Scan Comparison Site, Call for Sample Scans

Discussion in '35mm Cameras' started by Rafe B., Dec 16, 2003.

  1. Rafe B.

    Rafe B. Guest

    I've assembled a collection of film scan samples at

    <http://www.terrapinphoto.com/jmdavis/>

    The idea is to see just how much image detail
    can be pulled from 0.25" by 0.25" of film, or
    an equivalent area of a digicam image.

    Forget MTF curves and line pairs per millimeter.
    This is a simple gallery of images achievable
    in practice with good gear and good technique.

    There are samples here from several popular
    drum scanners (Aztec, Howtek, ICG) and CCD
    scanners (Nikon, Minolta, Imacon, Leaf) As well
    as a couple of digicams (Canon 1Ds, 10D.)

    I'd love to post many more. Good sharp scans only.

    Any camera, any lens, any film, any scanner.
    Cherry pick the images all you like.
    There are only a few simple rules:

    1. Image should represent 0.25" by 0.25" of
    film at native scan resolution.
    2. No unsharp masking
    3. Include a small overview image of the frame
    from which the scan was taken and indicate the
    film format (eg., 35 mm, 6x6, etc.)
    4. Compress as high quality JPG.
    5. Post to a web location and indicate the
    URL, or email JPGs directly to me at:

    rafe DOT bustin AT verizon DOT net

    Please, before posting, have a look at the images
    and scanners already represented. If your scanner
    is already represented, there's no need to post
    anything (and I will most likely ignore your submission)
    unless it clearly matches or beats other samples from
    that same model.

    Thanks. And enjoy.


    rafe b.
    http://www.terrapinphoto.com
    scan comparisons
    http://www.terrapinphoto.com/jmdavis/
     
    Rafe B., Dec 16, 2003
    #1
    1. Advertisements

  2. ....
    JPEG will destroy the details you want to have in your samples, and
    also remove much of the noise. This is the oposite of what you
    need. You shoud use TIFF with data compression or the like. But *NO*
    image compression!!

    --
    Paul Repacholi 1 Crescent Rd.,
    +61 (08) 9257-1001 Kalamunda.
    West Australia 6076
    comp.os.vms,- The Older, Grumpier Slashdot
    Raw, Cooked or Well-done, it's all half baked.
    EPIC, The Architecture of the future, always has been, always will be.
     
    Paul Repacholi, Dec 16, 2003
    #2
    1. Advertisements

  3. Rafe B.

    Bill Hilton Guest

    Hi Paul,

    I understand what you're getting at and always work in TIFF as long as possible
    (never use jpeg except for web images), but it wouldn't be practical for Rafe
    to be posting TIFFs due to file sizes.

    I'm not so sure you lose THAT much with just one conversion from TIFF to high
    quality JPEG. Could you post two samples, one of a TIFF and another of the
    same file converted to HQ JPEG that actually shows how much detail is destroyed
    and how much noise is removed? Would be interesting to see ... I don't think
    it's as bad as you indicate but am willing to be proven wrong.

    Haven't downloaded the files from Rafe's site yet but it sounds like a good
    idea to me.

    Bill
     
    Bill Hilton, Dec 16, 2003
    #3
  4. Rafe B.

    Bill Hilton Guest

    From: Rafe B.
    Rafe, this is a very good idea and well-put together, thanks. We need more
    posts like this on the net and fewer food fights.

    I especially liked the ones where the same film was scanned on different
    scanners, you can learn a LOT by looking at those (almost makes me want to turn
    in my Nikon 8000 for a drum scanner :). Also found the Vuescan vs Silverfast
    comparison interesting.

    A suggestion and a couple of nits to pick --

    1) It would be great if you could encode more info in the title since most
    people using this will likely download the files and check them carefully in
    Photoshop. For example "davis_n8000_velvia.jpg" or, for the drum scans, the
    actual scan rez used, like "davis_tango_5000dpi_provia100F.jpg" or whatever.
    At least I'd like to see that.

    2) Haven't looked at all of them but noticed a couple of things ... the
    Lawrence Smith submission says "Nikon 4000" in one column and "Nikon 8000" in
    another. While the software is pretty much identical for these scanners the
    hardware is a lot different and it's probably worthwhile knowing which scanner
    was actually used.

    3) Mispelled "Littleboy" as "Littelboy".

    Keep it coming, especially the ones that scan the same film on different
    scanners or with different software.

    Bill
     
    Bill Hilton, Dec 16, 2003
    #4
  5. Rafe B.

    David Chien Guest

    www.imaging-resource.com for many scanner test reviews and various
    images scanned for comparison of contrast, resolution, etc.
     
    David Chien, Dec 16, 2003
    #5
  6. Rafe B.

    Rafe B. Guest


    I beg to differ. There is no observable difference in this
    case between the JPG and the TIFF. No details have
    been lost. Furthermore, TIFFs can't be viewed by web
    browers and the file sizes of these images would be huge.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 17, 2003
    #6
  7. Rafe B.

    Ti Moute Guest

    I'm was reading about the scanned film vs digital images and it became
    clear to me that I was confused about something that hopefully you can
    clarify.

    Why is it that "For digital captures, the camera's largest native
    image size is taken to be 1" high"?

    Thanks
    Ti
     
    Ti Moute, Dec 17, 2003
    #7
  8. Rafe B.

    Rafe B. Guest


    It's somewhat arbitrary, unless you're using 35 mm
    as the reference, in which case it makes a good
    deal of sense.

    For starters, consider that 1.0" x 1.5" is a close
    approximation to the true size of 35 mm image.

    The 10D produces an image that's 2048 by 3072
    pixels. There's no "dpi" figure in this case, but if
    you equate this image to a 35 mm frame then you
    can think of the 10D image as a "scan" at 2048 dpi.

    Or think of it this way -- again, taking the 35 mm
    frame to be 1" x 1.5" -- my arbitrary sample
    images are 0.25" by 0.25" which is exactly
    1/24 of a 35 mm frame. Applying this same
    scaling to the 10D image you get a snippet
    of 512 x 512 pixels -- ie., 1/24 of the total
    10D image area.

    So the "condensed" version of this arbitrary
    rule is: the short axis of the digicam image,
    measured in pixels, can be thought of as
    the equivalent "dpi" of the digicam capture,
    so long as we understand that we're
    normalizing to a 35 mm frame that's taken
    to be 1" by 1.5"

    You might argue that this standard isn't
    appropriate for scanning backs and MF
    digital backs, but I'll cross that bridge
    when I get some sample images from
    one of those.


    rafe b.
    http://www.terrapinphoto.com
    scan comparisons
    http://www.terrapinphoto.com/jmdavis
     
    Rafe B., Dec 17, 2003
    #8
  9. Then you are wasting your time. You will have a collection that
    is limited by whatever the JPEG encoder someone uses is, and
    have lost the original data.

    Perhaps you should get a better browser, mine if fine with TIFFs.

    --
    Paul Repacholi 1 Crescent Rd.,
    +61 (08) 9257-1001 Kalamunda.
    West Australia 6076
    comp.os.vms,- The Older, Grumpier Slashdot
    Raw, Cooked or Well-done, it's all half baked.
    EPIC, The Architecture of the future, always has been, always will be.
     
    Paul Repacholi, Dec 17, 2003
    #9
  10. Rafe B.

    Ti Moute Guest

    If the digital image size of 1 inch is arbitrary, then I don't see how
    it can be used for comparisons.
    If the digital image size were arbitrarily 4 inches or 1/4 inch then
    the results would more than likely be different on your site.
    Wouldn't it be better to scan film at the same 2048 by 3072 pixels
    with the same color depth then compare the two images.
    Or either scan the film at maximum resolution and import the digital
    image at its maximum then convert both to 1000x1000 for a comparison.
     
    Ti Moute, Dec 17, 2003
    #10
  11. I've never been able to find any significant differences between original
    tiffs and low compression jpegs. Your comments are at odds with objective
    reality. At sensible compression settings, the vast majority of pixels are
    identical. Sure, at higher compression levels things get ugly, but not at
    low compression settings. Given the small amount of high frequency
    information (other than grain noise) in scans, the enormous tiff file macho
    in scanner users is quite misplaced.

    David J. Littleboy
    Tokyo, Japan
     
    David J. Littleboy, Dec 18, 2003
    #11
  12. Rafe B.

    Rafe B. Guest


    I'm afraid you quite missed the point.

    This arbitrary one inch rule is only used in
    an attempt to compare digicam captures to
    film scans. It does not apply to the film scans
    at all.

    I've explained the logic of it, so I'm not going
    to go through it again.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 18, 2003
    #12
  13. Rafe B.

    Rafe B. Guest



    Trust me, nothing's been lost with the JPG conversion.

    Those first two "perfect" scans are also JPGs and
    yet it's still abundantly obvious how much clearer
    they are than any of the film scans.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 18, 2003
    #13
  14. Rafe B.

    Rafe B. Guest

    Yes, of course it would. But given that 35 mm is
    presently the standard toward which most digicams
    aspire, there is special significance to the dimensions
    of the 35 mm frame -- and for simplicity's sake, I'm
    calling that 1" by 1.5".


    How do you scan film at 2048 x 3072 pixels?

    You could scan film at 2048 dpi, which would give you
    an image of around 2048 x 3072 pixels. And that would
    in fact give you a fairly decent 8x10" print, if that were
    true optical resolution we're talking about.

    But 2048 is generally recognized to be on the low side
    for scanning a sharp image on fine grained film. Even
    I'll agree that film holds something more than that.

    You are correct about one thing: a proper comparison
    needs to recognize that different dpi scans (of course)
    produce differently sized images from the same area
    of film.

    This is explained (albeit briefly) on the site. To "deal"
    with it, simply choose some arbitrary and reasonable
    standard image size and resample as necessary to
    that size.

    Eg. if you want to properly compare a Howtek scan
    at 4399 dpi to the Nikon at 4000 dpi, one way would
    be to upsample the Nikon scan sample from 1000 to
    1100 pixels on a side.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 18, 2003
    #14
  15. Rafe B.

    Ti Guest


    If you are talking about the information on the website, I'm not in agreement
    with it but since you answered my question with two threads, I will continue
    this in your other response with some reasons why.
     
    Ti, Dec 18, 2003
    #15
  16. Rafe B.

    Ti Guest

    I don't necessarily agree with arbitrarily picking numbers to suit an experiment
    for the sake of comparison. If you compared to other size film then you would
    end up with different dpi for the digital camera. But since you are using 35mm
    cameras, you have selected chosen 1" and I understand your reasoning but it
    still doesn't mean it is factually correct.
    Here is your first words on your page "How much image detail from 0.25" by 0.25"
    of film?"

    The would think best way to determine detail would be to see how big a picture
    could be produced from the film without losing its quality. The same would go
    for a digital image.

    Going the other way would seem to determine the quality of the scanner.
     
    Ti, Dec 18, 2003
    #16
  17. Rafe B.

    Rafe B. Guest


    What is so difficult to comprehend about this?

    In what way does my methodology compromise
    or "constrain" image quality?

    I've posted scans at full resolution based on
    a fixed area of film.

    It goes without saying (or so I thought) that larger
    film formats produce more total information.

    Choosing 1/4" by 1/4" was merely a convenient
    reference area, for several reasons:

    1. With the best film scanners typically
    returning images of 4000 dpi, the resulting
    snippet size is 1000 x 1000 pixels, which
    can be viewed on a decent hi-res monitor.

    2. Hopefully obvious, but using a small
    area also allows users to scan at the full
    capabilities of their scanners.

    2. Bandwidth issues for web browsing

    3. File size issues for web browsing


    rafe b.
    http://www.terrapinphoto.com
    scan comparisons
    http://www.terrapinphoto.com/jmdavis
     
    Rafe B., Dec 19, 2003
    #17
  18. Rafe B.

    stacey Guest

    ??? Even the basic KDE browser Konquerer opens them fine. Maybe you need a
    better browser/OS?
     
    stacey, Dec 19, 2003
    #18
  19. Rafe B.

    Alan Browne Guest


    Netscape 7 accesses TIF's via a viewer such as Quicktime.
     
    Alan Browne, Dec 19, 2003
    #19
  20. Rafe B.

    jjs Guest

    And TIFFs can be compressed.
     
    jjs, Dec 19, 2003
    #20
    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.