Re: Which scanners REALLY provide 36 bit output? HP?

From: Steve Underwood (steveu@coppice.org)
Date: Fri Dec 15 2000 - 19:36:36 PST

  • Next message: Ben Millhouse: "RE: [snapscan] Vuego 620S and ACARD scsi card"

    Bob Washburne wrote:

    > > CDs can fail in as little as a year
    > > (any make) despite the claims of 25 years for a pressed CD (which isn't that
    > > impressive anyway) and 70-100 years for CDR (which seems completely bogus). I
    > > live in a sauna, and things tend to fail fast, but even in dryer climates you
    > > can never rely of any digital medium. Some years ago in the UK we stored
    > > precious reels of data tape in a fully controlled environment to the makers
    > > (Ampex) spec. In just two years the layers of tape had coalesced so well we
    > > could saw through the block of tape like it was a block of wood. Of course such
    > > extreme failure doesn't always happen. However, since you cannot tell when it
    > > will, you cannot rely of these media. Close monitoring and recopying seems a
    > > useless precaution, when degradation can occur so fast.
    >
    > So far I have not experianced the problems you have. Yes, back in the
    > early 80's I had a large batch of nine-track tapes which went bad
    > (Graham Magnetics). We switched to Scotch Black Watch and never had a
    > problem since. I have yet to have a pressed CD go bad on me and the
    > only CD-R to fail has been the dirt-cheap generic.

    The tape I referred to was Ampex. At the time it was considered the best you can
    get. My experience with back coated tapes is they suffer less this kind of sticking,
    but have other problem which don't make them much better overall. Interestingly, I
    have also found Scotch tapes last better than others - though those silly ads for
    video tape that last forever seem to be inviting trouble. As a new tape Scotch gives
    one of the nastiest performances, but they do score well on lengevity. They still
    rarely last more than 10-15 years. Many early videoed TV programs are still
    available in a watchable form, but then many are not. Tape can last, but don't rely
    on it. I live in one of the world's most humid climates (HK). Here, we buy special
    video tapes and floppy disks with fungicide embedded in the tape coating. Without
    that a stored video tape fails in no time, though they can last well if used every
    day.

    I have a number of commercially pressed CDs, some less than two years old, which no
    drive I know of will read. They were fine the day I bought them. I have CDRs from
    Kodak, Mitsubishi and other top names that failed within months. Again, high
    humidity has no doubt aggrevated this, but I doubt it would take too long elsewhere.
    In the UK, which is a fairly dry climate for most of the year, I heard of quite a
    large number of failures in less than 10 years.

    > There are other philosophical problems to using digital media for
    > long-term archives. Accessability. My 8" floppies may still be
    > perfectly readable, but can *YOU* access them? You need:
    > 1) Hardware to access the media.
    > 2) Software to read the media
    > 3) Software to interpret the data (JPEG, TIFF, etc.)
    > None of which is guarenteed from decade to decade. And none of which is
    > apparent from eye-balling the media (can you determine what the format
    > of the data is on an unknown media?)

    I can probably find both a working 8" drive and a computer to use it with, if I look
    in my mother's loft. There are probably a number of collectors who could do that
    same. I doubt you have a disk that is still playable, though.

    > But there are great advantages as well:
    > 1) Ease of copying.
    > 2) No loss of signal from copy to copy. Once a pixel is digitised into
    > a "5" that 5 remains a 5 no matter which generation copy it is. And a
    > weak five will copy to a strong five.

    This is a weak argument in a world without perfect error correction. I have _never_
    seen this work out. I can make a thousand perfect copies of a disk tomorrow, but the
    same thing doesn't work out over time. There will always be uncorrectable errors
    over time, and often the media fails totally.

    > [...]
    >
    > 6) Reprint. As the current media, e.g. CD, becomes obsolete and a new
    > media becomes mainline, e.g. DVD-R, copy the data onto the new media and
    > re-publish. This is where the ECC (Error Correcting Code) format
    > becomes important.

    ECC does nothing for a corroded, mouldy or seriously warped medium. Cohesion,
    adhesion, and adhesive creep don't even need warm moist conditions to screw things
    up 100%.

    > 7) Find a child or grandchild interested enough to carry on after me.
    > Otherwise, the published versions in the libraries will have to suffice.

    This caring child will be looking sadly at a totally failed medium. If you keep the
    fragments of crumbing paper, those might still be readable.

    > At this point I will probably just keep my HP5370C unless someone
    > responds VERY quickly that there is a much better choice out there. It
    > has the 1200dpi, no problem. It has 14 bit A/D's and while the included
    > software only returns a gamma corrected 8 bits, tech support swears that
    > the hardware can return 12 bits to the computer.

    What you need is not more bits. You need a spectral shift. Your stained documents
    might really come alive, if you look at them through a part of the spectrum we
    cannot see well. You may find colours (probably in the IR band) where the stains
    don't show at all, and the text is beautifully clear. If you try an IR filter you
    might get interesting results. Most glasses will pass the high end of the IR band,
    and I think the CCD in the scanner will respond OK when you do a black and white
    scan. If you can find a suitable filter, its worth a try.

    No amount of clever signal processing will ever reconstruct the information you have
    lost. A scan of a stained image at R, G and B frquencies and 128 bit resolution will
    have little more genuinely useful information than one stored at 8 bit resolution.
    The RGB scanning has already lost the most interesting stuff, since it was designed
    to have very similar limitations to our own eyes. Today, signal processing software
    exists to deblur images, destain images and perform a number of other neat tricks.
    Whilst its pretty amazing to see the number plate of a getaway car resolved from the
    blur on a very slow photo, you can't actually construct useful information out of
    thin air. Signal processing can only reveal what is buried in the data.

    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 : Fri Dec 15 2000 - 19:29:18 PST