larger next spring. But won't be taking
    many photographs.

    Is that file size calculation correct? MP = 2.2 x 6400 x 2.7 x 6400 = 244.

    Yep. It is. (Doh, I know you are good with numbers<g>.)

    The problem with that, though, is that most photo editors require twice the
    RAM to do any transformation without swapping, so anything over .8 GB or so
    is going to be painful.
    Your experience with flatbed scanners has been more positive than mine or
    Rafes. I remain unconvinced that they provide photographically significant
    information above 1/2 the advertised ppi.

    (Also, from looking at the examples on Rafe's page and other stuff on the
    net, I remain unconvinced that the drum scanners are more than marginally
    better than the 4000 ppi Nikons, but that's another rant.)

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Aug 30, 2006
    #21
    1. Advertisements


  2. I came *this* close to acquiring a Screen DTS-1030 drum scanner
    a couple of weeks ago. Someone very nearby was selling it.
    The price was reasonable -- but it would have cut into my
    savings for a FF DSLR, and I do have my priorities.

    The 4990 is another one of those cases where I think I'm getting
    maybe most of "what's possible" (relative to drum scanner)
    but with a small fraction of the effort.

    If you think flatbeds and the LS-8000 are finicky for focus,
    let me introduce you to the drum scanner. I believe the
    wet-mounting schtick may really be necessary to get the
    last bit of resolution out of it.

    You know, given the incredible quality of stitched images I'm
    seeing, maybe Creo-Scitex had the best idea of all -- a hi-res
    flatbed that works by stitching adjacent "swaths" of the
    image.

    Just "aim" the CCD at 1/N of the width and thereby
    increase resolution by the same factor. Needs excellent
    optics and mechanics, but no new technology.



    rafe b
    www.terrapinphoto.com
     
    Raphael Bustin, Aug 30, 2006
    #22
    1. Advertisements

  3. Huh? I thought you had to wet mount to a drum scanner. Learn something new
    every day...
    I've largely not been paying attention to the high end. The pro labs here
    are insanely expensive, so I can't even try it for occasional scans.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Aug 30, 2006
    #23
  4. "Roger N. Clark (change username to rnclark)" <>
    wrote in message SNIP
    Yet it shouldn't be too difficult with a film frame with a sharp
    slanted edge.

    Assuming optimal focus, scanned at two perpendicular angles, it would
    allow to produce a close approximation of the PSF. That PSF can be
    used to either make a model specific High-pass filter for a fast
    restoration of resolution, or for a time consuming deconvolution
    method in frequency space.

    With the relatively large support PSFs I expect from most flatbed film
    scanners, it should not 'enhance' graininess too much, so a
    substantial improvement could probably be achieved. Such a PSF also
    provides insight in a quantifyable amount of 'empty' enlargement, or
    down-sampling without significant loss of resolution. It also would
    take a lot of speculation out of the resolution debates.
     
    Bart van der Wolf, Aug 30, 2006
    #24
  5. One can simply tape the film to the drum. Securely, of course.

    The problem is that the film will bow due to centrifugal force
    so focus won't be uniform. I discovered that after a few
    weeks with the Scanmate.

    Unfortunately, the ScanMate crapped out just as I was
    getting started with Kami and mylar sheets.


    I follow it out of curiosity, and if a cheap viable drum
    scanner appears in my neighborhood again, I may
    grab it. Ditto (maybe) for a Scitex hi-end machine.

    It won't make a huge visible improvement in the prints
    I can make on my 1800 or 7600. I've already got more
    pixels than I need or can use (eg. scanning 4x5" at
    2400 dpi, printing at 24x30".) That's a mere 6x,
    which may even fall into grain-sniffing territory.

    A lot of work though, taking that process from beginning
    to end.


    rafe b
    www.terrapinphoto.com
     
    Raphael Bustin, Aug 31, 2006
    #25
  6. There must be something more precise about a spinning drum than a driven
    scanner head, if that's what it's called, no? I mean, you'd think that if a
    spinning drum can be that precise, a slower moving head, that moves on one
    axis would be more precise.
     
    AKA Gray Asphalt, Aug 31, 2006
    #26
  7. I guess that prompts a question about why a flatbed can't have higher
    resolution.
     
    AKA Gray Asphalt, Aug 31, 2006
    #27
  8. "Roger N. Clark (change username to rnclark)" <>
    wrote in message SNIP
    They tend to pass very little light, and I could imaging diffraction
    being an issue just like in pinhole cameras.

    The slanted edge method, as described in the ISO 16067 standards for
    determination of scanner resolution, allows to sub-sample each sensel
    position approx.10 times. The ISO then only collects the edge
    responses at quarter pixel resolution, more than enough for an MTF
    upto double the Nyquist frequency (also allows to judge the aliasing
    response).
    True, but that would possibly tell more about positioning error than
    sub-sampling, and would require a lot of handling for a statistically
    solid solution. That is the nifty part about the slanted edge method.
    By scanning an edge that is slanted by approx. 5 degrees along the
    horizontal/vertical sampling direction, you can accurately sub-sample
    each sensel at 10 different positions. By scanning a stretch of a
    couple of hundred pixels the statistics are very stable.
    I use a razor blade in a slide mount, much cheaper and reasonably
    simple to make ... I know it's not perfect, but it works good enough,
    the procedure and results are very repeatable.

    Evaluation is very simple with a tool like Imatest that I use:
    <http://www.imatest.com/docs/tour_sfr.html>
     
    Bart van der Wolf, Aug 31, 2006
    #28

  9. Creo-Scitex makes flatbeds in the $15 thousand range.

    Precision mechanics and optics are still important -- and
    expensive.

    But basically a drum scanner uses microscope optics
    and images one point at a time. You can see the guts
    of one here:

    www.terrapinphoto.com/drumscansaga


    11,000 dpi optical -- that's pretty much beyond what
    any real film image holds, by a factor of two, and
    some would say by a factor of 3 or 4.


    rafe b
    www.terrapinphoto.com
     
    Raphael Bustin, Aug 31, 2006
    #29
  10. But it kind of doesn't answer a question about why the same type of
    electronics can't be applied, in a cost effective manner, to a moving scan
    head instead of a moving drum. Maybe I don't get it, though. : -)
     
    AKA Gray Asphalt, Sep 3, 2006
    #30
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.