Re: "Assigning" vs. "Matching" a color profile

Discussion in 'Digital Photography' started by me, Dec 29, 2009.

  1. me

    me Guest

    On Mon, 28 Dec 2009 11:11:08 -0800, isw <> wrote:

    >This is on a Mac, BTW.
    >
    >I have a large number of scanned slides bearing a color profile
    >(assigned by the scanner) that gives iPhoto fits; I'd like to change it.
    >
    >Using ColorSync, I can "assign" a different profile, or I can "match" to
    >a different profile, but I do not understand which I should do, or (more
    >importantly) what the difference is between the two. Further, I don't
    >know which profile I should move to: "Generic RGB"; "sRGB"; or what? The
    >images are my own, and will not be displayed on the web. I'd like to
    >keep them at the highest possible "accuracy" (whatever that means).
    >
    >A whole lot of googling has produced many descriptions of *how* to do
    >these things, but nothing on *why* or *which*.
    >
    >Can anybody shed some light, please?



    I've seen the other give and take on this and believe a fundamental
    concept is missing here. The profile assigned to the image by the
    scanner contains the information of how the colors mapped by the
    scanner correspond to the reference color space. A color space aware
    app is needed to read this. This then gets translated into the working
    color space of the app. Finally a monitor or printer profile is used
    to again translate this into the monitor or printer color space.

    Simply assigning a different color profile is not the answer. You need
    to convert to a different profile and then resave that image with the
    new profile. Not knowing anything about iPhoto I don't know if "match"
    to a different profile is what you want here, but suspect it might be.
    Check the help file for this function. Note that by doing this you
    will actually changing the color data in the image (IIRC)
     
    me, Dec 29, 2009
    #1
    1. Advertising

  2. me

    me Guest

    On Tue, 29 Dec 2009 09:45:48 -0800, isw <> wrote:


    >I'm OP, and I agree with what you've said, now that I finally understand
    >it. Now, the remaining question is, what is the preferred profile to
    >change to?
    >
    >Considering that there will certainly be improvements in the gamut
    >capabilities of display devices in the future, I'm thinking that sRGB is
    >probably too constraining. How about ProPhoto?



    First, what matters is the content of your images, not necessarily
    what colorspace they are in. Even though the scanners color space may
    be larger than sRGB or AdobeRGB, what matters is the actual colors
    contained. Can iPhoto show out of gamut colors? You could then see
    what colors are contained/maintained in a given color space. You also
    haven't mentioned what format you are saving in. A wider color space
    may bring about issues given the wider space must fit into a given
    amount of storage bits and hence there may be less graduations of a
    given color available.
     
    me, Dec 29, 2009
    #2
    1. Advertising

  3. me

    MikeWhy Guest

    "isw" <> wrote in message
    news:isw-4A37D4.20315329122009@[216.168.3.50]...
    > In article <>,
    > me <> wrote:
    >> You also
    >> haven't mentioned what format you are saving in. A wider color space
    >> may bring about issues given the wider space must fit into a given
    >> amount of storage bits and hence there may be less graduations of a
    >> given color available.

    >
    > Saving as "high quality" JPEGs; these are 35mm slides, scanned at 4800
    > ppi on a scanner that *claimed* to be able to handle it. But there's
    > nothing about the JPEG encoding process that forces a restricted gamut.
    > Storage space is not an issue, but I need to deal with iPhoto's problem
    > with the scanner's profile, and I don't want to throw anything away in
    > the process.


    Then you should archive the scans as they are, without converting the
    colorspace. Your choices for JPEG are 8 bit and 16 bit. As he pointed out,
    converting to a larger gamut requires more bits to maintain the same
    gradation in the larger space as 8 bits in the smaller color space. A
    lossless conversion will double the size of each file in the new colorspace.

    Would you mind posting or emailing a small crop of a 4800 ppi scan? I have
    never seen a scan that fine. Can your scanning service handle medium format
    and sheet film sizes? What technology do they use?
     
    MikeWhy, Dec 30, 2009
    #3
  4. me

    me Guest

    On Tue, 29 Dec 2009 20:31:53 -0800, isw <> wrote:


    >I don't think so. But that's not the issue. I don't care whether iPhoto
    >(or any other specific image handling app) can display the full gamut of
    >the images; what I want is to not lose something that might be usable on
    >*future* imaging equipment, whether a better screen, or a printer, or
    >whatever.


    Unless you have verified there are actual colors in the scan outside
    of sRGB or aRGB this obseeion with colorspace is just that, an
    obsession with little real implication. Given this course of reasoning
    IMO you would have been much better off saving the orignal scans to 24
    bit tiffs for the archival originals. Except if you have some raelyy
    special images I would doubt the color info lost would come anywhere
    close to the tonal graduation info lost saving to jjpeg.
     
    me, Dec 30, 2009
    #4
  5. me

    MikeWhy Guest

    isw wrote:
    > In article <hhfv18$s6c$-september.org>,
    > "MikeWhy" <> wrote:
    >
    >> "isw" <> wrote in message
    >> news:isw-4A37D4.20315329122009@[216.168.3.50]...
    >>> In article <>,
    >>> me <> wrote:
    >>>> You also
    >>>> haven't mentioned what format you are saving in. A wider color
    >>>> space may bring about issues given the wider space must fit into a
    >>>> given amount of storage bits and hence there may be less
    >>>> graduations of a given color available.
    >>>
    >>> Saving as "high quality" JPEGs; these are 35mm slides, scanned at
    >>> 4800 ppi on a scanner that *claimed* to be able to handle it. But
    >>> there's nothing about the JPEG encoding process that forces a
    >>> restricted gamut. Storage space is not an issue, but I need to deal
    >>> with iPhoto's problem with the scanner's profile, and I don't want
    >>> to throw anything away in the process.

    >>
    >> Then you should archive the scans as they are, without converting the
    >> colorspace. Your choices for JPEG are 8 bit and 16 bit. As he
    >> pointed out, converting to a larger gamut requires more bits to
    >> maintain the same gradation in the larger space as 8 bits in the
    >> smaller color space. A lossless conversion will double the size of
    >> each file in the new colorspace.
    >>
    >> Would you mind posting or emailing a small crop of a 4800 ppi scan?
    >> I have never seen a scan that fine. Can your scanning service handle
    >> medium format and sheet film sizes? What technology do they use?

    >
    > The scanner is a consumer-grade unit from Microtek; I acquired it a
    > few years ago for under $200. I'm under no illusions about how good
    > the scanner is with regard to dark noise and so on, but I wanted at
    > least to start out with high spatial resolution. I have several
    > thousand slides, and simply couldn't afford to have them
    > "professionally" scanned.
    >
    > I've posted four screenshots from GIMP here:
    >
    > http://profile.imageshack.us/user/isw/
    >
    > The images are in reverse order, but I couldn't figure out how to
    > rearrange them; I rarely post images.
    >
    > The ruler at the top of the screenshots (in mm) and the info at the
    > bottom give some scale to the images, and to the scanner's pixel size.
    > The actual pixels don't begin to show up until about 400%, where 1mm
    > is almost the full width of the image. A bit of fringing can also be
    > seen at high magnifications, but I really don't know whether it is
    > due to the scanner, or is in the original photo (which I can't access
    > right now).


    Thanks. That was pretty interesting. Yeah, I was curious about the fringing
    also. They're likely part of the grain structure, rather than sharpening
    halos from the scanner software. Those chemists in Rochester did some really
    things, back in the day.

    It might be instructive to look at a few slides under a microscope and
    compare them with what you see from the scanner. I went through the same
    exercise some years ago using an Epson flatbed. 1600 dpi was a bit of a
    stretch; I settled for 2400 dpi as a workable compromise. The biggest
    problems are flatness and focus. Not owning a drum scanner, I tried glass
    mounts and oil briefly. Unmounting, scanning, and remounting each slide
    proved even more tedious than it sounds. The lower resolution actually is
    quite reasonable and usable. I have no regrets about doing it the way I did.

    About the colorspace... I profiled the Epson with MonacoEZColor, and
    archived the scans unmodified. I have some doubts about the precision this
    chain can achieve, but it's more than adequate for my needs. I batch
    converted a "working set" to sRGB, and haven't bothered to dig through the
    original scans even once in all the time since.

    > The shot was made with a Nikon FTn using a 50mm Nikon
    > lens and a circular polarizing filter. In real life, that dish is 64
    > meters across; it's the "deep space" antenna at NASA's Goldstone site
    > in southern California.
    >
    > The whole file is about 1.3 MB, and I'll be happy to send it to you if
    > you like; just give me an address that won't mind an attachment of
    > that filesize.
    >
    > Isaac


    Good luck on your project.
     
    MikeWhy, Dec 30, 2009
    #5
  6. me

    me Guest

    On Wed, 30 Dec 2009 20:47:28 -0800, isw <> wrote:


    >Do you have any doubt that Kodachrome or Ektachrome slides can have a
    >greater color gamut than sRGB can reproduce? Adobe 98, I'm not so sure,
    >but it's certainly wider than sRGB.


    You just don't get it do you. Just because the medium, digital or
    analog, MAY contain a colorspace of a given size is one thing. Whether
    or not it is used is another. That is why I suggested looking at your
    scanned images with a color aware application which can be set to
    differing working color spaces and then show out of gamut colors.
     
    me, Dec 31, 2009
    #6
  7. me

    me Guest

    On Thu, 31 Dec 2009 09:35:12 -0800, isw <> wrote:


    >1) Images on Kodachrome originals *may* (depending on the subject
    >matter) have a wider gamut than sRGB can handle, and
    >
    >2) I have no intention of measuring, one-by-one, nearly three thousand
    >slides, to find out which ones in fact hold images needing that wider
    >gamut and which do not, so


    And no one said to do so. Use a little common sense. Pick a few of the
    more colorful shots and check them against aRGB. Is anything even
    close to being out of gamut? If not then just go with aRGB.


    >And right now I'm trying to learn enough to understand the best way to
    >do that. So far, my understanding is that sRGB *may* cause a
    >(content-dependent) loss of gamut vis-a-vis Kodachrome, and that
    >ProPhoto can handle it easily, but is probably too large a gamut unless
    >I'm willing to commit to doing everything to 16 bit precision (which I'm
    >not). So Adobe98 is my current "best guess" as to what I should convert
    >these images to.


    So then just do a couple. Compare them against the originals. Even try
    subtracting one image from another to see the diff. How big is it?

    >Now I need to figure out the least labor-intensive way to do that, and
    >why two methods, both provided by Apple, produce visibly different
    >results.


    Can't help here, since I'm a PCperson. But given decent SW this isn't
    that big a deal, with scripting, batch processing, etc.
     
    me, Dec 31, 2009
    #7
  8. me

    MikeWhy Guest

    "isw" <> wrote in message
    news:isw-5BBB07.09351231122009@[216.168.3.50]...
    > 1) Images on Kodachrome originals *may* (depending on the subject
    > matter) have a wider gamut than sRGB can handle, and


    You still need to consider the scanner's gamut. The hardware itself might
    not be up to recording the 'Chrome's full glory; cramming it through the
    scanner's profile will already have lost some of the raw information; and
    each subsequent remapping will lose still more. If you really want to worry
    yourself sick, the 'Chrome itself is an imperfect image of the original
    scene. ;)

    I read elsewhere you were using the manufacturer's supplied profile, and I
    wonder if I should even bring this up. It's better than nothing, of course,
    and by itself is probably sufficient to minimize unintended color casts.
    MonacoEZColor came with my Epson. It uses IT8 calibration targets, one for
    transparencies and another for reflective, to build device specific
    profiles. The differences between the stock profile and the custom profile
    were rather significant. Switching from one to the other, there were
    distinct color pops across the entire image, but none I would consider large
    or damaging.

    For that matter, I use X-Rite ColorChecker to profile my camera sensors.
    There's nothing really wrong with the images as they come out of the camera,
    but I find image fidelity improves with the custom profile. Just this
    afternoon, because it's still fresh on my mind, I grabbed a quick shot of
    soft sunlight caressing a bottle of Hoppe's oil sitting on the window sill.
    Before applying the camera profile, the bottle was rendered almost a dull
    red rather than Hoppe orange. You could almost feel the texture of the
    plastic in your hands looking at it onscreen. (Printing is a whole another
    exercise in anality. It's also better to not ask about the monitors and
    their calibration.)

    Again, there's nothing at all wrong with the images that come of the camera,
    even without profiling the sensor. It's the exact same situation as your
    stock scanner profile compared to a custom, measured profile. You wouldn't
    normally even think about it to complain, but once you start down this road,
    everything becomes a mess of profiling and obsessive white balancing and
    pixel peeping. You should know upfront where this is taking you.
     
    MikeWhy, Jan 1, 2010
    #8
  9. isw <> wrote:

    > What I "get" is:


    > 1) Images on Kodachrome originals *may* (depending on the subject
    > matter) have a wider gamut than sRGB can handle, and


    > 2) I have no intention of measuring, one-by-one, nearly three thousand
    > slides, to find out which ones in fact hold images needing that wider
    > gamut and which do not, so


    > 3) What makes sense to me is to handle all the images alike, and in such
    > a way that minimal information is lost *no matter the content*.


    First: calibrate your scanner. You'll need an IT8 Kodachrome
    Target (google!) for your scanner. Otherwise, you're loosing
    information right away at the scanner.

    Then, save the raw data from the scanner with the ICC profile from
    your scanner calibration. That way, you can decide in 10 years
    you want to do something different --- and you can start again
    right at the point the data came into your computer. That way,
    you loose no information past the scanning process.

    If you want to have 'final' images (alongside the above named
    RAWs!), it depends on what you want to do with them: "display
    on the web" or "consumer photo developing", sRGB and JPEG is the
    way to go.

    > So far, my understanding is that sRGB *may* cause a
    > (content-dependent) loss of gamut vis-a-vis Kodachrome, and that
    > ProPhoto can handle it easily, but is probably too large a gamut unless
    > I'm willing to commit to doing everything to 16 bit precision (which I'm
    > not).


    Why not?? After all, your primary task is 'loose minimal
    information', not 'stay in 8 bit and save disk space', isn't it?

    > So Adobe98 is my current "best guess" as to what I should convert
    > these images to.


    Not really. The best guess is writing a small program that sees if
    you loose information *for this given image* at a list of colour
    spaces (sorted by size) and choose the colour space and bit depth
    that looses none --- then run that over all your images and let
    your computer do the work. Don't forget to attach the correct
    colourspace to your images.

    > Now I need to figure out the least labor-intensive way to do that,


    have a program do the work, see above.

    > and
    > why two methods, both provided by Apple, produce visibly different
    > results.


    They shouldn't produce visibly different results outside
    minimal adaptions for the colour space ... you might be doing
    something wrong. Maybe as easy as converting to Adobe RGB, but
    not attaching that information to the image, so it gets shown as
    sRGB (and looks somewhat washed out).

    -Wolfgang
     
    Wolfgang Weisselberg, Jan 2, 2010
    #9
  10. me

    me Guest

    On Sat, 02 Jan 2010 10:39:12 -0800, isw <> wrote:


    >I suppose, but I have no idea what. In both cases (one manual, one
    >running an Applescript), it sure seems like the same thing should happen
    >(I've looked inside the script, which controls the same app that I use
    >to do the manual conversion).
    >
    >The difference between the output images seems to be no more than a
    >change in levels, because I can make a simple adjustment to the "paler"
    >one to make the two visually identical.


    What options are you given in converting? In PS CS2 you are given the
    Intent options of Perceptual, Saturation, Relative and Absolute
    Colormetric and all but the Absolute Colormetric all you to choose
    Black Point Compensation.
     
    me, Jan 2, 2010
    #10
  11. me

    me Guest

    On Sat, 02 Jan 2010 21:02:35 -0800, isw <> wrote:


    >> What options are you given in converting? In PS CS2 you are given the
    >> Intent options of Perceptual, Saturation, Relative and Absolute
    >> Colormetric and all but the Absolute Colormetric all you to choose
    >> Black Point Compensation.

    >
    >Based on what I've recently learned, "perceptual" is the best choice for
    >what I want to accomplish. I suspect that whether I use ColorSync
    >Utility or Preview (or the Applescript, which calls ColorSync Utility)
    >to do the deed, there's only one underlying algorithm (which Apple uses
    >for all color profile work). If any of those offers a black point
    >adjustment, I don't know where it is or how to find it. It's not even
    >clear to me if an arbitrary (i.e. user controlled) black point move is
    >appropriate during a color profile change.


    Finally had some time to pull out my copy of Real World Color
    Management by Fraser, Murphy and Bunting. The BP option is supposedly
    a "proprietary Adobe feature." By mapping black to black you can
    avoid: 1. clipping shadow detail if the source has a lower black
    point than the destination; and 2. having no real blacks if the source
    has a higher black point than the destination.
     
    me, Jan 8, 2010
    #11
    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. PTRAVEL

    Matching Color

    PTRAVEL, Aug 26, 2003, in forum: Digital Photography
    Replies:
    2
    Views:
    452
    Frogiswrong
    Aug 27, 2003
  2. Nick N

    matching color to printer?

    Nick N, Apr 15, 2004, in forum: Digital Photography
    Replies:
    2
    Views:
    326
    Marvin Margoshes
    Apr 15, 2004
  3. dtn
    Replies:
    3
    Views:
    542
    mark_digital
    Jul 16, 2004
  4. peter

    color matching question for web album prints

    peter, Aug 10, 2004, in forum: Digital Photography
    Replies:
    1
    Views:
    391
    Bill Hilton
    Aug 10, 2004
  5. me
    Replies:
    4
    Views:
    342
Loading...

Share This Page