transforming images rgb to cmyk

Discussion in 'Photoshop Tutorials' started by indide_designs, May 22, 2005.

  1. I want to know how can i transform rgb images to cmyk without loosing
    original colors in te printing process
    thanks
     
    indide_designs, May 22, 2005
    #1
    1. Advertising

  2. indide_designs

    Tacit Guest

    In article <42908a9b$0$12857$>,
    "indide_designs" <> wrote:

    > I want to know how can i transform rgb images to cmyk without loosing
    > original colors in te printing process


    Impossible. The laws of physics forbid it. Certain RGB colors can not be
    reproduced in the CMYK color space, period.

    Sorry...

    --
    Art, photography, shareware, polyamory, literature, kink:
    all at http://www.xeromag.com/franklin.html
     
    Tacit, May 22, 2005
    #2
    1. Advertising

  3. indide_designs

    Mike Russell Guest

    indide_designs wrote:
    > I want to know how can i transform rgb images to cmyk without loosing
    > original colors in te printing process
    > thanks


    Tacit is right, as usual, in his comment that exact conversion from RGB to
    CMYK is not possible. But as you might guess from the fact that printing
    works, the situation is far from hopeless.

    Certain very saturated colors - pure reds and blues for example - lose some
    of their saturation when you convert from RGB to CMYK. This is a limitation
    of the inks that are used in printing, and for that matter similar
    limitations exist for chemical based photographic printing processes. So we
    live with it.

    Considerable preparation is needed before you can create CMYK images
    reasonably well, targetting them for particular press conditions as well as
    compensating for the relatively minor color shifts that will inevitably
    occur. There is a world of misinformation out there (none of it from Tacit,
    BTW :) and I would recommend that you spend a week reading one of Dan
    Margulis's books to get a handle on the issues, then work with your printer
    to create CMYK images that meet the necessary total ink limit and dot gain
    specifications.
    ---
    Mike Russell
    www.curvemeister.com
     
    Mike Russell, May 22, 2005
    #3
  4. indide_designs

    Hecate Guest

    On Sun, 22 May 2005 14:35:20 +0100, "indide_designs"
    <> wrote:

    >I want to know how can i transform rgb images to cmyk without loosing
    >original colors in te printing process
    >thanks
    >

    You can't. It's a different colour space. All you can do is get
    nearest approximations.

    --

    Hecate - The Real One

    Fashion: Buying things you don't need, with money
    you don't have, to impress people you don't like...
     
    Hecate, May 22, 2005
    #4
  5. indide_designs

    Guest

    Hecate wrote:
    >
    > On Sun, 22 May 2005 14:35:20 +0100, "indide_designs"
    > <> wrote:
    >
    > >I want to know how can i transform rgb images to cmyk without loosing
    > >original colors in te printing process
    > >thanks
    > >

    > You can't. It's a different colour space. All you can do is get
    > nearest approximations.


    I recall reading somewhere that converting to/from LAB is "lossless". If
    that is true, will RGB to LAB to CMYK work?
     
    , May 23, 2005
    #5
  6. indide_designs

    Bill Hilton Guest

    >I recall reading somewhere that converting to/from LAB is "lossless".

    Draw a gradient in RGB mode and look at the histogram, convert to LAB
    and back to RGB and look at the histogram ... the "fuzz" is gone,
    telling you it's not completely lossless, though it's certainly good
    enough for most applications.

    >If that is true, will RGB to LAB to CMYK work?


    Actually converting from RGB to CMYK already goes thru LAB in
    Photoshop, so the answer is "no".
     
    Bill Hilton, May 23, 2005
    #6
  7. indide_designs

    Hecate Guest

    On Mon, 23 May 2005 13:07:08 GMT, wrote:

    >
    >
    >Hecate wrote:
    >>
    >> On Sun, 22 May 2005 14:35:20 +0100, "indide_designs"
    >> <> wrote:
    >>
    >> >I want to know how can i transform rgb images to cmyk without loosing
    >> >original colors in te printing process
    >> >thanks
    >> >

    >> You can't. It's a different colour space. All you can do is get
    >> nearest approximations.

    >
    >I recall reading somewhere that converting to/from LAB is "lossless". If
    >that is true, will RGB to LAB to CMYK work?


    No it's not and no.

    --

    Hecate - The Real One

    Fashion: Buying things you don't need, with money
    you don't have, to impress people you don't like...
     
    Hecate, May 23, 2005
    #7
  8. indide_designs

    tacit Guest

    In article <>, wrote:

    > I recall reading somewhere that converting to/from LAB is "lossless". If
    > that is true, will RGB to LAB to CMYK work?


    Converting TO Lab is lossless. Converting FROM Lab is not. When you
    convert a Lab image to CMYK, the same thing happens that happens when
    you convert RGB to CMYK--some of the colors are imnpossible to reproduce
    in CMYK, and those colors change.

    Certain colors can not be reproduced in CMYK, period. The laws of
    physics forbid it. There is no workaround; it does not matter how you do
    the conversion; it does not matter what modes you go through first. You
    simply can't get certain colors in CMYK, and no fiddling in Photoshop
    (or any other program) can change that. That's one of the fundamental
    rules of life you learn when dealing with print.

    However, it's not that bad. The colors look flat when compared to the
    RGB original; but when you see them on their own, when you make a
    printout or see them on press, they don't look bad at all.

    --
    Art, photography, shareware, polyamory, literature, kink:
    all at http://www.xeromag.com/franklin.html
     
    tacit, May 24, 2005
    #8
  9. indide_designs

    Guest

    tacit wrote:
    >
    > In article <>, wrote:
    >
    > > I recall reading somewhere that converting to/from LAB is "lossless". If
    > > that is true, will RGB to LAB to CMYK work?

    >
    > Converting TO Lab is lossless. Converting FROM Lab is not. When you
    > convert a Lab image to CMYK, the same thing happens that happens when
    > you convert RGB to CMYK--some of the colors are imnpossible to reproduce
    > in CMYK, and those colors change.


    Thanks for the clarification.

    > Certain colors can not be reproduced in CMYK, period. The laws of
    > physics forbid it. There is no workaround; it does not matter how you do
    > the conversion; it does not matter what modes you go through first. You
    > simply can't get certain colors in CMYK, and no fiddling in Photoshop
    > (or any other program) can change that. That's one of the fundamental
    > rules of life you learn when dealing with print.
    >
    > However, it's not that bad. The colors look flat when compared to the
    > RGB original; but when you see them on their own, when you make a
    > printout or see them on press, they don't look bad at all.


    By "Certain colors can not be reproduced in CMYK", I assume you meant
    out of gamut colors between color spaces. But some colors on a monitor
    that are not flagged as oog in PS still won't show up in Epson inkjet
    prints, after all the fuzz of monitor calibration and custom printer
    profiles. Is the PS oog detection faulty? Aside from oog detection, are
    there other ways to find out what colors won't print?
     
    , May 24, 2005
    #9
  10. indide_designs

    Mike Russell Guest

    wrote:
    [re conversion of RGB to CMYK]
    > By "Certain colors can not be reproduced in CMYK", I assume you meant
    > out of gamut colors between color spaces. But some colors on a monitor
    > that are not flagged as oog in PS still won't show up in Epson inkjet
    > prints, after all the fuzz of monitor calibration and custom printer
    > profiles. Is the PS oog detection faulty? Aside from oog detection,
    > are there other ways to find out what colors won't print?


    There are a host of technical issues with out of gamut colors. The ICC
    standard defines a "tag" that defines out of gamut colors.

    But the gamut tag is not always included in a profile, and when it is, it is
    not used consistently by different manufacturers. So programs like
    Photoshop generally ignore the gamut tag, perferring instead to rely on
    color values calculated using the profile. One rather simple method of
    doing this is to see if converting the color back and forth gives the same
    result. For example, converting the out of gamut RGB (255,0,0) into CMYK
    and back again gives a different RGB value from the original (RGB(199,3,3),
    say). When this happens, that particular color is considered out of gamut.

    There are other, better, ways to calculate out of gamut colors, and I
    suspect the Photoshop engine uses one of those. BTW, out of gamut colors
    are calculated by the engine, not by Photoshop itself. You can experiment
    by telling Photoshop to use the Microsoft (or Apple) ICM engine in your
    color prefs.

    But this only says that the profile generates unique values for a particular
    color, and not whether the printer itself yields a different color when that
    color and colors close to it in value are printed. Given the limits of the
    current technology, particularly the poor support of the gamut tag, the only
    way to tell for sure is to print the colors, and scan them with a photometer
    or spectrophotometer. You can't even look at the darn things to tell if
    they are different, because the differences are very subtle. No wonder
    there is so much confusion, so much being spent on calibration equipment,
    and relatively little result to show for it.

    But most of us realize that our inkjet printers do a pretty good job. If
    printers were watches, at this time in history they would accurate to a
    couple of minutes a day. As was the case with watches, there are those who
    require (or think that they require) greater accuracy, and are willing to
    pay for it. As has happened already with watches, at some point in the
    future printers will probably be extremely accurate and inexpensive. Until
    that happens, most of us can easily live comfortably with less than
    perfection.

    Here are some more thoughts that I hope will give pause to some of you who
    have bought into the conventional wisdom that larger gamuts are always
    better than smaller ones.

    1) What was the last out of gamut object you photographed? Blue sky, for
    example, is not really that saturated, particularly near the horizon where
    most of our cameras are pointed. Red objects - even bright bird plumage, is
    a far cry from RGB(255,0,0). For most of us, the answer to the question is
    "none".

    2) RGB is not the last word in color spaces. Consider pure yellow objects,
    which are not all that uncommon, are not at all well represented in the RGB
    color space. For most monitors RGB(255,255,0) is brighter, but less
    saturated, than CMYK(0,0,100,0) will be in print. Magenta and cyan objects
    have similar problems.

    Recent monitor developments have made an end run around this problem,
    resulting in the apparent miracle of a monitor capable of displaying the
    Adobe RGB color space directly. This is achieved by filtering the RGB
    phosphors to create purer colors, sharpening and stretching the three
    corners of the RGB gamut (to see these shapes, check out Curvemeister's
    Labmeter, a free gamut plotter image). But even purity can have its limits,
    and the extermely sharp spectral characteristics of these monitors are
    bringing a new problem, viewer metamerism, to the forefront. With this
    latest advance, color consistency is literally in the eye of the beholder.
    The color on these monitors simply look funny to some people.

    3) Consider that a larger gamut such as Adobe RGB sacrifices color
    gradation. An Adobe RGB image uses a smaller number of color values to
    represent the same range of colors than an sRGB image. This issue may be
    addressed by working in 16 bits, but there is a more serious problem to
    working in a large gamut space: the high probability that someone else may
    look at your image on an sRGB monitor and conclude that your work is too
    drab. Both of these issues - particularly the second one - are reasons for
    working in sRGB, particularly since real world objects such as pure blue sky
    do not come close to exceeding the gamut of even sRGB.

    4) Print has a much smaller gamut than a CRT. The big money in photography
    is still in printed images, whether they be published images, or wall-sized
    art.

    My suggestion, as always, is to trust what you can see and verify, and not
    spend too much time or money on getting your "watch" to run within one
    second of correct. Todays color and printer technology is excellent, and
    most of us - the vase majority - can live with its imperfections, provided
    we understand them, and base our understanding on common sense.
    --

    Mike Russell
    www.curvemeister.com
     
    Mike Russell, May 24, 2005
    #10
  11. indide_designs

    Tacit Guest

    In article <>, wrote:

    > By "Certain colors can not be reproduced in CMYK", I assume you meant
    > out of gamut colors between color spaces.


    Yes, that's correct. For example, pure RGB blue ain't never going to be
    reproduced in CMYK.

    > But some colors on a monitor
    > that are not flagged as oog in PS still won't show up in Epson inkjet
    > prints, after all the fuzz of monitor calibration and custom printer
    > profiles. Is the PS oog detection faulty? Aside from oog detection, are
    > there other ways to find out what colors won't print?


    Photoshop's out of gamut detection is remarkably good; however, there
    are many issues which may confound trying to identify out of gamut
    colors on an inkjet print.

    For starters, Photoshop's out of gamut detection uses the current CMYK
    setup; if this is incorrect (for example, if you're using a SWOP
    profile, intended for printing presses, rather than a profile for your
    inkjet printer), Photoshop will show you the colors out of gamut for a
    printing press, not your printer.

    Second, when you print CMYK to an inkjet printer, the printer's drive
    software converts from CMYK to RGB, then back to the printer's own CMYK.
    So the data the printer sees may not match the data Photoshop sees,
    since the printer driver's CMYK->RGB conversion isn't very good.

    Third, consumer-grade inkjet printers don't even use pure primary CMYK
    inks at all. In particular, the cyan ink that an inkjet printer uses
    isn't really cyan; it's too blue. This is done to get more vibrant, more
    saturated colors that consumer users like, at the expense of color
    accuracy. Professional inkjet printers use inks closer to CMYK
    primaries, and are easier to calibrate for press-accurate CMYK output.

    --
    Art, photography, shareware, polyamory, literature, kink:
    all at http://www.xeromag.com/franklin.html
     
    Tacit, May 25, 2005
    #11
  12. indide_designs

    Mike Russell Guest

    Tacit wrote:
    >... Photoshop's out of gamut detection uses the current CMYK
    > setup ...


    CMYK is indeed the default, but you may set the gamut warning to use any
    profile that may be used as a working space.

    LabMeter, which is a free download from Curbemeister, uses this feature to
    plot the gamuts of any suitable profile.
    --
    Mike Russell
    www.curvemeister.com
     
    Mike Russell, May 26, 2005
    #12
  13. indide_designs

    Guest

    Mike Russell wrote:
    >
    > wrote:
    > [re conversion of RGB to CMYK]
    > > By "Certain colors can not be reproduced in CMYK", I assume you meant
    > > out of gamut colors between color spaces. But some colors on a monitor
    > > that are not flagged as oog in PS still won't show up in Epson inkjet
    > > prints, after all the fuzz of monitor calibration and custom printer
    > > profiles. Is the PS oog detection faulty? Aside from oog detection,
    > > are there other ways to find out what colors won't print?

    >
    > There are a host of technical issues with out of gamut colors. The ICC
    > standard defines a "tag" that defines out of gamut colors.
    >
    > But the gamut tag is not always included in a profile, and when it is, it is
    > not used consistently by different manufacturers. So programs like
    > Photoshop generally ignore the gamut tag, perferring instead to rely on
    > color values calculated using the profile. One rather simple method of
    > doing this is to see if converting the color back and forth gives the same
    > result. For example, converting the out of gamut RGB (255,0,0) into CMYK
    > and back again gives a different RGB value from the original (RGB(199,3,3),
    > say). When this happens, that particular color is considered out of gamut.


    When soft proofing with Epson inkjet profiles, I can see an obvious
    shift in some monitor colors that won't be reproduced correctly in
    prints. Yet these colors are not detected as oog by PS. Does that mean
    the Epson profiles do not include the gamut tag? Is there a way to find
    out if a profile has a gamut tag?

    > There are other, better, ways to calculate out of gamut colors, and I
    > suspect the Photoshop engine uses one of those. BTW, out of gamut colors
    > are calculated by the engine, not by Photoshop itself. You can experiment
    > by telling Photoshop to use the Microsoft (or Apple) ICM engine in your
    > color prefs.
    >
    > But this only says that the profile generates unique values for a particular
    > color, and not whether the printer itself yields a different color when that
    > color and colors close to it in value are printed. Given the limits of the
    > current technology, particularly the poor support of the gamut tag, the only
    > way to tell for sure is to print the colors, and scan them with a photometer
    > or spectrophotometer. You can't even look at the darn things to tell if
    > they are different, because the differences are very subtle. No wonder
    > there is so much confusion, so much being spent on calibration equipment,
    > and relatively little result to show for it.


    Confusing is an understatement. Take the above situation as an example.
    My working space is rgb, soft proofing is in rgb, the file is sent to
    the Epson as rgb, the Epson driver will print in six cmyk inks.
    Meanwhile when soft proofing in PS, there is an Info Palette option to
    show Proof Color (italic rgb). What is the meaning and use of Proof
    Color? IOW, what is the relationship between Actual Color and Proof
    Color? Adobe must have thought that Proof Color has some use.

    > But most of us realize that our inkjet printers do a pretty good job. If
    > printers were watches, at this time in history they would accurate to a
    > couple of minutes a day. As was the case with watches, there are those who
    > require (or think that they require) greater accuracy, and are willing to
    > pay for it. As has happened already with watches, at some point in the
    > future printers will probably be extremely accurate and inexpensive. Until
    > that happens, most of us can easily live comfortably with less than
    > perfection.


    > Here are some more thoughts that I hope will give pause to some of you who
    > have bought into the conventional wisdom that larger gamuts are always
    > better than smaller ones.
    >
    > 1) What was the last out of gamut object you photographed? Blue sky, for
    > example, is not really that saturated, particularly near the horizon where
    > most of our cameras are pointed. Red objects - even bright bird plumage, is
    > a far cry from RGB(255,0,0). For most of us, the answer to the question is
    > "none".
    >
    > 2) RGB is not the last word in color spaces. Consider pure yellow objects,
    > which are not all that uncommon, are not at all well represented in the RGB
    > color space. For most monitors RGB(255,255,0) is brighter, but less
    > saturated, than CMYK(0,0,100,0) will be in print. Magenta and cyan objects
    > have similar problems.
    >
    > Recent monitor developments have made an end run around this problem,
    > resulting in the apparent miracle of a monitor capable of displaying the
    > Adobe RGB color space directly. This is achieved by filtering the RGB
    > phosphors to create purer colors, sharpening and stretching the three
    > corners of the RGB gamut (to see these shapes, check out Curvemeister's
    > Labmeter, a free gamut plotter image). But even purity can have its limits,
    > and the extermely sharp spectral characteristics of these monitors are
    > bringing a new problem, viewer metamerism, to the forefront. With this
    > latest advance, color consistency is literally in the eye of the beholder.
    > The color on these monitors simply look funny to some people.
    >
    > 3) Consider that a larger gamut such as Adobe RGB sacrifices color
    > gradation. An Adobe RGB image uses a smaller number of color values to
    > represent the same range of colors than an sRGB image. This issue may be
    > addressed by working in 16 bits, but there is a more serious problem to
    > working in a large gamut space: the high probability that someone else may
    > look at your image on an sRGB monitor and conclude that your work is too
    > drab. Both of these issues - particularly the second one - are reasons for
    > working in sRGB, particularly since real world objects such as pure blue sky
    > do not come close to exceeding the gamut of even sRGB.
    >
    > 4) Print has a much smaller gamut than a CRT. The big money in photography
    > is still in printed images, whether they be published images, or wall-sized
    > art.
    >
    > My suggestion, as always, is to trust what you can see and verify, and not
    > spend too much time or money on getting your "watch" to run within one
    > second of correct. Todays color and printer technology is excellent, and
    > most of us - the vase majority - can live with its imperfections, provided
    > we understand them, and base our understanding on common sense.
    > --



    Very well put. For hobbyists like myself, getting print colors "close"
    to monitor colors is all we need. But some colors I described above can
    be WAY off. That is like having a watch that is accurate to the minute
    most of the time, but can be off by hours some of the time. Very
    frustrating.

    Will Crockett is another who choose sRGB over other wider gamut space as
    working space:

    http://shootsmarter.com/infocenter.html
     
    , May 26, 2005
    #13
  14. indide_designs

    Guest

    Tacit wrote:
    >
    > In article <>, wrote:
    >
    > > By "Certain colors can not be reproduced in CMYK", I assume you meant
    > > out of gamut colors between color spaces.

    >
    > Yes, that's correct. For example, pure RGB blue ain't never going to be
    > reproduced in CMYK.
    >
    > > But some colors on a monitor
    > > that are not flagged as oog in PS still won't show up in Epson inkjet
    > > prints, after all the fuzz of monitor calibration and custom printer
    > > profiles. Is the PS oog detection faulty? Aside from oog detection, are
    > > there other ways to find out what colors won't print?

    >
    > Photoshop's out of gamut detection is remarkably good; however, there
    > are many issues which may confound trying to identify out of gamut
    > colors on an inkjet print.
    >
    > For starters, Photoshop's out of gamut detection uses the current CMYK
    > setup; if this is incorrect (for example, if you're using a SWOP
    > profile, intended for printing presses, rather than a profile for your
    > inkjet printer), Photoshop will show you the colors out of gamut for a
    > printing press, not your printer.
    >
    > Second, when you print CMYK to an inkjet printer, the printer's drive
    > software converts from CMYK to RGB, then back to the printer's own CMYK.
    > So the data the printer sees may not match the data Photoshop sees,
    > since the printer driver's CMYK->RGB conversion isn't very good.
    >
    > Third, consumer-grade inkjet printers don't even use pure primary CMYK
    > inks at all. In particular, the cyan ink that an inkjet printer uses
    > isn't really cyan; it's too blue. This is done to get more vibrant, more
    > saturated colors that consumer users like, at the expense of color
    > accuracy. Professional inkjet printers use inks closer to CMYK
    > primaries, and are easier to calibrate for press-accurate CMYK output.


    Agreed with all your points. Please see my response to Mike's post to
    see what's been bugging me.
     
    , May 26, 2005
    #14
  15. indide_designs

    Mike Russell Guest

    wrote:
    [re working spaces, versus print spaces]

    > Confusing is an understatement. Take the above situation as an
    > example.
    > My working space is rgb, soft proofing is in rgb, the file is sent to
    > the Epson as rgb, the Epson driver will print in six cmyk inks.
    > Meanwhile when soft proofing in PS, there is an Info Palette option to
    > show Proof Color (italic rgb). What is the meaning and use of Proof
    > Color? IOW, what is the relationship between Actual Color and Proof
    > Color? Adobe must have thought that Proof Color has some use.


    The meaning of actual and proof color, at least, is well defined. Actual
    color refers to the numeric pixel color values of your image. Proof color
    displays actual color values after converting from your working space to the
    last selected proof color space, as specified in View>Proof Setup. It's
    like a one pixel "Convert to Profile", with the target color space set to
    your proof space.

    The use and intent of Proof color is another issue. Photoshop is first and
    foremost a tool, and providing a motivation for each feature is in the
    documentation is difficult, and not necessarily Adobe's responsibility.

    One very powerful application for viewing proof color values is for
    verification of profile color operation. This is useful for people who
    produce color profiles, or who otherwise concern themselves with the input
    and output numeric values for conversions made by a particular profile. For
    example, set one info palette to actual", and another to "proof". Then move
    the cursor around your image to compare color values and look for trouble
    spots. Instead of saying "my colors look different", you could say that your
    epson color profile converts sRGB(255,0,0) to EpsonRGB(254,12,5).

    Another use would be to preview what CMYK color values would be produced for
    various parts of your image, without having to first convert to the proof
    CMYK color space.

    > Very well put. For hobbyists like myself, getting print colors "close"
    > to monitor colors is all we need. But some colors I described above
    > can be WAY off. That is like having a watch that is accurate to the minute
    > most of the time, but can be off by hours some of the time. Very
    > frustrating.


    Yes, or even a watch that simply stops some of the time. I'd be interested
    in exactly which colors look funny on your printer, how your printer is set
    up, etc. One quick think to try is to compare whatever profile setup you
    are using now with the Color Enhance setting for your printer. Terra cotta
    red, for example, happens to be a tricky color for some Epson papers.
    --
    Mike Russell
    www.curvemeister.com
     
    Mike Russell, May 26, 2005
    #15
  16. indide_designs

    Guest

    Mike Russell wrote:
    >
    > wrote:
    > [re working spaces, versus print spaces]
    >
    > > Confusing is an understatement. Take the above situation as an
    > > example.
    > > My working space is rgb, soft proofing is in rgb, the file is sent to
    > > the Epson as rgb, the Epson driver will print in six cmyk inks.
    > > Meanwhile when soft proofing in PS, there is an Info Palette option to
    > > show Proof Color (italic rgb). What is the meaning and use of Proof
    > > Color? IOW, what is the relationship between Actual Color and Proof
    > > Color? Adobe must have thought that Proof Color has some use.

    >
    > The meaning of actual and proof color, at least, is well defined. Actual
    > color refers to the numeric pixel color values of your image. Proof color
    > displays actual color values after converting from your working space to the
    > last selected proof color space, as specified in View>Proof Setup. It's
    > like a one pixel "Convert to Profile", with the target color space set to
    > your proof space.


    This part is easy to understand.

    > The use and intent of Proof color is another issue. Photoshop is first and
    > foremost a tool, and providing a motivation for each feature is in the
    > documentation is difficult, and not necessarily Adobe's responsibility.
    >
    > One very powerful application for viewing proof color values is for
    > verification of profile color operation. This is useful for people who
    > produce color profiles, or who otherwise concern themselves with the input
    > and output numeric values for conversions made by a particular profile. For
    > example, set one info palette to actual", and another to "proof". Then move
    > the cursor around your image to compare color values and look for trouble
    > spots. Instead of saying "my colors look different", you could say that your
    > epson color profile converts sRGB(255,0,0) to EpsonRGB(254,12,5).


    Here's when things get complicated. How close should the Proof Color be
    relative to the Actual Color in a "good" profile (for a rgb printer like
    Epson)? If there is a big mismatch, what can you do about it?

    For the Epson printer that uses six/eight inks, I think the problem just
    gets more complicated. Since the Proof Color rgb values will eventually
    be used to "drive" the inks by the Epson driver, the numbers may be very
    different from the Actual Color values.

    > Another use would be to preview what CMYK color values would be produced for
    > various parts of your image, without having to first convert to the proof
    > CMYK color space.


    If the working space is cmyk and the output is cmyk press printing,
    comparing Actual and Proof colors may make more sense. I'm unfamiliar
    with press printing, so some guess here.

    > > Very well put. For hobbyists like myself, getting print colors "close"
    > > to monitor colors is all we need. But some colors I described above
    > > can be WAY off. That is like having a watch that is accurate to the minute
    > > most of the time, but can be off by hours some of the time. Very
    > > frustrating.

    >
    > Yes, or even a watch that simply stops some of the time. I'd be interested
    > in exactly which colors look funny on your printer, how your printer is set
    > up, etc. One quick think to try is to compare whatever profile setup you
    > are using now with the Color Enhance setting for your printer. Terra cotta
    > red, for example, happens to be a tricky color for some Epson papers.


    It took me a while to come up with a reasonable example. Here's a small
    crop from an unedited scan file to illustrate a typical monitor/print
    mismatch problem. The crt monitor has been calibrated with a Spyder.

    http://pg.photos.yahoo.com/ph/roger8231/detail?.dir=/3b76&.dnm=7ac1.jpg&.src=ph

    In proofs and prints, A and B are the problem areas, while C and D are
    acceptable for my purpose. Using A as an example, here are its rgb
    values before and after Soft Proofing with the Epson 2200 hwm profile
    (first with perceptual intent, and then with relative colorimetric
    intent.)

    Actual Color = 57_30_24
    Proof Color (perceptual intent) = 86_19_6
    Proof Color (relative colorimetric intent) = 64_7_12

    While the perceptual intent proof is more pleasing, A and B lose their
    shadow impact. The relative colorimetric intent proof is just dull, but
    the B area retains some shadow. Test prints (convert to profiles and no
    color management at Epson) match pretty closely to the proofs.

    If you have the 2200 (or 1280) Epson profiles, do you consider the
    proofs and prints of this image "closely matching" the monitor image?
    (i.e. This is as good as it gets with these printers and profiles.)
    What, if anything, do the Proof Colors tell you? Is the problem because
    A and B are out of gamut (no oog is detected)? Can a custom profile fix
    this kind of problem? For those who don't use profiles, can their prints
    of this image match what they see on their monitors? If yes, how?
     
    , Jun 1, 2005
    #16
  17. indide_designs

    Mike Russell Guest

    wrote:
    [re relation of original RGB image color to a particular printer's proof
    color]

    > Here's when things get complicated. How close should the Proof Color
    > be relative to the Actual Color in a "good" profile (for a rgb
    > printer like Epson)? If there is a big mismatch, what can you do
    > about it?


    The actual color values are less important than the answers to these
    questions:

    1) Is the profile of excellent quality?

    2) How much are adjacent color values changing in the proof when you modify
    your original color values slightly. I think of this as the amount of
    "elbow" room a particular color has in the proof, and it amounts to how
    close to the edge of the printer's gamut your desired color is?

    3) Is the printer color space unduly distorted in the particular area where
    your color falls, or are there other topological problems with the lightest
    or darkest areas of the gamut?

    Using Photoshop's info palette's proof color feature only gives you an ant's
    eye view of what is occuring with your color. Curvemeister's free download,
    LabMeter, addresses this issue more globally, since you can locate the Lab
    coordinates of your color on the image, and see if you are located in a
    "fat" part of the gamut, or off in "blue Florida" (my name for the dark
    peninsula of blue that most RGB and CMYK profiles have in the darker range
    of the profile, known as the threequartertone), or the yellow Aleutians.
    Some bits of gamut are even located on little "islands", or bits of ragged
    coastline, and although these tend to be areas of very low luminance, it
    strikes me that these areas would not be very useful in actual printing.

    > For the Epson printer that uses six/eight inks, I think the problem
    > just gets more complicated. Since the Proof Color rgb values will
    > eventually be used to "drive" the inks by the Epson driver, the numbers

    may be
    > very different from the Actual Color values.


    If you are interested in this sort of thing, Epson has described how this is
    done, at least in part, in their patent documents. You're correct that it
    is complex, but the nice thing is there is no need to worry about it.

    >> Another use would be to preview what CMYK color values would be
    >> produced for various parts of your image, without having to first
    >> convert to the proof CMYK color space.

    >
    > If the working space is cmyk and the output is cmyk press printing,
    > comparing Actual and Proof colors may make more sense. I'm unfamiliar
    > with press printing, so some guess here.


    The numeric comparison is of more interest to folks who work more directly
    with the numbers, though the numbers can be of interest to anyone who is
    curious about the inner workings of profiles. They are also helpful for
    discussing printing problems, as in this case.

    [re actual colors that don't print well]

    > It took me a while to come up with a reasonable example. Here's a
    > small
    > crop from an unedited scan file to illustrate a typical monitor/print
    > mismatch problem. The crt monitor has been calibrated with a Spyder.
    >
    >

    http://pg.photos.yahoo.com/ph/roger8231/detail?.dir=/3b76&.dnm=7ac1.jpg&.src=ph
    >
    > In proofs and prints, A and B are the problem areas, while C and D are
    > acceptable for my purpose.


    I'm getting the following values from your scan for these locations. BTW
    There is no embedded profile in your image, so I'm assuming you are using
    Adobe RGB because the image looks dull as dishwater in sRGB. The color
    numbers were copy/pasted using software, I did not type them in manually
    :):

    A RGB(54,29,25)
    B RGB(59,31,19)
    C RGB(25,20,16)
    D RGB(88,56,36)

    and for use with LabMeter, which is a free curvemeister download, here are
    the same colors in Lab:

    A Lab(13,18,12)
    B Lab(15,19,16)
    C Lab(5,1,2)
    D Lab(27,19,24)

    C is relatively dark, and requires less color accuracy because the eye is
    more forgiving of very dark colors. D is the brightest of the four colors,
    so it occupies a wider part of the gamut, giving it a better shot at
    retaining its hue for a good match. This turns out to be important for this
    particular profile.

    > Using A as an example, here are its rgb
    > values before and after Soft Proofing with the Epson 2200 hwm profile
    > (first with perceptual intent, and then with relative colorimetric

    intent.)
    >
    > Actual Color = 57_30_24
    > Proof Color (perceptual intent) = 86_19_6
    > Proof Color (relative colorimetric intent) = 64_7_12


    These are very interesting numbers, for a couple of reasons. First, they
    are "fatter" (brighter and more saturated) than the corresponding original
    numbers. I was very puzzled by this, and downloaded some of the Epson 2200
    profiles from www.epson.com. The profiles that I downloaded are horrible.
    If the yours are similar, this explains the problem.

    The quick check is to look at the individual channels - red is not bad, but
    green and especially blue show very prominent banding and light spots in
    just the areas you are mentioning. In particular, the shadow as you
    approach the chin rest gets lighter, instead of darker, for these channels!

    You may verify this for yourself more systematically as follows. Lay down a
    gradient - I recommend setting dither off and smoothness to zero in the
    profile editor to get a clean gradient. Then convert it using one of the
    Epson profiles. Look at the green and blue channels, and you will see
    substantial banding near the threequartertone, right in the area of concern
    for the colors you mentioned.

    The free Profile Plotter action at Curvemeister.com will do this more
    systematically, producing a graph of all three channels. If your profile is
    like mine, you will see some rather nasty bends in the red and green curves,
    right where the trouble is.

    > While the perceptual intent proof is more pleasing, A and B lose their
    > shadow impact. The relative colorimetric intent proof is just dull,
    > but the B area retains some shadow. Test prints (convert to profiles and
    > no color management at Epson) match pretty closely to the proofs.


    Yes, the banding is worse in the RelCol profile.

    > If you have the 2200 (or 1280) Epson profiles, do you consider the
    > proofs and prints of this image "closely matching" the monitor image?
    > (i.e. This is as good as it gets with these printers and profiles.)
    > What, if anything, do the Proof Colors tell you? Is the problem
    > because A and B are out of gamut (no oog is detected)?


    The proof colors tell me that the profile is making the shadows
    substantially lighter, even adding an apparent glow to the darkest areas of
    the image that simply does not belong there.

    > Can a custom profile fix this kind of problem?


    Yes, it can, however there are substantial hazards to using a scanner, as I
    suspect happened for these particular profiles, rather than a
    spectrophotometer or good quality colorimeter to generate a printer profile.
    Shadows are easy to print accurately, but difficult to scan accurately, so
    scanner based profile packages are fighting an uphill battle when it comes
    to generating accurate profiles for the darkest 25% of the image.

    > For those who don't use profiles, can their prints
    > of this image match what they see on their monitors? If yes, how?


    Use of a good profile is great - and you may use the profile plotter action
    to judge this for yourself and ensure that there are no sharp bends or other
    irregularities in the shadows. Otherwise, try using the PhotoEnhance mode
    of the Epson driver and see if you like the results. Although not profile
    based, it is capable of good results provided you stick to the Epson inks
    and the paper types that are mentioned in the driver.

    If your profiles do not show this problem, it is still possible that
    something else is wrong - if you believe this is the case, perhaps you could
    make the icm files available somewhere and I could look at them more
    closely. Otherwise, all you need to do is get hold of some better profiles
    (or use PhotoEnhance) and you'll be on the air with better looking shadows.

    Thanks for an interesting problem.

    Labmeter is located here:
    http://www.curvemeister.com/tutorials/LabMeter/index.htm

    and the Profile Plotter is here:
    http://www.curvemeister.com/downloads/profileplotter/index.htm

    The RGB and Lab values were converted to text using the Curvemeister plugin.
    This functionality, and many more color related features, is supported in
    the demo version:
    http://www.curvemeister.com/downloads/cmdemo/index.html
    --

    Mike Russell
    www.curvemeister.com
     
    Mike Russell, Jun 2, 2005
    #17
  18. indide_designs

    Guest

    Mike Russell wrote:

    > Thanks for an interesting problem.


    I should thank you for taking your time to look at the image and follow
    up in detail. If this "interesting" problem is caused by these popular
    Epson printer profiles, I should not be the only one troubled by it.
    Since no one else is participating in this thread, can we continue off
    line? That way, I can send you images and profiles, etc. I do have many
    comments and questions.
     
    , Jun 3, 2005
    #18
  19. indide_designs

    Hecate Guest

    On Wed, 01 Jun 2005 13:36:40 GMT, wrote:


    >Here's when things get complicated. How close should the Proof Color be
    >relative to the Actual Color in a "good" profile (for a rgb printer like
    >Epson)? If there is a big mismatch, what can you do about it?
    >

    Here's where you go wrong and why you're having trouble fully
    understanding it. The Epson printer isn't an RGB printer. No printers
    are RGB printers. Printers are CMYK devices. The fact that you need
    to send the Epson an RGB files doesn't change the fact that it prints
    CMYK.

    Hope that helps,

    --

    Hecate - The Real One

    Fashion: Buying things you don't need, with money
    you don't have, to impress people you don't like...
     
    Hecate, Jun 3, 2005
    #19
  20. indide_designs

    Mike Russell Guest

    wrote:
    > Mike Russell wrote:
    >
    >> Thanks for an interesting problem.

    >
    > I should thank you for taking your time to look at the image and
    > follow up in detail.


    > If this "interesting" problem is caused by these
    > popular Epson printer profiles, I should not be the only one troubled
    > by it.


    There are hundreds of people scratching their heads over poor profiles.

    > Since no one else is participating in this thread, can we
    > continue off line? That way, I can send you images and profiles, etc.
    > I do have many comments and questions.


    I'd like to continue this discussion in some sort of public forum. I think
    the problem is of wide interest, in spite of the lack of other people
    participating, I think there are some lurkers who are benefitting. The
    curvemeister yahoo group is another possibility. The group has an upload
    area for files and images, and there are other experienced people who will
    chime in.
    --
    Mike Russell
    www.curvemeister.com
    ..
     
    Mike Russell, Jun 4, 2005
    #20
    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. Mpaintnz

    Unsubtle: 12bit RGB to 1bit CMYK

    Mpaintnz, Aug 10, 2003, in forum: Photoshop
    Replies:
    1
    Views:
    272
    Mike Russell
    Aug 10, 2003
  2. murmur

    help converting RGB photo to CMYK

    murmur, Aug 27, 2003, in forum: Photoshop
    Replies:
    3
    Views:
    2,489
    murmur
    Aug 28, 2003
  3. Martijn Postmus

    Batch-script needed for transforming images

    Martijn Postmus, Sep 2, 2003, in forum: Photoshop
    Replies:
    1
    Views:
    255
  4. Otari Kew
    Replies:
    3
    Views:
    239
    Tom Thackrey
    Sep 17, 2003
  5. StuTheDog

    Transforming multiple images to same size

    StuTheDog, Jul 8, 2006, in forum: Photoshop Tutorials
    Replies:
    2
    Views:
    280
    Harry Limey
    Jul 9, 2006
Loading...

Share This Page