Scanners and 48 bit data

Discussion in 'Digital Photography' started by Don Stauffer in Minneapolis, Feb 22, 2005.

  1. Most scanners claim to provide 48 bit data. What application software
    can use 48 bit data? PS, PSP, ??

    Does the Twain standard import 48 bit data from scanner to application?

    If I want to save the image before I am finished editing it, what file
    formats retain 48 bit color depth?
    Don Stauffer in Minneapolis, Feb 22, 2005
    #1
    1. Advertising

  2. Don Stauffer in Minneapolis

    Scott W Guest

    Both Photoshop and Photoshop Elements 3 will work in 48 bit mode and
    yes you can get the file through the twain interface.

    For saving the file in 48 bit more I use the native file for Photoshop,
    psd files, you can also save in a number of other formats including jpg
    2000 and tiff.

    Scott
    Scott W, Feb 22, 2005
    #2
    1. Advertising

  3. Don Stauffer in Minneapolis

    rafeb Guest

    Don Stauffer in Minneapolis wrote:
    > Most scanners claim to provide 48 bit data. What application software
    > can use 48 bit data? PS, PSP, ??
    >
    > Does the Twain standard import 48 bit data from scanner to application?
    >
    > If I want to save the image before I am finished editing it, what file
    > formats retain 48 bit color depth?



    Photoshop certainly can use and manipulate 48 bit
    image files. ACDSee can usually view them (but
    sometimes chokes on very big ones.)

    ..PSD and .TIF formats can be used to save these files.

    I don't know much about TWAIN anymore. In Windoze
    it's been superceded by WIA, but in any case I find
    myself using standalone apps, both for my scanners
    and my digicams.


    rafe b.
    http://www.terrapinphoto.com
    rafeb, Feb 22, 2005
    #3
  4. Scott W commented courteously ...

    > Both Photoshop and Photoshop Elements 3 will work in
    > 48 bit mode and yes you can get the file through the
    > twain interface.
    >
    > For saving the file in 48 bit more I use the native
    > file for Photoshop, psd files, you can also save in
    > a number of other formats including jpg 2000 and tiff.


    Scott, being a PSP 9 user with no knowledge of PS at all,
    does PS and PSE really do everything in 48 bit?

    48-bit color comes up fairly often these days in a variety
    of NGs. Again, being ignorant of 48-bit color, I always
    ask people "can you tell the difference vs. 24-bit?" and
    "on what do you display it or print it?".

    Not to mention, I have to ask "what original could
    possibly be scanned with a 48-bit scanner that actually
    has that many colors?"

    Last I looked, Windoze was 32-bit, and the extra 8 bits
    are for transparency, rather than additional colors.

    --
    ATM, aka Jerry
    All Things Mopar, Feb 22, 2005
    #4
  5. Don Stauffer in Minneapolis

    Scott W Guest

    All Things Mopar wrote:
    > Scott W commented courteously ...
    >
    > > Both Photoshop and Photoshop Elements 3 will work in
    > > 48 bit mode and yes you can get the file through the
    > > twain interface.
    > >
    > > For saving the file in 48 bit more I use the native
    > > file for Photoshop, psd files, you can also save in
    > > a number of other formats including jpg 2000 and tiff.

    >
    > Scott, being a PSP 9 user with no knowledge of PS at all,
    > does PS and PSE really do everything in 48 bit?
    >
    > 48-bit color comes up fairly often these days in a variety
    > of NGs. Again, being ignorant of 48-bit color, I always
    > ask people "can you tell the difference vs. 24-bit?" and
    > "on what do you display it or print it?".
    >
    > Not to mention, I have to ask "what original could
    > possibly be scanned with a 48-bit scanner that actually
    > has that many colors?"
    >
    > Last I looked, Windoze was 32-bit, and the extra 8 bits
    > are for transparency, rather than additional colors.
    >
    > --
    > ATM, aka Jerry

    I don't know about PS but PSE 3 will do a lot in 48 bit mode but not
    all. It is handy to work in 48 bit mode as you have more dynamic range
    and you can do repeaded adjustments to the levels with out degreading
    the image, once you convert to a 24 bit bitmap you can't tell the
    differnace.

    Scott
    Scott W, Feb 22, 2005
    #5
  6. Don Stauffer in Minneapolis

    Chris Brown Guest

    In article <Xns9605716827A6AReplyToken@216.196.97.131>,
    All Things Mopar <> wrote:

    >48-bit color comes up fairly often these days in a variety
    >of NGs. Again, being ignorant of 48-bit color, I always
    >ask people "can you tell the difference vs. 24-bit?" and
    >"on what do you display it or print it?".


    I think, you may be asking the wrong question. The real use for 48 bit
    colour is to avoid quantisation problems (clipping and rounding) which occur
    due to image processing and the use of wide-gamut colourspaces (e.g.
    Adobe-RGB). For final presentation, 24bpp is perfectly adequate, but 48bpp
    is useful as a working format to avoid image degredation.
    Chris Brown, Feb 22, 2005
    #6
  7. Don Stauffer in Minneapolis

    RSD99 Guest

    "What application softwre can use 48 bit data?" The ones I know about are

    PhotoShop CS

    PhotoShop Elements 3 (limited)

    Picture Window Pro

    Corel PhotoPaint

    CinePaint (customized version of "The GIMP" ...
    but it's pretty much a "rolling alpha release" at this point)

    "If I want to save the image before I am finished editing it, what file
    formats retain 48 bit color depth?" The ones I know about are


    PhotoShop (PSD)

    TIFF

    *Probably* Corel PhotoPaint's proprietary format




    "Don Stauffer in Minneapolis" <> wrote in message
    news:...
    > Most scanners claim to provide 48 bit data. What application software
    > can use 48 bit data? PS, PSP, ??
    >
    > Does the Twain standard import 48 bit data from scanner to application?
    >
    > If I want to save the image before I am finished editing it, what file
    > formats retain 48 bit color depth?
    RSD99, Feb 22, 2005
    #7
  8. Scott W commented courteously ...

    > I don't know about PS but PSE 3 will do a lot in 48
    > bit mode but not all. It is handy to work in 48 bit
    > mode as you have more dynamic range and you can do
    > repeaded adjustments to the levels with out degreading
    > the image, once you convert to a 24 bit bitmap you
    > can't tell the differnace.


    Thanks, Scott.

    So, you're working on the histogram and other aspects of
    the image that can benefit from the additional bit width?

    I hang out on a couple PSP news groups, where 48-bit color
    comes up often. Natually, the Corel crowd pooh-poohs this,
    but I'm sure they are working on it.

    However, some of the very creative photographers and
    digital artists always ask "where's the beef?". One or 2
    steadfastly maintain that nearly all of the extra 48 bits
    is just noise at each end of the histogram.

    Natually, I don't know, since I don't have a camera or a
    scanner which will go beyond 24-bit. And, if you've seen
    any of my posts, my requirements as well as my PSP 9
    expertise, is a long ways below where I'd see any
    advantage to 48-bit. But, I'm trying to stay on top of the
    growing trend in the industry so I am ready for it when it
    becomes of interest to me.

    >
    > Scott
    >
    >




    --
    ATM, aka Jerry
    All Things Mopar, Feb 22, 2005
    #8
  9. Chris Brown commented courteously ...

    > I think, you may be asking the wrong question.


    I probably am asking the wrong question, Chris, as I don't
    understand enough about this nor have any hardware that I
    can use to test my "theories".

    > The real use for 48 bit colour is to avoid
    > quantisation problems (clipping and rounding)
    > which occur due to image processing and the use of
    > wide-gamut colourspaces (e.g. Adobe-RGB).


    I understand Adobe RGB vs. the "standard" sRGB color
    space, but not very well. Are you doing your image
    processing on RAW images, I assume?

    > For final presentation, 24bpp is perfectly adequate,
    > but 48bpp is useful as a working format to avoid image
    > degredation.


    I'm aware from talking to friends in Chrysler's Photo
    Imaging department that there *are* things you can "see
    with 32-bit or 48-bit color, I've been shown examples of
    banding and posterization in 24-bit that go away with high
    width. But, these guys have hundreds of thousands of
    dollars invested in computer hardware, cameras, printers,
    and color calibration software/hardware.

    With that bloviating preface, could you continue my
    education and explain briefly what you mean by "avoiding
    image degradation"?

    Thanks, Chris.

    --
    ATM, aka Jerry
    All Things Mopar, Feb 22, 2005
    #9
  10. All Things Mopar wrote:
    []
    > Natually, I don't know, since I don't have a camera or a
    > scanner which will go beyond 24-bit. And, if you've seen
    > any of my posts, my requirements as well as my PSP 9
    > expertise, is a long ways below where I'd see any
    > advantage to 48-bit. But, I'm trying to stay on top of the
    > growing trend in the industry so I am ready for it when it
    > becomes of interest to me.


    Actually, you have a Nikon 5700 which shoots in RAW mode if you wish. I
    understand that in that mode you get 12 bits per colour, so you can
    already generate 36-bit images. The 12-bit data is a linear value, the
    8-bit JPEG data nearer a log or gamma-corrected value, so the available
    dynamic range is similar in each, but the accuracy of representation in
    the 8-bit data isn't as high as in the 12-bit data. {But does that
    matter?]

    Broadly speaking, it's one of those awkward choices - 8 bits per colour is
    fine if everything is 100% exposed correctly and the dynamic range of the
    scene isn't too large. Having the 12-bit data gives you a little more
    margin. But computers work better with either 8-bit or 16-bit colours
    rather than 12-bit, so your choice is between 24-bit RGB and 48-bit RGB.
    By the way, there are those who argue that PhotoShop only works with
    15-bit colour data, not 16-bit data. Whatever, the lower few bits are
    just noise.

    Cheers,
    David
    David J Taylor, Feb 22, 2005
    #10
  11. David J Taylor commented courteously ...

    > Actually, you have a Nikon 5700 which shoots in RAW
    > mode if you wish. I understand that in that mode you
    > get 12 bits per colour, so you can already generate
    > 36-bit images.

    [snip]

    Didn't know that about my NEF files. Thanks, David!

    > Broadly speaking, it's one of those awkward choices
    > - 8 bits per colour is fine if everything is 100%
    > exposed correctly and the dynamic range of the scene
    > isn't too


    That makes perfect sense to me...

    > Having the 12-bit data gives you a little more
    > margin. But computers work better with either
    > 8-bit or 16-bit colours rather than 12-bit, so your
    > choice is between 24-bit RGB and 48-bit RGB.


    Yes, computers have to be on even byte boundaries,
    although these days, I believe that the CPU operates most
    often on a 32-bit "word".

    > By the way, there are those who argue that PhotoShop

    only
    > works with 15-bit colour data, not 16-bit data.
    > Whatever, the lower few bits are just noise.


    That's exactly what I've heard/read, but I wasn't sure
    enough of my "facts" to be dogmatic with my post.

    Thanks for the clarifications...

    --
    ATM, aka Jerry
    All Things Mopar, Feb 22, 2005
    #11
  12. Don Stauffer in Minneapolis

    Chris Brown Guest

    In article <Xns9605846F66D33ReplyToken@216.196.97.131>,
    All Things Mopar <> wrote:
    >
    >> The real use for 48 bit colour is to avoid
    >> quantisation problems (clipping and rounding)
    >> which occur due to image processing and the use of
    >> wide-gamut colourspaces (e.g. Adobe-RGB).

    >
    >I understand Adobe RGB vs. the "standard" sRGB color
    >space, but not very well. Are you doing your image
    >processing on RAW images, I assume?


    RAW images and scanned film. The latter almost certainly doesn't pick up
    anything but noise in the least significant 8 bits of each channel, but I
    work in 48 bits to avoid quantisation losses whilst processing in Photoshop.

    >> For final presentation, 24bpp is perfectly adequate,
    >> but 48bpp is useful as a working format to avoid image
    >> degredation.

    >
    >I'm aware from talking to friends in Chrysler's Photo
    >Imaging department that there *are* things you can "see
    >with 32-bit or 48-bit color, I've been shown examples of
    >banding and posterization in 24-bit that go away with high
    >width.


    You can construct examples within a 24 bit range where the transitions
    between adjacent bands can be seen, but it's generally not a problem in real
    photographic images.

    But this does highlight the reason to work in 48 bits - 24 bits is only just
    adequate for display, as you have hinted at above. Processing the image will
    lose some of the useful information in the low-order bits. By switching to
    48 bits for processing, those low-order bits don't contain anything useful
    anyway. If you process in 24 bits, however, you may well end up with an
    image that only has, say, 18 bits worth of real information at the end.

    >With that bloviating preface, could you continue my
    >education and explain briefly what you mean by "avoiding
    >image degradation"?


    An analogy may help - imagine you have the following set of numbers, which
    represent some image data:

    12
    4
    25
    8
    13

    Now we want to do some operation on them, which represents some operations
    in Photoshop - let's divide them all by 3 and add them, but we'll do it
    twice - once using just integers throughout, and again using one decimal
    place:

    divide by 3:

    4 4.0
    1 1.3
    8 8.3
    2 2.6
    4 4.3

    Now let's total them:

    19 20.5

    Even if we present our final result as an integer, we have got a more
    accurate answer by working at a higher precision in the second case. That's
    the main benefit from working in 48 bits - you keep the fractional parts of
    operations, parts which get discarded in 24 bits. After a lot of processing,
    all those missing fractions add up (or rather, don't).
    Chris Brown, Feb 22, 2005
    #12
  13. Chris Brown commented courteously ...

    [snip]
    > You can construct examples within a 24 bit range where
    > the transitions between adjacent bands can be seen, but
    > it's generally not a problem in real photographic

    images.

    This makes sense...

    [snip]
    > An analogy may help - imagine you have the following
    > set of numbers, which represent some image data:

    [snip]

    Yes, that does help. I now have a better understanding of
    the underlying mathematics of this, as looked at from a
    raster graphics format.

    And, this tells me that it'll be a *long* time, if ever,
    before I get into situations with my "documentary"-style
    digital photography where either RAW or 48-bit color would
    be useful! <grin>

    Thanks for the explanation!

    --
    ATM, aka Jerry
    All Things Mopar, Feb 22, 2005
    #13
  14. All Things Mopar wrote:
    >
    > I'm aware from talking to friends in Chrysler's Photo
    > Imaging department that there *are* things you can "see
    > with 32-bit or 48-bit color, I've been shown examples of
    > banding and posterization in 24-bit that go away with high
    > width. But, these guys have hundreds of thousands of
    > dollars invested in computer hardware, cameras, printers,
    > and color calibration software/hardware.
    >
    > With that bloviating preface, could you continue my
    > education and explain briefly what you mean by "avoiding
    > image degradation"?
    >
    > Thanks, Chris.
    >


    The choice of either clipping or compressing (curve shaping).

    This problem predates the digital camera. It was a problem in film
    also. Photographic paper had a density range of less than 2 (it is
    about 50:1 in best papers). Negative film, however, has a wider density
    range, most a density of three (1000:1), some even more.

    It was never possible to print the whole tonal range one found in a good
    neg onto paper. One had a number of choices. Print for shadow detail
    (concentrate on blacks), print for highlight detail (concentrate on
    whites), or use a lower contrast paper that compressed the range of the
    tonal range, lowering contrast but retaining both ends of the range.

    We can do the same thing even more easily with digital, since contrast
    changes and curve bending are simple clicks of a mouse. But you cannot
    print data that isn't in the file. Having more dynamic range in the
    original file (whether from camera or scanner) allows you the same
    creative choice that film printers had. One does not need to clip or
    compress until ready for the final print, and you can choose which to do.

    If one is scanning prints, it is a moot point 'cause the prints have
    limited range themselves, but with a transparency/negative adapter, I
    will be working with wider dynamic range originals.
    Don Stauffer in Minneapolis, Feb 23, 2005
    #14
  15. Don Stauffer in Minneapolis

    Scott W Guest

    Don Stauffer in Minneapolis wrote:
    > This problem predates the digital camera. It was a problem in film
    > also. Photographic paper had a density range of less than 2 (it is
    > about 50:1 in best papers). Negative film, however, has a wider

    density
    > range, most a density of three (1000:1), some even more.


    If should be pointed out that just because negative film has a density
    range of 1000:1 does not mean that is the dynamic range it captures,
    because it is non-linear, there is some compression, it captures light
    levels more in a range of 4000:1. It is just the range of 4000:1 is
    compressed into a range of 1000:1. You can do the same thing with
    prints, if you print at low contrast you can make a print that captures
    a dynamic range much larger than it is capably of reproducing, of
    course it will look like crap. It is very easy to compress a range of
    1000:1 to a range of 50:1 and print it, people don't tend to do this
    since the prints will look very poor.

    Scott


    Scott
    Scott W, Feb 23, 2005
    #15
  16. > If one is scanning prints, it is a moot point 'cause the prints have
    > limited range themselves, but with a transparency/negative adapter, I
    > will be working with wider dynamic range originals.
    >

    I have a couple of discussions on the value of 16 bit over 8 bit files
    in the tips section of my web site. Many people claim not to be able
    to see a difference, but I feel that starting with 16 bit before doing
    any large amounts of contrast or brightness adjustment improves the
    final image slightly.
    Take a look at the tests and judge for yourself.

    --
    Robert D Feinman
    Landscapes, Cityscapes and Panoramic Photographs
    http://robertdfeinman.com
    mail:
    Robert Feinman, Feb 24, 2005
    #16
    1. Advertising

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

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. Replies:
    0
    Views:
    717
  2. Homer J. Simpson
    Replies:
    5
    Views:
    715
    Jim Barry
    Jan 19, 2009
  3. David B.
    Replies:
    0
    Views:
    457
    David B.
    Jan 20, 2009
  4. Lanwench [MVP - Exchange]
    Replies:
    0
    Views:
    403
    Lanwench [MVP - Exchange]
    Jan 21, 2009
  5. sandi_ro
    Replies:
    2
    Views:
    501
    sandi_ro
    Jan 28, 2009
Loading...

Share This Page