RELATED MATTERS

Assigning vs. Converting Color Spaces

In Photoshop you can alter color spaces through assignment or through conversion. These are two very different processes. When you assign a different color space to an image, the values attached to each pixel remain the same, but the way they are interpreted changes. Thus, suppose you have an image with an sRGB profile attached. If you assign the Adobe RGB profile, the sRGB values for each pixel will remain unchanged, but they will be interpreted as Adobe RGB values in determining colors. If, for example, one assigns an Adobe RGB interpretation to the sRGB fire-engine red triple (250, 20, 10), the result will be a much brighter, more vivid red. Thus, the appearance of the image will change as you assign different profiles, but the the RGB values will not. Assignments can only be made within the same color model, say one RGB space to another RGB space.

When you convert from one profile to another, by contrast, the values attached to each pixel actually change, and their interpretation changes to match. So, if you convert from sRGB to Adobe RGB the values attached to, say, fire-engine red pixels will actually change from the sRGB values corresponding to fire-engine red (250, 20, 10) to the Adobe RGB values corresponding to that same fire-engine red (215, 27, 19), but, at the same time, the profile name will change to indicate how to interpret the altered value appropriately.

It should be noted that there could be some change in appearance when converting spaces, particularly when going from a large- to a small-gamut space, since some colors in the larger space have no equivalent in the smaller. Just what this change will be depends on whether the image contains problematic colors and, if it does, the rendering intent employed, which is explained below. However, as a matter of practice, with an image lacking large areas of strongly saturated colors, you will be hard pressed to see any difference on your screen display or your print as you convert from one working-space to another. A significant change in appearance can occur if you convert color models (which is possible), say from an RGB space to a CMYK space.

Sometimes you may get an image that is untagged – an unspecified color space. With the proper preference settings, Photoshop (and any good image editor) will notify you of this situation when you first open the document and give you options for proceeding. You would rarely want to opt to convert it at the outset to your default working space. Rather, opt to open it "as is" (without color management). It may look quite weird initially, and that is all right. Go now and assign different profiles to it until you find the one that makes it look right. This will allow you to identify which set of RGB values is present on that image. The "missing profile" will often be sRGB, so this is a smart choice to try first. At this point you can either leave that profile assigned to the image, or you can convert to another color space. Before converting, however, you must first assign the color space you found to be correct, and then convert it. The first assignment provides the profile that properly interprets the colors corresponding to the values initially on the image, and the subsequent conversion changes those values (and its profile) to whatever color space you wish thereafter.

Gamut

Color spaces differ greatly in the range of colors they contain (their gamut). The LAB working space is very broad, containing roughly 97% of all visible colors.  Similar percentages for the more popular working spaces are ProPhoto RGB (91%), Beta RGB (69%), Adobe RGB (50%), and sRGB (35%). Photo-quality inkjet printers may have color spaces that can be broader than, say, Adobe RGB in some color dimensions and narrower in others. Different computer monitors have widely varying gamuts. Most are quite narrow, similar to sRGB, and others, like so-called wide-gamut monitors, have gamuts that are as broad as Adobe RGB. Monitors may have gamuts that are broader than your inkjet printer in some dimensions and narrower in others. It is clear, then, that as we change color spaces, going from the camera to the image editor or raw processor, to the monitor, to the printer or web, we are often unavoidably changing gamuts and so encountering colors in one space that may or may not exist as the same color in another.

Note that the sRGB color space has a significantly narrower gamut than the other major spaces, and this means that some colors may be (likely will be) clipped when you use it or convert to it. Once done, these colors are lost can never be restored, and if you've chosen sRGB as your camera's color space, the jpegs it produces (but not the raw files) will be permanently color-stunted from birth. As we have seen, this may not be a problem if one's only intent is to produce images for the web, but if you intend processing for eventual printing on your own photo-quality inkjet printer, a broader space, such as Adobe RGB or ProPhoto RGB, may be more suitable – although many people are perfectly happy with sRGB no matter what. Adobe RGB is quite popular for effecting a wider gamut and, as a practical matter, it is an excellent choice. ProPhoto RGB is even broader, but contains many colors that cannot be reproduced on any monitor or printer, which is potentially dangerous. A space this wide is really most appropriate if one is shooting raw and processing in 16-bit mode. It has far fewer advantages if one is dealing with an OOC jpeg that started its life with a narrower gamut, Adobe RGB at best, and an 8-bit depth. Many people like to process raw files in ProPhoto RGB (it's LR's working space) because it helps maintain a fuller set of colors during the various processing transformations.

Rendering Intent

Both ProPhoto RGB and Adobe RGB have gamuts that contain colors that are beyond what some printers or monitors can reproduce – these are the so-called out-of-gamut colors. So some color alterations must inevitably happen in the transition from image to display or print. Just how this alteration is carried out is what is meant by Rendering Intent or simply Intent. There are several possible intents, but two are relevant for photographic purposes: Relative Colorimetric and Perceptual. To understand the difference between these two, consider the following. Suppose we have two color spaces, a larger one that encompasses a smaller one. With great oversimplification, but good pedagogy, picture these as two concentric circles with different radiuses and a common center representing their white points. The space with the larger radius represents the greater gamut, so part of its color set lies outside the one with smaller gamut.

When using the Relative Colorimetric intent to reduce the larger to the smaller, the white points (the common center) are first made the same. Then all the colors depicted by points lying in the intersection of the two circles (the colors common to both spaces) remain essentially unchanged, while all those lying outside the smaller space (the out-of-gamut colors) are simply moved to the point nearest to them on the perimeter of the smaller space. The "impossible" colors of the broader space are thus truncated to their closest "possible" color in the smaller space. With perceptual intent, the reduction is done more nearly as if you were to shrink the larger circle down into the smaller one, the colors collapsing inwards towards the white center. In this shrinking, some of the colors that are common to both gamuts may move only slightly, if at all, while others may be moved quite a bit; the idea is to try to keep the overall "perception" of the color relations "pleasing," whatever that is supposed to mean.

Just how different the application of the two intents will be depends very much on the image – on how many colors are out of gamut and how far they are out. With some images there will be very little difference between the two, and with others the difference can be great. And when the difference is great, it can be difficult to predict which will produce the image more to your liking. Soft proofing, discussed below, can be of some help here, but often you will need to compare prints made with each intent to make a final judgement.

The previous discussion applies essentially the same any time you convert from one color space to another. If an Adobe RGB image in Photoshop, for example, is displayed on a monitor that has a gamut that differs greatly from Adobe RGB, your monitor's profile along with a rendering intent helps make this conversion of spaces work. Likewise, when you convert working spaces in your image editor, say from Adobe RGB to sRGB or vice versa, an intent is applied to determine the way the conversion takes place. You can choose this intent in Photoshop's Conversion Options portion of the Color Settings menu.  As to which to pick, if you go by published recommendations, you might as well flip a coin.  Relative Colorimetric tends to preserve color better, and it is the one I like, but . . .


Soft Proofing

Soft proofing allows you to see on your monitor an emulation of how an image would look as a print from your printer. We have seen how your monitor has a profile that allows colors coming to it to be translated to be displayed correctly. A soft proof allows a selected printer's profile to intervene to "correct" the monitor's profile so that what is displayed on the screen is made to look like (emulate) what would be coming from your printer. Sometimes this works well, and sometimes it is misleading, depending on how good the monitor and printer profiles are, the appropriateness of the room lighting, the brightness setting of the monitor, and numerous other factors. Clearly soft proofing provides one way of assessing which rendering intent is more suitable when printing a given image. In Photoshop, you select the printer profile to use for soft proofing in the Proof Setup item in the View menu.

Soft proofing not only allows you to view different rendering intents but also to emulate the paper whiteness (provided your printer profile is both printer- and paper-specific). This is useful since the paper has considerably less brightness and dynamic range than the monitor. Soft proofing can also allow you to see the effect of compensating for the black point. Papers are not able to produce blacks as black as they can be in the image file, so, without compensation, a range of the darkest shades in the image will simply be collapsed into the darkest black possible on the paper. Black-point compensation attempts to make the image's black the paper's black and taper the remaining shadows appropriately to retain good relative definition. In Photoshop it is usually recommended to use black-point compensation when using the Relative Colorimetric rendering intent; with the Perceptual intent, it makes no difference since black-point compensation is applied regardless of the black-point checkbox setting.

If your monitor is well calibrated and your monitor and printer profiles are good, soft proofing can be a real boon in getting a rough idea on your monitor of how your printed image will look. So, after you have put the finishing touches of post-processing on an image and have it looking great on your monitor, and before you consider printing it, save a copy of the file in that state so you can return to it later if needed. Then (re)open and duplicate the image in the image editor and turn on soft proofing for the duplicate. Your great image will now, no doubt, look somewhat disappointing and dull. Some people recommend turning your head as you turn on soft proofing so you don't see the immediate contrast. This is actually a pretty good idea. You can now make adjustments to the soft-proofed version (tonal, saturation, etc.) while toggling between it and the original – or put the two versions up side-by-side – to try to make it look more like the original. You'll never be fully happy, but you do the best you can. Then size/resample, sharpen (if needed), convert to 8-bits (if needed – some printers take 16-bit files), and print – being sure, of course, to use your printer's profile.


NOTES

1. An earlier version of this work appeared as a multi-part post on the dPreview D300 forum in 2008.  It benefited greatly from comments made by members at that time.  I wish to express my gratitude to Detail Man for the application of his fine-tooth comb, and well-known attention to detail, to the current version.  I alone, however, am responsible for this content and any possible erroneous constructions.

2. I call this a walkthrough because I intend it to be relatively brief and to the point. It is not intended to, nor can it, substitute for the well known and substantially longer treatises on the subject by Bruce Fraser and Andrew Rodney. There is also a classic font of information on color spaces by Bruce Lindbloom, whose entire site is worthy of exploration. My hope here is to provide a reasonable overview of the ballpark while being sure to touch all the bases without going too far out of the baselines.

3. This is also commonly called embedding the profile. However, there is a distinction that can be made between tagging and embedding. Tagging simply fills an appropriate EXIF entry with the name of the relevant profile. Most cameras do this with the chosen jpeg space, sRGB or Adobe RGB.  Some processing software, like Lightroom, actually embed the entire icc profile as part of the meta data in their exported file. Different image-creation devices behave differently in this regard. Some embed, some tag, some do both, and some do neither. When an image only has a tag, it is up to the image-viewer to supply the relevant profile information, which it may or may not be able to do. When an image has neither profile nor tag, many image-viewers assume sRGB, but some just punt. 

4. As part of the process of color determination, most major raw processors automatically apply input camera-profile information. Camera-specific default profiles are provided for most cameras, but custom profiles may also be made and installed by the user. Lightroom and ACR, for example, use "Adobe Standard" profiles for most cameras, with specialized choices for some, all of which can be found in the Camera Profile pull-down menu of the Camera Calibration pane. Many people find it advantageous to create custom profiles for their particular camera. Different raw processors provide specialized information for creating and installing camera profiles. LightRoom and ACR users can make use of the Adobe DNG Profile Editor or the ColorChecker Passport software for this purpose. Some information for creating these profiles can be found here.

5. Some third-party processors, such as LR, ACR, and Silkypix, also transparently apply any lens-distortion corrections (rectilinear and possibly also chromatic aberration) that are included in the file's meta data.

6. There is a significant interplay between appropriate monitor brightness and the viewing-room ambient light. Many people keep their monitors too bright – possibly, but not always, because the ambient room light is too bright. This is not only tiring on the eyes, but it can cause troubles when printing. One frequently hears the complaint that prints are too dark: the image looked fine on the monitor, but not in print. This has likely occurred because the monitor's excessive brightness results in a screen image that looks fine even though the image file has been inadequately brightened for the printer. It's worth experimenting with the monitor's brightness level during calibration to find one that best suits what goes to the printer. This is likely to result in a monitor brightness that at first seems low, but you will soon get used to it and, likely, grow to love it. I like mine at 90 cd/m^2 or even less. So does my printer – but you will want to find settings that work for your situation.

7. Photographic prints have significantly less dynamic range than a monitor, upwards of 3 to 4 stops less. So there is simply no way that a printed image can have the same impact as what you see on your monitor. Darks will not seem as dark and highlights will not seem as bright. And, if you're using a wide-gamut monitor, the printed colors, even at their best, can look significantly less vibrant than the screen versions. But that's an unfair comparison; they will look fine when you get to "see them on their own." So, use your profiles to get the best from your printer that you can, and wait a while after printing to fully evaluate the effort.  Give time to let the monitor image be forgotten, and be sure to view the printed image under good light because what dynamic range a print has is due to reflection.