sRGB versus Adobe RGB

Discussion in 'Digital Photography' started by Dave Addison, Nov 24, 2005.

  1. Dave Addison

    Dave Addison Guest

    What are the pros and cons of using sRGB versus Adobe RGB when shooting
    - I'll be using a Canon 350D (Rebel XT)?

    Any work on the images will be done in Photoshop CS, with different
    purposes for the output - sometimes web, sometimes print.
     
    Dave Addison, Nov 24, 2005
    #1
    1. Advertisements

  2. Dave Addison

    Bill Hilton Guest

    Dave Addison writes ...
    AdobeRGB has a wider gamut (more colors), sort of like buying the
    Crayola box with 256 crayons instead of 192 crayons when you were a
    kid. Your camera can capture more colors than sRGB (actually more than
    AdobeRGB), if you've calibrated your monitor it will display more
    colors than sRGB (if the monitor is any good) and if you're printing on
    a 6,7 or 8 color inkjet or a Chromira or LightJet laser printer then
    AdobeRGB is better because these printers have a wider gamut than sRGB.

    sRGB is much better for the web, where most applications don't support
    color management profiles and most people's monitors are uncalibrated.
    It's also the color space of choice for the on-line print shops or
    places like Wal-Mart or Costco, where the assumption is they are
    receiving mostly jpegs from people who don't understand color
    management. If you're printing on a 4 color business-class inkjet
    (instead of the 6-8 color 'photo' printers) sRGB is fine too.
    Just use AdobeRGB and when you are working on a file for the web make a
    copy and do Image > Mode > Convert to Profile to convert to sRGB. If
    you don't do this the saturated colors get dumped when you make the
    jpeg for the web. Basically you can go from AdobeRGB to sRGB easily
    but you can't reverse the process in any meaningful fashion, so it's
    better to use AdobeRGB out of the camera if you're going to print with
    a better printer (or if you just want to keep as much info in your
    files as possible).

    Bill
     
    Bill Hilton, Nov 24, 2005
    #2
    1. Advertisements

  3. Dave Addison

    Celcius Guest

    Hi Dave!
    Bill has answered your question very thoroughly. I might add, that if you
    use Photoshop to edit your images and are using Adobe with color settings
    (Shift + Ctrl +K in Windows) at Adobe RGB (1998), you're better off using
    Adobe RGB with your camera.
    Marcel.
     
    Celcius, Nov 24, 2005
    #3
  4. Dave Addison

    Mardon Guest

    Despite having read numerous articles on colour management, I'm still
    struggling with understanding colour workflow; especially how the
    monitor's hardware device profile fits in. Can you help? I use a
    digital still camera with a device profile of Adobe RGB (1998). I use
    Photoshop CS2 and Adobe RGB (1998) working space. I embed an sRGB
    profile when saving for the web, but I embed Adobe RGB (1998) for
    print images because I have access to a printer that handles Adobe RGB
    (1998) profiles. I also have a digital camcorder with an NTSC device
    profile and edit the video in Adobe Premiere Pro with an NTSC working
    colour space. The final video product, of course, is viewed on an NTSC
    TV. I have dual monitors and calibrate them with a Spyder2Pro
    colorimeter. What I do NOT understand is which colour profile I
    should chose with the Spyder software in order for everything to look
    correct on my monitors. Spyder allows me to choose several different
    profiles that can be loaded into my video card but Adobe RGB (1998) is
    not one of them. They include NTSC, sRGB, 2.2-6500, and several
    others. How do I know which of these profiles to use when calibrating
    the monitors? How can a monitor calibrated to NTSC look correct when
    I'm using the Adobe RGB (1998) working colour space in CS2 or how can
    a monitor calibrated to sRGB look right when I'm using Premiere with
    an NTSC working colour space. I guess I just don't get the monitor
    calibration and how it relates to everything else. Can you help?
     
    Mardon, Nov 24, 2005
    #4
  5. Dave Addison

    Mark Roberts Guest

    For monitor viewing, sRGB is better. For print AdobeRGB is better.

    I generally work in AdobeRGB or ProPhoto RGB for printing and archiving.
    When I make a smaller image for web display I convert to sRGB as part of
    that process. (To convert profiles, open in Photoshop and use "Convert
    to Profile" under the EDIT menu.)
     
    Mark Roberts, Nov 24, 2005
    #5
  6. Dave Addison

    eawckyegcy Guest

    You can buy 192 crayons of length 1, or you can buy 256 crayons of
    length 0.75 (or whatever the ratio actually is). You can't ignore the
    encoding.
     
    eawckyegcy, Nov 24, 2005
    #6
  7. Dave Addison

    Bill Hilton Guest

    Mardon writes ...
    This article explains the underlying basics well ...
    http://www.creativepro.com/story/feature/13605.html ... Bruce is one of
    the authors of the "Real World Photoshop" series and "Real World Color
    Management", where I learned this stuff (which I agree isn't intuitive,
    but makes sense once you understand the basics behind it).
    The colors are represented by RGB numbers in a file but each "device"
    (monitor, printer, scanner) represents these numbers differently. So
    you generate a monitor profile by displaying known RGB values on the
    screen and the puck (if you are doing a hardware calibration) measures
    what is actually displayed. The calibration program knows what it
    started with (displays) and now knows what those numbers (colors) look
    like on the screen after it measures them so it creates a translation
    table (actually a matrix for a monitor profile) that "translates" the
    numbers in the image file so the colors look as accurate as possible on
    the screen.

    It's the same for a printer profile only you print a test pattern and
    this is read back by the calibration software to determine which colors
    can be printed accurately (the "gamut") and how to translate the
    numbers in the file to different numbers for the printer to get the
    closest match between what you see on the screen and what is printed
    out. You can turn on 'proof color' in Photoshop's Info palette to show
    what the translated RGB values will be, for example.

    So anyway, what I described is for a single device under a set of
    conditions (monitor profile is most accurate for one monitor under
    similar ambient light conditions, printer profile is specific to one
    printer/paper/ink combo, for example). These are called "device
    profiles" because they are specific to a device and are different from
    "working space" profiles, which are abstractions chosen for a given
    gamut that won't necessarily precisely match any specific device.
    No, the camera doesn't have "a device profile of Adobe RGB (1998)".
    Actually it's hard to create a "device profile" for a digital camera
    because as the light changes colors look different, so you can't
    generate an accurate profile unless it's in a studio setting where the
    lights are identical for each shot, typically a product photographer
    using strobes. "AdobeRGB" is a working space profile, which is
    abstract, not device-specific.
    OK, that works ...
    Instead of embedding an sRGB profile you should convert first from
    AdobeRGB to sRGB. Then you can embed or not as you wish (few browsers
    will see the profile though). If you just embed sRGB into an AdobeRGB
    file the saturated reds etc get dulled down.
    You would typically get better results soft proofing to the specific
    printer/paper/ink profile thru Photoshop. Doing it the way you
    describe simply means the printer software/firmware will convert to the
    internal profile but if you are soft-proofing you have more control and
    can preview it accurately on the screen before printing.
    I think this is just a starting point for calibration, ie, to get the
    right gamma etc. The software will eventually load your
    device-specific monitor profile settings into the video card registers,
    not an abstract working space profile. At any rate, NTSC is probably
    the closest to AdobeRGB (do you have SMPTE as a choice, I think that's
    what AdobeRGB was called in V5 and earlier?). They (Colorvision)
    probably would have to pay a small royalty fee to include AdobeRGB so
    have included something with a similar gamut instead. I'd just email
    Colorvision support and tell them you're using AdobeRGB as your main
    working space so which profile setting do they recommend for this
    initial set-up step ...
    The monitor profile should let you view the colors accurately with any
    color-managed application, you can switch between files with different
    working spaces without a problem. For example I use mainly AdobeRGB
    but also use Ektaspace for film scans and sometimes ProPhotoRGB for
    highly saturated digital camera images (the reds don't clip with this
    much wider space) and sRGB for web images. The four different color
    spaces I use basically define the gamut or range of colors but the
    monitor profile will accurately convert those colors for viewing on the
    screen, at least the ones that are within the gamut or range of colors
    that my monitor can display.
    I think you are pretty close to getting it ... make sure you understand
    the difference between a device-specific profile like Spyder is
    creating for your monitor and the abstract working spaces like
    AdobeRGB, and ask Colorvision which profile they recommend as a
    calibration starting point for your range of working spaces (it's just
    a start point, the actual video card settings are generated by the
    profile software) and you should be fine. If you want to really dig
    into color management the "Real World Color Management" book I
    mentioned earlier is excellent.

    Bill
     
    Bill Hilton, Nov 24, 2005
    #7
  8. Dave Addison

    kctan Guest

    In film photography, we select type of film to use according to the type of
    subject to shoot. We have films for portraiture, landscape, general...etc.
    Different camera and lens also produce its own color rendition; all in all
    is to simulate color of the original scene. When it is in slide or
    transparency form, a lightbox is needed for viewing. Calibrated light in the
    lightbox got to be standardize for consistent viewing among viewers.

    In digital photography, we create digital image file by assigning a tagged
    profile
    (analogous to type of film) accordingly to the type of color rendition based
    on subject to
    simulate. To display this image, we need a calibrated monitor (analogous to
    calibrated lightbox) for consistent viewing among viewers. So we need
    colorimeters like Eyeone, Spyder, Monaco...etc to characterize and calibrate
    the monitor.

    Remember that the digital image file will never change it colors information
    (unless you convert) but can be simulated to many different types of color
    rendition by assigning a known tagged profile (acts like a multipurpose
    film). For
    profile, there are input profile
    (camera and scanner), output profile (monitor and printer) and editing
    profile (Image editing software like PS). Color management means getting all
    the descriptions of all these profiles, calculate the different with
    reference to a known standard color model to produce a standard and
    consistent color visual.

    All monitors cannot display that much colors contain in adobeRGB color
    space, therefore no point to have that profile. Choose the type of color
    space profile you are displaying the video. Don't confuse with monitor
    profile and working space profile. They are 2 different profiles that needed
    to be known by color management engine to do calculation. Remember, there
    are many profiles interacting in color management and not just one.
     
    kctan, Nov 24, 2005
    #8
  9. Dave Addison

    Mardon Guest

    Thanks for taking so much time to provide a very complete answer.
    Thanks also for the book suggestion. I think it's an understatement
    to call colour management "less than intuitive! ;-)

    As you suggested, I've emailed Pantone about which monitor Target to
    choose in their Spyder2PRO software. Their quoted response time is 3
    to 4 days. I've included below a copy of how I worded the question. I
    hope it's OK. I've also shown below all of the monitor Target choices
    contained in the Spyder2PRO software. There is no SMTPE. The
    associated help file says, "If there is not an existing Target that
    contains the values that you desire you may create a new Target by
    clicking the 'New' button" but it does not make it clear what to do
    after that. I guess they incorrectly assume I know what I'm doing.

    -----------------------------

    Email to Pantone
    Subject: Which Monitor Target To Chose with Spyder2PRO?
    Question: I am a colour workflow Newbie and I do not understand which
    Target I should choose from the drop-down list on the "Select target"
    page of the Spyder2PRO software. I use a working space of Adobe RGB
    (1998) in Adobe CS2 and the NTSC working space in Adobe Premiere Pro
    1.5. Which target does this mean I should choose for my monitor in
    Spyder2PRO?

    ------------------

    Spyder2PRO software choices for Monitor Target:
    1.8-5000
    1.8-6500
    1.8-Native
    2.2-5000
    2.2-6500
    2.2-Native
    Cineon
    ITU-R Rec. BT.709
    NTSC
    PAL SEACAM
    Print Standard
    sRGB
     
    Mardon, Nov 24, 2005
    #9
  10. Dave Addison

    Bill Hilton Guest

    Mardon writes ...
    I'd be interested in their reply if you can remember to post it to this
    thread ...
    Ah, I think I see now ... after seeing your complete list I'm pretty
    sure all they want you to choose is a gamma and a white point ... I use
    gamma 2.2 and whitepoint 6500, which I'm almost certain is the
    "2.2-6500" choice on your drop down. The 'working space' choices each
    have a default gamma so maybe that's why they included those, in case
    you didn't know which gamma/whitepoint pair to pick. The monitor cal
    software I use asks for gamma and whitepoint, which is why I'm making
    this assumption. Note that it then calculates a custom white point
    based on the changes you make to the individual RGB guns (CRT), which
    is why I said earlier this was likely just a starting point anyway. At
    any rate, would be nice to see their reply ...
    Yes, but when you finally fight thru all the jargon it all makes sense
    :)

    Bill
     
    Bill Hilton, Nov 24, 2005
    #10
  11. Bill,
    Yes, I agree, but then I come across things like this:
    The printer, graphic arts, adobe, monitor, etc community are still
    using CIE 1931 data, yet newer data, which supposedly is more
    accurate, yet shows different responses, e.g that of 1964:
    http://www.clarkvision.com/tmp/eye-response.1931+1964.v1-w-1.gif

    So if the newer eye response functions are more accurate, what does that
    say for the accuracy of current calibrations and profiles using the old
    profiles? (I'm not expecting an answer, expressing that there is confusion
    at the lowest level data, so at some level one wonders what is color
    accuracy? Then people probably have differences in their color response,
    so it can get really confusing, even when you understand the principles!)

    If anyone knows of a reference to color response variability between
    people, I would be interested.

    Happy Thanksgiving,
    Roger
     
    Roger N. Clark (change username to rnclark), Nov 25, 2005
    #11
  12. Dave Addison

    Dave Addison Guest

    Thanks everyone for the very informative and in-depth replies. Greatly
    appreciated!

    Dave
    ================
     
    Dave Addison, Nov 25, 2005
    #12
  13. Dave Addison

    Mardon Guest

    My wife is about to order "Real World Color Management" online from
    chapters.indigo.ca for a Christmas present. :) There are two versions
    listed: (1) one published by Prentice Hall, September 2004 and (2)
    another published by Peachpit Press, January 2003. Both are by Bruce
    Fraser. Is it correct to assume that only the 2004 edition is needed?
     
    Mardon, Nov 25, 2005
    #13
  14. Dave Addison

    Bill Hilton Guest

    Mardon writes ...
    I have the 2003 editon from Peachpit (early adapter, what can I say)
    but the field is changing rapidly so I'd go with the 2004 edition now.

    Bill
     
    Bill Hilton, Nov 25, 2005
    #14
  15. Dave Addison

    Bill Hilton Guest

    Roger Clark writes ...
    Hi Roger,

    I'm more the engineer, interested in applying the theory in practical
    ways, while you're the scientist, interested in the underlying theory I
    guess :) At any rate in the "Real World Color Management" book this is
    touched upon briefly. My understanding is the 1931 tests involved
    measuring the actual reponses of a few hundred people to various colors
    and light intensites to come up with things like "delta-E" (smallest
    difference in color that the eye could discern) and the LAB space
    (range of all colors the human eye can see). According to the RWCM
    book, in 1964 they re-ran some tests "that used larger color samples
    that illuminated a wider angle of the fovea and found a slightly
    different tristimulus response". Or in English, maybe the LAB space is
    a bit bigger than the 1931 tests would imply. The author just mentions
    this in passing, saying "It's rare that you'll encounter the 10 degree
    (ie 1994 test) observer" data. (The 1931 tests were 2 degree).

    One of the Photoshop gurus, I think JP Caponigro, wrote that LAB
    describes 100% of the colors the average eye can see, saturated film
    can capture maybe 75% of those colors at most, a good monitor can
    display perhaps 50% and most 4 color CMYK presses can print about 25%
    of those colors. The Real World book doesn't say this but my
    assumption is that since practical color management is mainly concerned
    with accurately translating colors within this 50% range of colors you
    can view and print that the 1964 tests aren't very relevant, since they
    seem to simply expand the LAB space slightly at the boundaries. But
    that's just my interpretation of it.

    The ICC is actually a consortium of graphics companies that create the
    specs for the color management workflow. They have a website at
    http://www.color.org/ and if you log on you'll see "Ask Phil" where you
    can send in questions about things like this. Would be interesting to
    hear their response as to why they use the 1931 CIE model instead of
    the 1994 data so if you "ask Phil" let me know what he says.

    Bill
     
    Bill Hilton, Nov 25, 2005
    #15
  16. The 1964 data is 10 degrees and the 1931 is 2 degrees. The eye simply
    reacts differently to color patches that are spread across 10 degrees than
    smaller 2 degree ones.

    The big problem is however not a at the XYZ level. It is quite possible that
    if you take colors from nature, you convert them twice to XYZ (once using the
    1931 data and once using 1964) and then you display or print them correctly
    that you end up with basically the same colors in the print.

    The big problem is at the perceptual level. Things like simultaneous
    contrast. Or what if the viewing environment does not match the environment
    in which the CIE data was obtained, etc. What if you have to reduce contrast
    because the original scene was more contrasty than can be printed.

    And then there is the problem that printing systems are calibrated using
    5000K light but are rarely viewed that way. Even if the print is correct at
    5000K, it may not be at 2800K.
     
    Philip Homburg, Nov 26, 2005
    #16
  17. You can if you use 16-bits/channel.
     
    Philip Homburg, Nov 26, 2005
    #17
  18. Dave Addison

    eawckyegcy Guest

    In the absence of error bars, who can say whether or not these are even
    _different_, let alone one being "better" than another.
    Not exactly what you had in mind, but I'll mention that those who had
    cataract surgery prior to 1980 or so can see into the ultraviolet. At
    least my mother could, and so can this guy:

    http://starklab.slu.edu/humanUV.htm

    I'm not sure if modern cataract surgery gives it's victim's, er,
    patients the same ability today.
     
    eawckyegcy, Nov 26, 2005
    #18
  19. Dave Addison

    Markeau Guest

  20. Dave Addison

    Colin D Guest

    I had cataract surgery on both eyes about 6 years ago, done through a
    4mm slit cut into the top of the eyeball, the natural lenses were
    removed with a phacoemulsifier (an instrument which emulsifies the lens
    with ultrasonic vibrations and removes the result by suction) and
    preformed soft plastic lenses fitted, designed to restore focus at
    infinity. My immediate reaction to my vision when the eye-shields were
    removed next day, was that everything looked so white, really really
    white, where before the view was - in hindsight - yellowish. The
    opthalmologist said that was due to two things, natural lenses tend to
    yellow as one gets older, and they appear to have a UV cut-off like a
    skylight filter, whereas the replacement lens has neither.

    I found that I could easliy tell a monochrome monitor from a vga monitor
    when both are switched off, by the color of the screen. Vga monitors
    are gray, while mono monitors are distinctly bluish - but my wife cannot
    see any difference.

    This gives rise to some royal arguments between us about color balance
    in the prints I produce - her line is that my vision sees too far into
    the blue, where hers is natural. (I haven't an answer to that one!)

    It's interesting, though.

    Colin D.
     
    Colin D, Nov 26, 2005
    #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.