FAQs and Tips

(Go back to main Profiles page.)

August 16, 2006 (Revision September 16, 2006)

1) These ICC profiles are Mac and Windows compatible
2) System Requirements?
3) Installation
4) How does one choose which complete sets to use?
5) See two-dimensional gamut plots here
6) How do you use the RGB spaces and their Chroma Variants?
7) Is it OK to convert into the Chroma Variants instead of into the master space? No.
8) Is there a way to apply the effect of a Chroma Variant to just one part of an image? Yes.
9) Questions and Tech Support
10) Check for clipping
11) In Photoshop's Preferences, go to...
12) In Photoshop, to see what you can't see,...
13) It's all simulations in Photoshop
14) In Photoshop, to avoid the default soft proof setup...
15) I developed most of these...
16) What is quantization error?



1) These ICC profiles are Mac and Windows compatible. For optimal cross-platform compatibility between Windows and Mac OS X, the use of the .icc suffix is important. Note that for any ICC profile to be usable with Mac OS 9, it must have a type and creator code, which will always be stripped away if an archive, such as a .zip file, is decoded on a Windows machine, or if it is not zipped with the "preserve Mac contents" option enabled. Therefore to get profiles which are compatible as-is with both OS 9 and OS X, be sure to unzip the profiles on a Mac, either in OS X or in Classic/OS 9. When my profile sets are downloaded and unzipped on Windows, two files will appear for each profile, a big one (the profile) and a little one (the Mac type and creator code info) with the same name but beginning with ._ (dot underscore). In this case, simply delete the small files. Mac OS X can recognize and therefore use a profile if it contains either the type and creator code information, or the .icc suffix.


2) System Requirements? Mac OS 9 or X, Windows 98 or later, Adobe Photoshop 6 or later, or any other application which can both convert from profile to profile and assign profiles to images and which supports color management simulations to show you correct color on screen. If you don't have a calibrated and profiled monitor, you can't get to square one with digital imaging. In that case, I recommend the software that I mostly designed, ColorBlind Prove It! for OS X or Windows, available from www.color.com.


3) Installation:
To use ICC profiles, first copy them into these two locations for Windows (2000 or XP):

C:Windows/System32/Spool/Drivers/Color
and
C:/Program Files/Common Files/Adobe/Color/Profiles/Recommended

or these two locations for Mac OS X:

Library/ColorSync/Profiles
and
Library/Application Support/Adobe/Color/Profiles/Recommended

and these two locations for Mac OS 9 and Classic:

System Folder/ColorSync Profiles
and
System Folder/Application Support/Adobe/Color/Profiles/Recommended

The first location is mandatory for each OS and the second location is useful if you'd like these (or any other) profiles to show up at the top of lists in Photoshop. Putting a profile in both locations will not cause the profile to be shown twice in Adobe's lists because they check for duplicates while making their lists. So you might want to only keep those variants which you are apt to use often in the second location.

If you still use any Classic apps on a Macintosh, be sure to also add the profiles to the old System Folder's ColorSync Profiles folder. You can put an alias for that folder into the OS X profiles folder to avoid the need to move your profiles to the new location or keep copies in two system locations.

When profiles are listed in Photoshop and in most, but not all imaging applications which are ICC-aware, the name of the profile shown in the list is an internal name, not the actual file name of the profile. Usually both names are the same, but when they aren't it can be quite confusing! In OS X, if you double-click a profile, that will launch ColorSync Utility and you will be able to click on the desc tag in the profile and see its internal names (up to three different ones are possible!).


4) How do you choose which complete sets to use?

If you are intent on keeping your pictures in 16-bit per channel form from the source space (camera profile or scanner profile) to the printer space, i.e. never allowing image data that is in the RGB working space to be in 8-bit per channel form, then the simplest and most economical thing to do may be to just use ProPhoto RGB and my complete set built for it, for all your pictures, though a few purple-blues that may be found in transparencies will not fit into ProPhoto. I prefer my DCam 4 master profile to ProPhoto for use with digital cameras when a giant, general-purpose space is desired, but the differences are negligible when the data are in 48-bits, and ProPhoto has the current advantage of being supported by the closed architecture of some of the popular RAW processing applications. For digital cameras, my default recommended space is DCam 3, however, assuming no RAW converter limitation, regardless of initial bit depth intentions.

If your pictures will be in 8-bit form while in any working space, then you have the opportunity to optimize image quality by minimizing image quantization, by using a space which is not overly large for the image or poorly shaped relative to typical image color ranges. In that case, read on to learn more about which of my several spaces will be most useful for you. Or if you already have existing converted and/or edited images in one of my spaces or in Adobe RGB, then you'll want to consider the complete sets for those master profiles as well. Finally, even if you are working with 16-bit per channel image data, you may find that using DCam 3 or even DCam 2 for digital captures, depending on how colorful your subject matter gets, or Chrome Space 100 for scans of transparency film, is preferable to using a giant space like ProPhoto or DCam 4 because it imposes clipping boundaries on lots of color range that will map relatively poorly into printer profiles and may therefore help to provide some discipline with regard to editing and help you to avoid hue shifting and major lightness shifting problems later on.

More Detail (very long, sorry!):

A) If you are making new scans of transparencies, choose Chrome Space 100 and its chroma variant set. If you have important existing scans of transparencies or color negatives in Ekta Space PS 5, chose the chroma variant set for Ekta Space PS 5 and optionally consider adding the Chrome Space 100 complete set, because it can be used to open up deep shadow detail quite noticeably by assigning any of its profiles to the image (not by converting!). This is useful with scans of transparencies because most such scans are short on shadow detail and because you can get shadow detail this way without taking a hit on the histogram of files converted into the spaces of well-behaved printers, unlike relying on RGB curves alone. See this shadow detail comparison.

B) If you are making new scans of color negatives, choose DCam 3 and its chroma variant set as the best all-purpose color neg scanning space, but you can also make good use of others among the five DCam sets, especially DCam 2, for the same reason that DCam users can — the range of colors encoded in a color negative is apt to vary greatly and it's ideal to work in a space that isn't a lot larger than you need, especially if your data may ever be in 8-bit form while in the working space. I have found in my own work over the last several years that scanning color negatives into Chrome Space 100 has worked quite well, so for simplicity and keeping your costs down, consider using Chrome Space 100 for all film scanning. The differences have to do with just when and where you might see clipping in scans of negatives that include very saturated colors. Chrome Space 100 and Ekta Space PS 5 cover a gamut which matches the dye sets of transparency films closely, not the range of colors that we (or digital cameras or color negatives) see in the world, but the transparency spaces are large enough that they contain the great majority of colors in most images from any source. If you are scanning film and you are able and willing to always keep your scans at 16-bits per channel, then you can safely consider using a very large space such as ProPhoto RGB or DCam 4, but your files will be cumbersome and usually take many times longer to work with, depending on available RAM. I have found that if you make your original scan at 16-bits per channel with a very high quality scanner, and you make gross image adjustments while still at 16-bits, that then converting down to 8 bits for all of the finer adjustments has been entirely adequate for fine print quality, while making the processing of my files bearable (450 MB, 16-bit files can still bog down fast machines a lot).

C) If you are making new DCam captures, the five DCam spaces offer a very wide range of choices, covering a range of gamut volumes of more than four-to-one (see the relative gamut volume diagrams here and here and the individual 2D gamut plots with each space's detailed description on the About the Master Spaces and Chroma Variant Sets page, and also on my Gamut Plots page.

My new set of five DCam spaces starts with the very small gamut of DCam 1, which is 11% smaller than sRGB, and it runs up to the huge gamut of DCam 5, which is about 23% bigger than Pro Photo (were it not for gamut clipping by Lab, its relative Lab volume would be greater still). DCams 2, 3 and 4 are all general-purpose spaces. DCam 4 and DCam 5 are recommended for images in 16-bit form only, though this is by no means a hard and fast rule. DCam 5 is simply huge and covers the entire visible gamut, though Lab precludes some of those colors from being used in any ICC workflow of today.

DCam 1 is a specialized space, intended for minimizing the quantization error of images that contain no very strong colors, and which also contain no very light or dark colors other than near-neutrals. As spaces get narrow, they also loose headroom, making light non-neutral colors clip more often. See illustration showing headroom differences between my five DCam spaces. DCam 1 is not intended as a general purpose space for anyone's full range of work, though it will often hold the full range of colors in images without strong colors in them and volume for volume it holds more useful colors than sRGB. For optimal processing quality with such images in 24-bit form, this space is ideal.

DCam 2 is exactly twice the volume of DCam 1, and it's 23% larger in volume than Adobe RGB, but it's much more efficiently shaped than Adobe RGB, yielding perhaps 50% more usable gamut volume. DCam 2 is ideal for people who either tend to make pictures with all-natural colors in them or who avoid very strong colors, and/or who exercise care in RAW conversion to avoid expanding colors beyond those actually present in the scene. It is a conservative, general-purpose design aimed at minimizing quantization error without clipping for people who avoid the most brilliantly colored subject matter (e.g. some of the most colorful flowers, neon signs, exceptionally colorful artificial fabrics, brilliant sunset highlights, colored foils, and the like). Once captured and converted without clipping into DCam 2, images can still be rendered as quite colorful by using DCam 2's chroma variant set, with no danger of clipping as the colors are expanded, and with further gamut reductions if the chroma variants are used to shrink the image's colors, further minimizing quantization error.

DCam 3 is also a general-purpose capture space, thirty percent larger in gamut volume than DCam 2, making it better suited to either a wider range of subject matter, including neon signs, the most brilliant flowers, and the most brilliant artificial fabric colors, paints, etcetera, or for use with somewhat less brilliant colors together with more adventurous RAW conversions. DCam 3 is my default recommended space for general DCam use, in part because it's not extremely large, yet it is big enough for almost anything, especially given the great extent to which it can shrink or expand with the help of its chroma variants. The master space should be big enough to accept your incoming colors without clipping (clipping causes 255's or zero's in non-neutral colors in the file after conversion). DCam 3 isn't as easy to work with (to avoid clipping) as are the two still larger spaces (4 and 5), but it does help to impose a more reasonable discipline onto the photographer, who may otherwise be tempted to move colors about in the various imaging processes in ways that put them far beyond the limits of all monitors, present or future, not to mention all printers, present or future. If a color doesn't fit into DCam 3 because it's moderately saturated but too light, i.e. because there isn't enough headroom in DCam 3, then it won't fit into any printer or display gamut either, so this discipline tends to be useful.

DCam 4 is also a general-purpose capture space, very similar in volume and form to Kodak's ProPhoto, but with my perceptually linear tone curve and slightly different gamut boundaries which I arrived at through my independent analysis, after having started off with a much more different design. Traditionally, color experts have advised that images should be in 16-bit form when used in ProPhoto, and I can't really argue with that contention, however, this is certainly not a black and white issue. Quantization error, though quite real and even obvious in 3D plots of images converted into various spaces, can be very difficult to see in an actual image, and so, although I find it personally offensive to see an image occupying a tiny percentage of its working space, and to know that the number of unique colors in the image may be cut in half by putting them into the larger space if the image is in 8-bit per channel form, I also know that the damage done to the file might never contribute to any visually detectable problem whatsoever. Then again it might. Certainly it is wise policy to avoid unnecessary damage to image files at each of the many steps through which they must pass. But then, one doesn't want to sweat too long and hard over counting the angels that can dance on the head of a pin. The many experiments I have performed have usually revealed a very small increase in graininess when comparing files with smooth, digitally captured skies converted into my larger spaces, compared with my smallest space. This doesn't mean that the differences might not combine with other processing, or under different circumstances, to create more obvious and distressing artifacts, but neither can I guarantee that such circumstances will ever occur. This is part of the reason that I determined that I should provide a full range of tools from which people can choose, given their own idea of what is just right for them. There are simply too many ways to make digitally captured images for me to be certain what will be best for you regarding working space gamut volume, or to conduct all the experiments necessary to be certain of the infinitude of possible outcomes. What I have done is to make certain that the exact range of colors captured for each of the five volume levels in my final set is the very best match to the colors of the world, as processed by a variety of raw converters, with consideration given to the relative importance of avoiding clipping in particular colors. DCam 2 was the most difficult to design, as it seeks to cover everything, without having the gamut to do so. The other designs, to varying degree, flowed from its design, except DCam 5's, which was simply a function of the human gamut as defined by the CIE in Yxy space.

DCam 5 is an unprecedented, giant RGB space, which fully (and just barely) encompasses the entire boundary of the human visual gamut (per Yxy D50 space). Some of the colors that people can see are nevertheless effectively excluded from what can be stored in any RGB working space on account of the limitation that standard encodings of Lab space impose upon the entire workflow because Lab is used by the CMMs (some of the greens and yellows we can see, plus a tiny sliver of blues are outside of the Lab gamut). Still, DCam 5 will hold a lot of colors that DCam 4 and ProPhoto will not. Such colors are nevertheless rare except for very light colors, where DCam 5 excels at avoiding clipping. Still, this extra headroom, as I call it, is all color that can't be replicated in print or on a monitor, and then some, so DCam 5 might be looked upon as a way to process RAW files with negligible clipping on conversion, leaving the problem of fitting colors into smaller gamuts for later. This could be helpful for batch conversions, though my sense of DCam 5 is that its added head room is more than you need, at least when you're doing the white point adjustments well in the RAW conversion, which is important to do since that's the only place it will ever be possible to get the gray balance of your image just right. If you want to use it for batch conversion, DCam 5 is available without chroma variants as part of a master spaces-only deal on my Products page. Data in 16-bit form should be completely safe from quantization damage from being stored in giant RGB spaces, whether the original AtoD conversion in the scanner or camera was 16-bit, 14-bit, or 12-bit. Still, for best quality with data in 8-bit form, don't use a giant space, but do avoid clipping through careful image adjustment.

If there is any reason to worry about DCam 5, it would be on account of its blue primary being located outside the bounds of positive x and y values, making both the y and Y values for blue negative. Given that this is a unique property among profiles that I am aware of being in use today, I must assume that it is possible that this profile, and/or its chroma variants, may prove to not work correctly with some imaging software, though my own testing has shown it to work fine in Photoshop and a few other applications, suggesting that we're OK there.

So, if you generally work with a conservative sense of color, DCam 2 should be ideal if you wish to optimize quality for files that may move into 8-bits per channel form at some point while still in the working space. If you want to cover the full range of Munsell colors and pretty much the full range of colors you are apt to find in nature, including all of the most brilliant flowers, as well as the overwhelming majority of saturated artificial colors, DCam 3 is ideal. If you want even more headroom to avoid clipping contrasty, moderately saturated colors or to encompass both the full range of what your camera can see and what can be printed on any printer (without relying on the chroma variants' ability to encompass those printer colors later on), then try DCam 4. And if you have important images with muted color and fairly neutral highlights, DCam 1 can be used to achieve unusually low levels of quantization of the data in 8-bit per channel files. Finally, if you prefer the open range, DCam 5 will give you unprecedented working room for an RGB workflow. And don't forget that just because a printer can print a color, doesn't mean you need to be able to print that color! The only printer colors that won't fit into DCam 3 are fairly dark, super-saturated bluish greens, which are wildly more saturated than any naturally occurring color in vegetation, for example.

It's apt to not be a simple choice to pick the one space that will be best for all work. My rule of thumb will be to use whichever DCam space I can convert my RAW capture into without clipping, while taking the initial edits in the RAW converter into account and modifying those edits to stay within the reasonable bounds imposed by the five working spaces. If I can't comfortably make a capture fit into 1 or 2, I'll try 3, and if not that, then I'll use 4.

If you want to feel free to convert from 16-bit format to 8-bits, it's best to stick with the three smaller spaces, and the smaller the better, so long as it's big enough to work without clipping. Avoiding clipping is the top priority, but I prefer to balance out the need for discipline in image rendering (the printers and the displays all have less bright-colors gamut than these spaces do — except DCam 1, especially the printers!) and to optimize the quantization error situation by not using a giant space routinely (i.e. DCam 4, ProPhoto or DCam 5).

My general recommendation for an all-purpose digital camera working space is DCam 3, J. Holmes, unless you are committed to staying in 16-bits per channel and want the ability to convert out directly to your working space from those RAW converters which have closed color management support, in which case ProPhoto is probably the better choice. DCam 2 is enough better than Adobe RGB for holding lots of colors with minimal quantization error problems that I'd like to see it available in cameras as an alternative to the highly arbitrary and limiting Adobe RGB design (it actually shares the red and blue primaries of sRGB, and merely increased the lateral gamut in the green direction, though it also increases the headroom gamut of the reddish colors.) See Adobe vs. sRGB, DCam 2 vs. sRGB, DCam 2 vs. Adobe, and DCam 3 vs. Adobe, 2D comparison plots). See 2D Gamut Plots for these and more, including Five DCams, and DCam 4 vs. ProPhoto. Remember that my spaces have the advantage with respect to their tone curves fitting well-behaved output devices best, which give them a quality advantage over other archiving spaces. Otherwise DCam 4 would be quite similar to ProPhoto in use with respect to clipping (though I do think it will clip fewer actual image colors than ProPhoto by a very small margin).

If you want the most headroom going into a working space for new digital captures (the least clipping of very bright or very dark moderately saturated colors), by all means choose ProPhoto, DCam 4 or even DCam 5, but they are large enough to include a great many colors that you'll never see in your subject matter, and have enough headroom (and toe room, as it were) to hold lots of colors that you can't see on any display, let alone from any printing system. And remember that you can get into more saturated colors than each master space can hold (up to a 99% chroma increase) by assigning the chroma variants for that master space to the image.

If you still need to look at the spaces to decide which one will suit your needs the best, you can purchase all six of my non-free spaces for $30 on the Products page, sans chroma variants, as insurance against choosing the wrong complete set(s).

One more important detail: don't forget that some RAW converters (Adobe Camera Raw v3 and Lightroom beta, and Canon's DPP v2, for example) do not allow you to choose your own working space into which to convert from the built-in camera profile. Capture One, RAW Developer, DxO, and Raw Shooter all do, and hopefully we won't have to wait too long to see Canon and Adobe open up to user choice of working spaces. If you are forced to convert into Adobe RGB or ProPhoto upon leaving ACR, then your best bet will usually be to stay in that space, and avoid the further conversion into one of my spaces, unless you'd like to convert down to 8-bits per channel, in which case you'd want to be sure you were in a space which was at least not unnecessarily large. That is part of the reason I am offering chroma variant sets for those two spaces as well. Certainly avoiding quantization error by using a smaller space demands that you either convert directly into that space during RAW conversion at 16-bits or convert into an intermediary space without clipping, then convert into the smaller space, then drop down to 8-bits, to achieve the full benefit of lower quantization. In other words, it's best if your conversions are all done with 16-bit data. Also, the quantization damage you are most likely to see in an image is that in the vertical direction, so avoid conversions between spaces with widely differing tone curves.

D) Finally, if you have existing images in Adobe RGB or ProPhoto, and especially if they have been edited with adjustment layers in Photoshop, or if you just want to stick with these existing industry standard spaces, you can get my chroma variant set for either of them and still enjoy the benefits of highly convenient and superior quality control over color. Also, if you're going to always keep your data, whether from film scans or digital cameras, in 16-bit form, using ProPhoto with my variant set for it is an easy and effective solution to achieving excellent quality, despite the space being huge and its tone curve being less than ideal, as the high bit depth of the data should protect the image from the deleterious effects of those two qualities.


5) See two-dimensional gamut plots here. And a few QuickTime 3D movies.


6) How do you use the RGB spaces and their Chroma Variants?

Short version:
A) Convert into the master space before you begin editing the image.
B) Assign any chroma variant for that master space until you like the effect and as often as you like.
C) When you're done editing the image, convert into your output profile, usually a printer profile.

Long version:
Install the profiles by copying them into the correct locations on your system as explained here. Convert into the master space for the set before you begin editing in Photoshop, preferably in the scanning or RAW conversion software. Do not assign the master space (unless you are in Ekta Space and want more shadow detail so you're assigning Chrome Space 100).

Failing that (if the RAW conversion software forces you to convert into another working space), do the conversion into the master space before you start editing in Photoshop. Select Relative Colorimetric, no Black Point Compensation, and dithering off. Be sure the data are still in 16-bit per channel form.

Don't worry that your camera was set to either Adobe RGB or sRGB — when it saves RAW files, those settings mean (almost) nothing — the file is still RAW, i.e. in camera space, not in Adobe RGB or sRGB, when you open it in the RAW converter. RAW converters usually assign camera profiles to the image automatically and the file is nearly always converted from that camera profile into an RGB working space when you save it from the converter. Some converters let you pick any camera profile you like and those converters are most likely to let you also pick any working space you like, into which to convert.

Make sure that you don't clip at each conversion (check for zeros or 255s that weren't there before you converted — see #10 below). Make sure the image is still in 16-bit format as you convert, so the conversion will do negligible damage, assuming no clipping. Either drop down to 8-bits per channel after the initial conversion into the working space or after your final conversion into printer space (better, but more processing burden). The former has been my own workflow with scans of color film for many years and my carefully-made 24-bit per channel files have proven to be robust and free of any detectable artifacts of limited bit depth, largely because they started life as 16-bit sensor output. With digital captures and today's software and computing equipment it's more practical to keep the files in 48-bit form until the printing tools force the drop down to 8-bits per channel (i.e. until the very end), if they do, so most people using digital cameras are choosing that option. Most "16-bit" camera files are really only 12-bit though, so the deeper shadows of these images have been significantly harmed by this quantization limitation. See Sensor Shadows.

If you should forget and you need to convert into the master space after you begin editing, don't flatten the image during conversion if you want to avoid another processing hit, but the effects of any adjustment layers on the image will then change when you convert the image layer from one space to another, because the adjustment layers are RGB operators and you'll be changing the RGB values in the file, so you will probably have to tweak any or all adjustment layers to get what you had before, including adjustment for clipping, contrast, and tonality. That's why it's important to convert into the master space right at the start.

If you have existing edited files with layers in another working space which happens to be either Adobe RGB or ProPhoto, then it would probably be best to stick with those spaces and use my chroma variant sets for them, so you don't have to re-work your existing edits.

Once you have done the initial conversion into the master space, assign any chroma variant for that space to the image at any time prior to output conversion, to change the amount of color in the image (In CS2 that's Edit > Assign Profile... and in Photoshop CS or earlier that's Image > Mode > Assign Profile...). Do not convert into the chroma variant! Assigning a variant will never cause clipping, unlike traditional saturation adjustments, or adjusting chroma in Lab and then converting back to RGB.


7) Is it OK to convert into the Chroma Variants instead of into the master space? No! It doesn't work (right, if at all). Photoshop CS2 treats a conversion into a chroma variant as if it were an assign, and assignment is what the control is all about, so just assign it to be sure everything works right. It is structurally impossible for these profiles to give a normal conversion result when converting into them, and often it won't work at all, so don't mess with it. Convert into the master space, then assign the variants. Converting from the variants is entirely fine, and also entirely necessary in order for them to function.

Converting from one profile to another keeps the colors the same, while changing all the numbers in the file. Assigning a new profile to an image changes all the colors in a color managed viewing environment only (like Photoshop), not by changing the numbers in the file, but by redefining what they mean.

I know this is terminally confusing until you get it, but just think about this: Photoshop 6 and later do something very tricky, but very good: when some of the data from an image on your hard drive is brought to the screen for you to see, the image data flowing to the screen only is converted from the source profile for the image to your monitor profile so the colors will appear as they should, i.e. as the source space defines them to be. If you turn on Proof Colors in the View menu, then Photoshop shows you image data on screen that has instead been converted from the source profile to the selected printer profile and then back to your monitor profile, so you can see on screen how going out to the printer will affect the image, if at all. In order for all this trickery to work, the engineering must keep track of two made-on-the-fly, fancy lookup tables that allow instantaneous conversion of numbers coming from the hard drive to the screen into new values, for each open image, to either give you a source simulation or an output simulation (aka soft proof). If you convert a file from one space to another while you are viewing it in Photoshop, the two fancy, concatenated lookup tables (called color worlds) for each image must be replaced with new ones which reflect the proper conversions from the new profile to your monitor profile and out to printer space and back to the monitor, and these tables are built in the blink of an eye and swapped into use at the very instant that the newly converted image file data begins arriving on the display! Therefore converting an image from profile to profile usually appears to have no effect on the color (save for gamut mapping limitations), yet all the numbers in the entire file will have changed — they just get re-defined at the same moment that they have changed, so they wind up appearing not to have changed, for the most part. Converting a file to new numbers injures the file a little bit, as all digital processing does. Assigning a new source profile does no injury whatsoever, although it appears to change the colors on screen in Photoshop or in a print a great deal more than converting does. Assigning a new source profile again means that the two simulations get updated instantly, but this time the appearance of the image data changes. There is a certain irony in this. To finish the story, if you assign a new source profile, then any conversion, be it to the screen or to the printer, involves new results, so what you see on screen in the way of altered chroma, etcetera, winds up being largely the same in a print, since the image data are converted from the new source profile to the printer profile before printing. Maybe if you read that enough times it will begin to make sense.


8) Is there a way to apply the effect of a Chroma Variant to just one part of an image? Yes. You will have to copy the main image layer into a new document and assign a variant that has the color shift you want, relative to the master profile. Then convert the image from the variant to the master profile, then reinsert the image as a layer right on top of the main image layer, keeping the file numbers constant (not allowing Photoshop to convert from the source profile to the main image profile when you paste). Then use the mask of the upper image layer to control where the image has the local adjustment. For example, fill the layer mask with black and then paint with white or gray to make any region of the upper image layer visible. I have never needed to do this, but it's possible, should you want it.


9) Questions and Tech Support: In theory they're all answered here somewhere, but if you want to ask me about my spaces or chroma variants send me a note and I'll get back to you when I can. If you are confused about how color management works (as everyone is while learning it) it is not my responsibility to unconfuse you about it! Please keep that in mind. Profile sets do not come with a free education in color management, beyond what you can glean here. For that, you may want to take one of my workshops or arrange for some one-on-one tutoring, if I have time (see my workshops page for more info).


10) Check for clipping in Photoshop by opening the Levels tool or a Levels adjustment layer and holding down the Option key (Mac) or Alt key (Windows) and simultaneously moving the end sliders of the Input Levels part of the tool. The shadow slider should make the whole image turn white, except for any values that are already as low or lower than the value of the current slider position. The highlight slider should make the whole image turn black, again except for any values that are already as high or higher than the value of the current slider position. The clipped areas are color coded so you can tell which channel(s) are clipping.

It is particularly important that this clipping warning, unlike some, shows you clipping in any individual channel.

To check for clipping in Adobe Camera Raw 3, keep the Shadows and Highlights checkboxes checked and look for the colored warning for 255s or black that will be happening as the file is mapped into the currently selected RGB working space (sRGB, ColorMatch, Adobe, or ProPhoto). This great feature should be more widely used in Photoshop and elsewhere, because checking for inadvertent clipping is a necessary task as many points in a fine imaging process. Unfortunately though, the shadow warning is designed to only show you pixels where all three channels have gone to zero! Notice how clipping decreases as you go from selecting sRGB to Adobe RGB to ProPhoto RGB, most often due to more headroom.


11) In Photoshop's Preferences, go to the File Handling page, and set Maximize File Compatibility to Never or Ask. When "file compatibility" is maximized, it means that Photoshop saves layers in a primitive version of the .psd file format, which doesn't compress the layers and so the files saved to your hard disk will tend to be huge. In case you need to send somebody a .psd file which they might not be able to open, because they're not using as recent a version as you, or other outdated software that can only open earlier format .psd files, you can go ahead and "maximize compatibility" when you save. In case you need to do the local chroma trick in #8 above, this may help you to greatly minimize the additional file size increment. Same goes for brushed-in curves layers, one of my favorite things!


12) In Photoshop, to see what you can't see, set up a soft proof of your own monitor, and then enable the Gamut Warning feature! Colors from image values which have been altered to make them fit into your screen's gamut will be marked as out of gamut. Go to View > Proof Setup > Custom and select your current monitor profile as the "Device to Simulate". Click OK, then enable the Gamut Warning option in the View menu. This can only work if your source profile for an image has a bigger gamut than your monitor profile does.


13) It's all simulations in Photoshop: To avoid terminal confusion about what you see in Photoshop, it's vital to understand that what you are seeing, whenever you view an image in Photoshop, is one of two kinds of simulations (as I tried to explain in #7 above), either:

A) What I call a source simulation, where the image's numerical values are being converted for screen display only, from either the embedded source profile or, if none is embedded, from the current default source profile (working space profile) which is set in the Color Settings dialog, to your monitor profile, so that colors will ostensibly appear correctly on screen. The exception is when the current source profile is your monitor profile, in which case there is no simulation when a soft proof is not enabled.

Or

B) A soft proof, i.e. Proof Colors, is enabled in the View Menu (Command-Y Mac/Control-Y Windows). This causes a fancier conversion, from the source profile to the selected output profile (usually a printer profile) and then back to the monitor profile. This let's you see how image colors will be scrunched to fit into printer space. It also lets you enable the Gamut Warning (Shift-Command-Y Mac/Shift-Control-Y Windows) which will overlay a color of your choosing onto all source colors that are outside the gamut of the destination profile. This is a very useful learning tool. For it to be of value when editing an image, you must set up the correct simulation in View > Proof Setup > Custom...


14) In Photoshop, to avoid the default soft proof setup being "Working CMYK..." for every image you open, enter the setup you like in View > Proof Setup > Custom, and then hold down the Option/Alt key (the typical hiding place in Photoshop for important, unlabeled functions) and click the "—> Default" button.


15) I developed most of these color spaces and matching chroma variant sets to solve my own imaging needs, but I offer all of them for sale to the public here on my site or by contacting me by email, as indicated on the Available Products page.


16) What is quantization error anyway? When image colors exist or are mapped into any RGB space, the space functions like a giant 3D grid of finite, tiny locations where a color can be placed. An image in 8-bit per channel form can sit in a 3D array of 16.7 million locations, or 256 planes in each of the three dimensions of the cube. Colors can be at the precise locations in the grid, but not in-between. If the RGB space is defined so as to cover a wider range of colors, the spaces between the steps increase proportionally. Since most images cover only a few percent of the volume of the RGB space, an image may exist in just a couple of dozen or fewer planes in vertical orientations. If such a space is enlarged to twice the radius, the number of planes of image info will be cut in half in each lateral dimension. The number of vertical steps (horizontal planes) doesn't change as RGB spaces get wider or narrower, rather the preservation of vertical steps is best optimized by keeping the tone curves of spaces being used as the source and destination for conversions similar. Preserving the vertical steps during image processing is more critical to image appearance than preserving horizontal steps, to a point. When one looks at a precise 3D model of the colors present in an image, as thousands of points hanging in space, one can see what look like the rows in an orchard viewed from a passing car. The gaps between the trees run out in several directions, but in 3D instead of 2D. When comparing two converted files, one into a small DCam profile and one into a larger DCam profile, the difference is that the gaps between all the rows have grown wider, and the number of locations where originally distinct colors can be placed is no longer enough to keep all of the colors distinct, or to keep differences between colors as subtle.