Skip to main content
Participant
January 14, 2009
Question

Assign- Convert - Proof Colors

  • January 14, 2009
  • 51 replies
  • 15476 views
In Photoshop, I can choose to assign or convert to profile. This seems straight forward, I think.

I was given images in ProPhoto RGB color space (8 bit). Just as a test, I did two things. I created a Proof setup that used Adobe RGB and Peceptual Rendering intent. Then, I also went to Assign Profile. Both cases the image was darker and lacked detail (given the smaller color space)...

My working space for RGB is Adobe RGB with perceptual rendering intent.

Next I turned the proof colors off and I converted the file to Adobe RGB. In this case, the color changed very little (was not dark at all).

I thought I understood that if you assign a profile, the appearance on screen will differ (and the numbers will stay the same), If I convert, the numbers actually change.

question: Why would the image look very dark when I was assigning the profile, but look the same as the original when converting?
    This topic has been closed for replies.

    51 replies

    Participating Frequently
    January 16, 2009
    >> I beg to differ (about Conversions degrade color information)...There is nothing "degraded" about a well-done conversion

    Hmm... you may be right, Marco.

    I recall I got that bit of info in one of my forum discussions with Bruce Fraser or from one of his books, but I hope I am not confusing people with my recollections.

    To a lay person (like me) who is trying to conceptualize the basic theory that Converting a higher-gamut space like Adobe RGB or ProPhotoRGB to smaller-gamut spaces like a press CMYK or web sRGB space would have to clip or compress and/or stir up some colors to happen I guess that would be the basis of my point.

    In any case, it is probably a good idea for most people to make as few profile Conversions as possible based on simple common sense principles like if we Convert AdobeRGB to sRGB -- and back again -- we have probably lost (degraded) some color information that we can never get back.

    Let alone if we Assign (or Assume) sRGB to an AdobeRGB file -- hammer it back into shape in CMYK SoftProof mode and save it back out with no profile ;)

    How that theory may apply to the actual rocket science is over my pay grade and more likely suited to your technical expertise.

    +++++++

    BTW
    >> Converting to a profile changes the color numbers in the file, but keeps the appearance unchanged (except for gamut mismatches, in which case the rendering intent decides how the out-of-gamut colors in the source are mapped to the destination).

    That totally rocks...
    January 16, 2009
    >To a lay person (like me) who is trying to conceptualize the basic theory that Converting a higher-gamut space like Adobe RGB or ProPhotoRGB to smaller-gamut spaces like a press CMYK or web sRGB space would have to clip or compress or stir up some colors I guess that would be the basis of my point.

    Sure, perfectly reasonable. Yet, my point is that it is the very purpose of a conversion to optimize an image for its intended output. No "degradation" there: only the achievement of a "best fit" within the circumstances.

    At the same time, a converted image, specially one converted to a standard CMYK color space from an RGB space like AdobeRGB or ProPhotoRGB, may well have undergone some substantial reduction of its color gamut (though not necessarily). That's why one should go back to the source file for further conversions to other color space destinations, in order to retain all the color gamut in the source that fits into any given destination.

    It makes no sense, for example, to cut the source's gamut to SWOP and then convert that converted file to Sheetfed, only to risk losing some of the color brilliance that could have been retained in Sheetfed in a conversion made directly from the source image.

    >That totally rocks...

    That's very kind of you. Thanks.
    Ramón G Castañeda
    Inspiring
    January 15, 2009
    csullivan,
    >I do have a good understanding of Assign and Convert.

    Not wanting to appear rude, condescending or disrespectful, but like Marco, I really question the veracity of that statement. :/
    January 16, 2009
    Just as "G. Ballard" says re: "csullivan":

    >if you know (and have Honored) the Source Space in Photoshop, WHY are you still talking about ASSIGNING a different profile to it it is kind of a pointless experiment once you understand the point.

    Exactly. What is the point, after converting, of assigning a profile that's different from the one that the file has been converted to?

    Unless one is still unclear about the distinction...in which case it's better to admit it.
    Participating Frequently
    January 15, 2009
    MORE ABOUT ASSIGN PROFILE

    Bruce Fraser wrote in a previous Adobe forum:

    (You could change profiles a thousand times via Assign Profile and the image would not degrade) "Inasmuch as the numbers in the file wouldn't change, this is true. But it would display incorrectly, and convert to any other space incorrectly, so it's fair to say that while the integrity of the data hasn't been compromised, and you can rescue the file by assigning the correct profile, for all practical purposes, it's hosed."
    Participating Frequently
    January 15, 2009
    >> once you have a "translation" (conversion) and you...re-translate back or to a different language (profile) that something may be lost in the translation....

    EXACTLY

    Conversions degrade color information...I like to keep my color conversions to an absolute minimum.

    For example:

    Capture in Camera RAW format
    Open in Adobe Camera Raw 16-bit ProPhotoRGB
    Edit in Adjustment Layers
    Save MASTER FILE in that format
    Open and resize and Convert to the Target Space
    Save the production copy

    That leaves my workflow with ONE conversion and fills my production copy with as much color information as possible.
    January 16, 2009
    >Conversions degrade color information...

    I beg to differ. Conversions are
    i necessary
    in order to create the best fit of an image with its intended output medium (a press or other). There is nothing "degraded" about a well-done conversion to the output color space -- quite the contrary.

    Though it's true that, once converted, such a file is of limited usefulness for
    i further
    conversions, for which it's far better to go back to the source image and start over.

    >I like to keep my color conversions to an absolute minimum.

    Well, I see no reason to convert more than just once, from source to intended output -- once for each output directly from the source.
    Participating Frequently
    January 15, 2009
    >> I do have a good understanding of Assign and Convert.

    if you know (and have Honored) the Source Space in Photoshop, WHY are you still talking about ASSIGNING a different profile to it it is kind of a pointless experiment once you understand the point

    if you want to change the Source Space (Source Profile) -- CONVERT

    if you want to "Soft Proof" a Target Profile/Space in Photoshop -- use View> Proof SetUp: Target Profile
    >> how Photoshop uses previews

    The Color Management System CMS CONVERTS the SourceSpace/File/Profile into MonitorRGB (the custom "calibrated" monitor profile) and PROOFs the color accurately on the screen -- this happens on the fly, automatically behind the scenes in Ps (we cannot turn it off since version 6).

    The file (the document Source Space) is independent of Photoshop:

    The Color Management System CMS, Photoshop, ONLY uses the monitor profile for one thing: To PROOF source file on the monitor (the monitor profile has zero to do with how the file prints).

    The Color Management System CMS, Photoshop, ONLY uses the printer (target) profile Print Space for one thing: To PROOF source file on the paper (the printer profile Print Space has zero to do with how the file looks on the monitor)...www.gballard.net/psd/cmstheory.html
    Participant
    January 15, 2009
    Thank you to both of you for technical and non-technical. The analogy of language is very good and would also support the theory that once you have a "translation" (conversion) and you attempt to re-translate back or to a different language (profile) that something may be lost in the translation....

    I do have a good understanding of Assign and Convert. Your information just supports that. What I did not know is how Photoshop uses previews.

    What I thought was that if a had a ProPhoto RGB file and assigned Adobe RGB that the preview would reflect in gamut colors, which in turn, would look very bad given the differences in size of color space. Convert (From ProPhoto), would actually change the pixel values to fit inside of the gamut of the destination profile (Adobe RGB).

    Why is it that:
    If all colors in the image from ProPhoto fit inside of the Adobe RGB profile to begin with, that the preview would chaange dramatically. Is it due to the size of the gamut of ProPhoto (i.e. the distance between color values)?

    and just for clarification of the preview mode. If I do assign Adobe RGB profile in this instance, why do I not get the choice of rendering intents? I am assuming that the preview is in gamut colors (and others are clipped?)

    thanks
    January 15, 2009
    > I do have a good understanding of Assign and Convert.Yes, I believe a know the difference.

    Are you sure? Your earlier explanation felt a bit muddled:

    > In simple terms, I believe that assigning a profile, assigns a working "set of colors" (color space) for the person who is correcting. It does not change the color value of the pixels, but is simply a visual reprsentation of the image when confined to the assigned colors. and by embedding it, the next user down the pike will know which defined "color set" was used to make corrections. Converting actually will map colors from one space to another and uses the actual destination spaces pixel values in the file.... right?

    I'm not sure. In
    i actually simple
    terms, the difference is this, as others have pointed out and I'm just restating:

    b Assigning a profile
    leaves the color
    i numbers
    in the file unchanged, but changes their
    i appearance
    (their
    i "meaning"
    in device-independent terms).

    b Converting to a profile
    i changes the color numbers
    in the file, but
    i keeps the appearance unchanged
    (except for gamut mismatches, in which case the rendering intent decides how the out-of-gamut colors in the source are mapped to the destination).

    I think it's important to understand this distinction and be able to state it as clearly as possible.
    Participating Frequently
    January 15, 2009
    I somehow missed that thread, Ramón, but here is one of the best explanations I've read:

    Eric Chan - 9:30am Nov 21, 08 PST (#71 of 82)
    Camera Raw Engineer

    I have not read the whole thread, but the easiest way to think of it is in terms of the language analogy.

    Tagging/assigning is identification.

    Converting means translation while preserving meaning.

    Here's an example.

    Suppose you hear a stream of words coming over the radio while driving your car. You don't understand it. Your friend sitting next to you says, "Oh, that sounds like Korean." That's tagging, or assigning. Your friend is identifying what the language is. There's no conversion/translation happening. Your friend may be correct, or incorrect. (For all you know, the real language might actually be Japanese.) Regardless, the central point is that this is what tagging/assigning means.

    But now suppose you actually want to understand what those words mean (i.e., what the radio program is about). Then somebody has to translate it for you. This is "conversion." You ask your friend to translate the "Korean" to English, a language you do understand. What you hear from your friend are words in English, which are different from the original words coming over the radio (which are in Korean). But the meaning is the same (**). Going back to color management, "conversion" means that in general the underlying numbers change during the conversion process, but the color appearance stays the same.

    Eric

    (**) This assumes your friend actually knows Korean and translated it correctly.
    Ramón G Castañeda
    Inspiring
    January 15, 2009
    csullivan,

    The only time you use assign is when some moron hands you an untagged profile and you need to make an educated guess as to which color space said moron was using when he/she created and saved the file.

    CONVERT to profile strives to preserve the colors while changing the numbers. Assign maintains the numbers and changes/wrecks the colors.

    Maybe this old post will help you:

    Ramón G Castañeda, "Definitions for profiles" #3, 16 Nov 2008 5:17 pm
    Participating Frequently
    January 15, 2009
    ASSIGN VERSUS CONVERT TO PROFILE
    >> Why would the image (change when) assigning the profile, but look the same as the original when converting?

    That behavior is 100% to be expected. I like to explain the difference between Assign Profile and Convert To Profile like this: http://www.gballard.net/psd/assignconvert.html

    The only time most users ever need to touch Photoshop> Edit> ASSIGN PROFILE is if the Source File is Untagged (and we need to PROOF it and/or edit it in a color-managed application like Photoshop).

    On the other hand, Ps> Edit> CONVERT TO PROFILE is a feature most of us use on a daily basis for example, Converting our high-gamut RGB working space to a target CMYK space, an sRGB web space, or a different space (profile) to hand it off.

    As a hard general rule, always USE EMBEDDED PROFILE first, then CONVERT TO PROFILE if it is not in the desired color space (profile).

    PROOF SETUP

    Photoshop> View> Proof SetUp: Profile
    AKA "Soft Proofing" is generally used to preview a target device profile (like a specific Printer/Paper/Ink profile, or a specific CMYK press profile).

    In most cases, Photoshop's Soft Proofing feature is very confusing in the theory department unless we understand why we are using SoftProofing.

    In this case, I would first get the Assign-Convert theory nailed down...and use Proof SetUp as sparingly as possible.

    BRUCE FRASER wrote the book(s) on Proof SetUp (Soft Proofing in Photoshop) http://www.creativepro.com/articles/author/127446
    Participant
    January 15, 2009
    Yes, I believe a know the difference.

    In simple terms, I believe that assigning a profile, assigns a working "set of colors" (color space) for the person who is correcting. It does not change the color value of the pixels, but is simply a visual reprsentation of the image when confined to the assigned colors. and by embedding it, the next user down the pike will know which defined "color set" was used to make corrections. Converting actually will map colors from one space to another and uses the actual destination spaces pixel values in the file.... right?

    If this is the case....
    I still ask the question above