Adobe RGB V sRGB

Discussion in 'Digital Photography' started by Russell, Jun 20, 2005.

  1. Russell

    Colin D Guest

    Agreed, but if the color space provided by the RGB filters in the camera
    did not provide the colors to fill the chosen colorspace, then using
    that space or gamut is pointless.

    Consider a camera that has blue and green filters, but no red filter.
    The resulting image would have no red component other than a possible
    reduced response from the slope of the green filter, which would then be
    rendered as green. Such an image could not fill even a sRGB gamut, let
    alone a wider one. If the filters in camera cannot supply all the
    colors in a given color space, then it's pointless using that space. or
    so it seems to me.

    Colin D.
    Colin D, Sep 16, 2005
    1. Advertisements

  2. Russell

    John Bean Guest

    But not harmful either. Someone suggested that ProPhoto
    somehow is "worse" than Adobe RGB, and that's simply not
    No point. Lets consider reality instead, and the colour
    gamut of any Bayer sensor doesn't fit in either sRGB or
    Adobe RGB.

    Less clipping occurs if you use ProPhoto RGB when converting
    raw images. If you're unconvinced then a lot of expert
    opinion says this is so. Perhaps you haven't read Bruce
    Fraser on this subject, to name just one.
    John Bean, Sep 16, 2005
    1. Advertisements

  3. Russell

    Mike Russell Guest

    Not to single out John, but I think beginners should be very careful about
    separating wheat from chaff in this particular thread.

    The "expansion" of ProPhoto RGB can have serious consequences, particularly
    for 8 bit per channel images. ProPhoto RGB, as with any very wide gamut
    space, can result in unacceptable noise, and posterization of the image. In
    16 or 32 bit per channel this is not the case, however not all of us will
    choose to work at that bit depth, particularly when no clear advantage to
    working in ProPhoto RGB has been demonstrated.

    Because of it's immense gamut, ProPhoto RGB also produces artifacts, in some
    situations, when converting to Lab. Try it yourself - create a white to
    blue gradient and convert it from ProPhoto RGB to Lab - see the black band
    at about the three quartertone? This occurs in 8 as well as 16 bit images.
    I'm not so sure. Adobe RGB is actually quite large, for an RGB space, and
    it may well be sufficient for many of the sensors out there.
    Fraser is a great guy, and I have read his books, but why is it necessary to
    appeal to authority?

    Seeing is believing. If ProPhoto RGB has a decisive advantage, it should be
    easy to come up with a photograph in which use of ProPhoto RGB results in a
    markedly better image - printout, screen, what have you, than the same image
    exported in Adobe RGB or sRGB.

    Again, not to put John on the spot here, but in all seriousness, I'm
    genuinely interested. Does anyone have such an image? For example, a raw
    file where the ProPhotoRGB image contains detail or color that goes away
    when the same image is exported as Adobe RGB?

    I suspect I will wait a long time before being presented with such an image,
    so I will say this: Given the great number of new ideas that are being
    tried right now, it's a safe bet that many of them are simply wrong. Lack
    of an image demonstrating a particular principle is a good indication that
    that principle is not going to pan out, in the long run.

    Until a photograph that demonstrates the superiority of ProPhoto RGB
    appears, I think those of us who work in Adobe RGB and sRGB may rest easily,
    whatever the "authorities" may say.

    Indeed, even if one or two images do turn up, I believe it is the case that
    for the vast majority of photos, ProPhoto is overkill, and 16 bits per
    channel is overkill. Toolmaker that I am, I say this with respect for those
    who take excellent photographs, and who believe otherwise.
    Mike Russell, Sep 16, 2005
  4. Russell

    Hecate Guest

    And some disadvantages, as I've already tried to point out to him.
    And I find it at least interesting that semi-pro and pro cameras
    usually provide AdobeRGB. :)
    Because, I suspect, he thinks that he's read the book and understood
    it. I've read the book too and he hasn't. He's making the assumption
    that what Fraser says applies whatever the situation whereas Fraser
    makes clear that isn't the case.

    And anyway, just because Fraser says something doesn't make him 100%
    right all the time. Like any other human, he has his biases. OTOH, I
    usually find him pretty much on the money :)
    Precisely. Though, of course, as I've done before, I'll have to
    disagree about the 16bit comment :)


    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, Sep 18, 2005
  5. Russell

    John Bean Guest

    It's not an appeal Mike, it's a reference to an accepted
    authority - which I'm clearly not - who offers a point of
    view similar to the one I was making. Do you never quote
    such references?
    John Bean, Sep 18, 2005
  6. Russell

    Mike Russell Guest

    I find experiments and examples much more convincing than simply telling
    someone that a third person feels a certain way.

    But, for the moment, I'll break that pattern, and accept Bruce as an
    authority. Here goes:

    I'm certain - enough to lay money on it - that Bruce has never provided, in
    all his books, articles, and posts, a raw image that does demonstrably
    better in ProPhoto RGB than in Adobe RGB.

    If you believe he has, please point me to the image. But if you agree that
    Bruce has not provided such an example, I suggest that this as authoritative
    evidence, by your own lights, that no such image exists.
    Mike Russell, Sep 18, 2005
  7. Russell

    John Bean Guest

    Mike, you sound like a zealot who has something to prove,
    especially in your disagreements with Bruce Fraser. I'm not
    playing your game.

    FWIW, I agree entirely that it's a bad idea to switch
    between Lab and RGB when using very large RGB spaces, and I
    entirely disagree that 16-bit is an overkill. With that in
    mind, and the fact that at the start of the thread I stated
    I work with 16-bit ProPhoto RGB conversions from raw, I have
    seen no evidence to indicate that this space is in any way
    harmful to the data in the way that was being suggested.

    I've been working this way since before the time that Fraser
    started recommending ProPhoto RGB and my comments in this
    thread have been based on my personal experience, not by
    "misunderstanding" Fraser's books as seems to have been
    suggested (not by you). Each to his own I guess, but
    condescention and "prove it" challenges don't strengthen
    your argument one little bit.
    John Bean, Sep 18, 2005
  8. Russell

    Mike Russell Guest

    It should be simple, then, to provide an example image that benefits from
    the ProPhoto 16 bit treatment..
    Mike Russell, Sep 18, 2005
  9. Russell

    Ray Fischer Guest

    Just to add fuel to the fire: Photoshop now has some support for
    32-bit images. Why the hell would that ever be useful given that
    cameras capture only 12 to 16 bits? It's handy when combining multiple
    images taken with different exposure settings. If you have a really
    high contrast scene it allows you to bracket the photo and then
    combine the images. You can then adjust the resulting 32-bit image
    without losing data at the extremes.
    Ray Fischer, Sep 18, 2005
  10. Russell

    Mike Russell Guest

    This format is called HDR and was created in the course of work by people
    like Paul Devevec, and others, to represent luminance maps, which are used
    to create realistic lighting effects. .

    There are situations requiring hibit data, and I'm familiar with those from
    my experience with medical imaging, which generates tremendous dynamic
    ranges for CT data. But this does not mean that every image, or even most
    images, benefit from being in hibit.
    Yes, this is how it is done, and there are any number of interesting
    applications. I predict that some day, perhaps relatively soon, there will
    be a camera with firmware that creates an HDR image directly, and in the
    more distant future a detector that captures floating point images directly
    in hardware. Again, does this mean that every image captured today will
    benefit from this specialized ability? Obvioulsy not.
    I have seen some stunning shots resulting from combining several exposures,
    in terms of dynamic range, which you are talking about, and in terms of
    space with images taken at different focus planes. Here are the best
    examples I've seen from a technical and artistic standpoint:

    BTW - the same technique may be spread across time as well. When
    multi-exposure technique is applied to images taken at different times, you
    get the well-known time lapse shot. If the detector plane is shifted
    laterally, you can get rid of aliasing using multiple exposures. BTW -
    these are applications of concepts discussed in Robert Cook's ground
    breaking paper on distributed ray tracing, published in the 1984 issue of
    Siggraph, and still bears reading today:

    Clearly, each of these multi-exposure techniques can produce images that
    could not be produced in any other way. Does this mean every image should or
    can be done by combining multiple images? No, and I'm sure that neither
    you, nor any of the practitioners who create such photos would make this

    In this the creators of multi-exposure images differ from the small number
    of folks who believe that every high quality image requires ProPhoto/16,
    apparently doing so with not a single actual image to demonstrate that this
    is the case.
    Mike Russell, Sep 18, 2005
  11. Russell

    Bill Tuthill Guest

    I have seen ProPhotoRGB images that showed more dynamic range
    than 8-bit images (can't remember whether AdobeRGB or sRGB)
    but it's likely that the greater dynamic range could have been produced
    just as easily in 8-bit using the Shadow/Highlight tool or something.

    Theoretically, isn't it worse to switch 8-bit images between AdobeRGB
    and sRGB, than to stick with one or the other? Doesn't each switch
    result in conversion problems?
    Bill Tuthill, Sep 18, 2005
  12. Russell

    Mike Russell Guest

    Hi Bill,

    I'll bet you two adult sized life jackets that you're remembering wrong, and
    this was an issue with 16 bit versus 8 bit, and not ProPhoto. ProPhoto has
    a larger color gamut, but it is a gamma 2.2 space with the same dynamic
    range as Adobe RGB.
    Theoretically, yes, and I certainly wouldn't convert back and forth for no
    reason, but the conversion problems are negligible for operations like
    saving for web.
    Mike Russell, Sep 19, 2005
  13. Russell

    Mike Russell Guest

    I was mistaken. ProPhoto is gamma 1.8, and Adobe RGB is 2.2. This can make
    a difference in shadow detail for certain images, with a slight advantage
    going to ProPhoto.

    But Bill still has to find the image to get his two life jacket reward :)
    Mike Russell, Sep 19, 2005
  14. Russell

    Bill Tuthill Guest

    You're right that it might have been entirely 16 bit versus 8 bit, but I'm
    certain it was a ProPhotoRGB versus some old-style colorspace. Unfortunately
    the images have gone off the Web, otherwise I would have provided URL.

    Off topic: The new MTI Big Buoy is a nice life jacket!

    As I alluded to, the new Shadow/Highlight tool in Photoshop CS 8
    seems to do an excellent job of increasing the visible dynamic range.
    There are other ways of doing this, but Shadow/Highlight seems easy.

    Does have a higher signal to noise ratio
    than the newsgroup?
    Last time I checked the latter was high-volume and booooooooring.
    Bill Tuthill, Sep 19, 2005
  15. You can't really disentangle the two issues. If you use an 8 bit per
    pixel space, you want to pick an encoding that actually uses as many of
    the 256 available values as possible to avoid quantization errors. If
    sRGB contains all the colours you care about, then it's better than
    Adobe RGB for that image because the possible colours are more densely
    spaced in colour space. If sRGB doesn't work, but AdobeRGB does, then
    again it's better than ProPhotoRGB if it gives denser sampling and
    better use of the limited number of pixel codes.

    On the other hand, if you're going to pay the cost of 16-bit channels
    anyway, then quantization error becomes a non-issue and you can pick the
    colour space using other criteria - such as making it larger than
    anything you'll ever need for any image. But it's the 16-bit channels
    that allow you to do that.

    Dave Martindale, Sep 19, 2005
  16. Russell

    Mike Russell Guest

    Hi Tut,

    ProPhoto is as old as any other color space. It used to be one of Kodak's
    proprietary color spaces, dating back to at least 1998, when it was called
    ROM RGB and Master RGB. It was designed to contain the gamut of E6 films,
    and all films for that matter. It's an input profile, intended to capture
    all the colors of a negative or positive film. Adobe RGB and sRGB are
    output oriented profiles, centered on colors that are useful for display

    Think of ProPhoto as a giant 6 mpg Hummer, with custom springs so high you
    need to custom weld steps on the side of your car to get in and out. It's
    an impedement, IMHO, for ordinary photographs to use it, and ProPhoto needs
    a 16 bit garage.

    If you're driving an sRGB Prius, or an Adobe RGB 98 Mustang, you get
    excellent transportaion, and can park in either the 8 bit or 16 bit garage.

    Having a large garage does not mean you need a large car, nor are people
    with smaller cars necessarily in need of getting a larger one.
    Wow - slick. Pockets too. I'm ashamed now of my bulky dayglo orange ones.
    OTOH, any life jacket is nice once you're in the drink.
    S&H is definitely a hot tool, and I highly recommend that people get
    familiar with it. For those who haven't upgraded to CS2, ts behavior may be
    duplicated by an action
    LOL - yes. I shouldn't laugh, but what can you do. It's actually getting
    bad enough to degrade the usefulness of this group somewhat.

    A good dose of block sender gets rid of most of it. I block nearly anyone
    who responds in kind to a thread cross posted to more than a couple of
    groups. Ignoring individual conversations is probably the best move since
    the names change around.

    What's new, sort of, about this trolling is that the person doing it poses
    as several different people who argue with one another in a way that tempts
    people to intervene, drawing more unsuspecting people into the fray.
    Mike Russell, Sep 20, 2005
  17. Russell

    Bill Tuthill Guest

    Interesting. I did not know all this, but had heard of Master RGB (and
    thought it was something different). Kodak had some great ideas, including
    the YCrCb compression method for original PhotoCD.
    So can I use ProPhotoRGB to carry around soldiers without adequate armor
    against roadside bombs (improvised explosive devices) whilst generating
    handsome profits for General Motors?

    I'll check out for a bit. I'm curious whether
    anybody there likes SaveForWeb. So far it hasn't been discussed much.
    What does "Ducky" mean in the EXIF header?
    Bill Tuthill, Sep 20, 2005
  18. Russell

    Mike Russell Guest

    Kodak had some great ideas, including
    There were a bunch of good ideas in PhotoCD. I think it got smothered
    because Kodak kept the PhotoCD creation software too close to the vest.
    This killed what might have otherwise been a good standard, and perhaps a
    juicy market. Not that it matters now, as the film digitizing market
    LOL, and here'sWide Gamut RGB:
    Save For Web is widely used. For some reason it's not discussed much in
    Great questions today, Bill. I did a quick search - Adobe seems to sign
    it's jpegs with the name "ducky" in the exif header, perhaps as a sort of
    hatched Easter egg. My private theory is that it's to honor "Ducky
    Doolittle", the runner up for the Ole no Moiré image:

    Mike Russell
    Mike Russell, Sep 21, 2005
  19. Russell

    Bill Tuthill Guest

    Despite wide gamut color, that thing is ugly looking!
    Photoshop really isn't a good tool for writing JPEG. Not only does
    SaveForWeb fail to warn about non-SRGB colorspace, it produces 2x2
    chroma subsampling even at the highest (IJG 98) quality levels, which
    seems to me downright dumb. My tentative guideline would be to use
    1x1 above IJG 85-90, and 2x2 below.

    The old-style SaveAs JPEG dialog in Photoshop does not provide
    separate control over chroma subsampling, although maybe that's
    not really necessary. At higher Q levels (7-12) it produces
    JPEG files that aren't any better looking than IJG libraries do,
    but are larger, especially if default Preview isn't turned off.
    At lower Q levels (1-4?) Photoshop does produce good looking JPEG
    compared to IJG, but I'm not sure this is an important thing to do:
    usually photographers want the highest quality possible, and webspace
    is much cheaper than it used to be.

    JPEG Optimizer, free with PaintShopPro, might be better, but I have
    not yet investigated.
    Interesting theory!

    Seems that only SaveForWeb JPEG is enHeadered (is that a word?) Ducky.
    Bill Tuthill, Sep 21, 2005
  20. Russell

    Hecate Guest

    Then again, some of us use Fireworks.... :)


    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, Sep 26, 2005
    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.