About the Master Spaces and Chroma Variant Sets

(Jump to Available Products page for pricing or purchase.)

(Go back to main Profiles page.)
August 16, 2006 (revision September 10, 2006)

Summary:

(Skip ahead to Profiles and Sets further down this page.)

First, I have my original pair of profiles built for scans of transparencies (aka chromes), each with the same gamut, but different tone curves. Second, I have a new quintet of profiles built specifically for use with digital cameras, all sharing the proprietary, perceptually linear tone curve of the Chrome Space 100 transparency profile, in a useful and very carefully crafted progression of gamut volumes. The middle one of the quintet is also particularly well-suited for scans of color negatives.

Chrome Space 100, J. Holmes
Ekta Space PS 5, J. Holmes
DCam 1, J. Holmes
DCam 2, J. Holmes
DCam 3, J. Holmes
DCam 4, J. Holmes
DCam 5, J. Holmes

I also have sets of 29 Chroma Variants for each of my seven spaces, plus sets for Adobe RGB
® (1998) and ProPhoto RGB — nine sets in all. If the last thing you need in your life is more bewildering choices to do with your computer, check out my FAQs and Tips page, which contains advice on choosing sets.

Working Space Issues:

(See also All About RGB Working Spaces for even more detail than the following short essay.)

So long as the colors in your images coming from the source profile (usually a scanner profile or a digital camera profile in a RAW converter) don't clip while entering any RGB space, the space is large enough for that image, as it is being processed during capture. Clipping means essentially that colors are pushed to contain values of 255 or zero in one or more channels (red, green or blue), because the colors were originally outside the boundaries of the RGB space by virtue of being lighter, darker, or more saturated than the nearest colors contained by the RGB space. Colors being clipped are thus hitting a wall of the RGB cube and image is being lost as adjacent pixels cease to have different values in one or more channels. Avoiding clipping when entering RGB spaces is the most important thing regarding the use of these spaces.

Avoiding using spaces which are more than a little larger than they need to be is of some benefit too, when
the image data are in 8-bit per channel form (or may be at any point during their processing) prior to converting out of the working space.

Also, using a space with a tone curve that matches idealized output tonality is a good idea for any archive, in my opinion, to protect the image against further loss of levels, again if the image data are in 8-bit form or may be, at any point prior to converting out of the working space.

Using a white point of D50 is a good idea, because color management workflows rely mainly on D50 white points: the Profile Connection Space (PCS) is D50 and the output profiles are nearly always D50. Using spaces with gamut shapes that match the nature of the input gamut is a good idea to optimize quantization efficiency.

Using spaces which are large enough yet not too large for your whole range of work may also be a good idea (it's good for convenience to only use a single space, but may not squeeze the last bit of quality that's possible by using more than one working space).

Designing large-gamut spaces for photography should entail the use of unreal colors in the space, particularly the blue and sometimes the green primary regions, because that yields the best encompassment of the important ranges of colors for the various uses, despite the fact that such spaces can be inadvertently abused by graphic artists who might put a color value into them like (0, 0, 255). Maximum blue will correspond to an imaginary color in many spaces and will often yield a very broken color from a printer profile. Absent colors being painted into photographs, photos rarely wind up containing unreal values, and essentially never values from the extreme corners of the space. When they do it's only because somebody pushed the colors while editing far beyond what is visible on any kind of monitor.

Should you find it necessary to convert from one master space to another, it's best if the spaces share the same tone curve, and it can also be useful if the hues of the primary colors of each space are similar so that the spaces can be used in unexpected ways. For example, Chrome Space 100 makes an amazingly good scanner profile for some if not most of the pre-ICC color management class of drum scanners, providing unmatched deep shadow detail often without spoiling the color accuracy too much compared with an actual scanner profile.

Finally, it's a wonderful thing to have a Chroma Variant set(s) for the working space(s) you use, because with it you can control the colorfulness of your images in a much more satisfactory way than is otherwise feasible. Given chroma variant sets to complement a working space, it makes perfect sense for the working spaces to be input-centric in their design, not output centric! The space needs to be sized to fit the incoming colors, not the gamuts of the printers! This allows for an elegant level of quantization efficiency, excellent color and tonal control and ease of use, all at the same time.

My working spaces are built around all of these principles and more.

When building a working space for containing scans of transparencies, it makes a lot of sense to use a single space for all such scans, because the gamut of the color film is large but not huge, because the various film types have very similar gamuts, and because the colors of images in those gamuts tend to utilize a pretty substantial fraction of the gamuts, in part because the film exaggerates the color of the lower middletones quite a bit. Digital cameras are different. Their gamuts tend to be much larger than those of processed transparency film, but at the same time, digital captures are unlikely to exaggerate subject colors very much, and therefore the range of colors that any given image actually occupies within a working space is likely to be smaller, at least in the lower middletones. Most importantly, just because a camera's gamut is large doesn't mean its images' gamuts are. But one can't be sure that some images might not contain brilliant colors that we want avoid clipping, and many images with ordinary levels of color still can clip because colors get too light or too dark and hit the sloping roof or floor of a given RGB cube (the cube tilted with one corner straight up and one straight down).

Indeed, brilliant colors are not the most likely to get clipped entering an RGB working space from a larger source space, be it a digital camera profile or a scanner profile, especially when the image has been edited in the RAW converter, as it always is, or in the scanning software. It's colors that have more modest saturation but which are close to white or black that create gamut busting problems most often. Progressively larger RGB spaces offer more room near white and black for such colors, but less and less so as the color approaches neutrality -- like differently sloped pointy roofs on a house that has the same maximum height in each case. This illustration shows this very well. My approach to this problem of widely varying needs for digital capture has therefore been to design a complementary quintet of spaces for digital cameras, so that images can be stored to better advantage — in a space that's about right for them, regardless of the need or preference. Quick checks of the gamut warning feature of the RAW converter (if it has one) should reveal which space is the right one to use for a given image in a given state of adjustment. Alternatively, one can, with 48-bit data only, stick with one of the larger spaces for everything, or with a mixture of 48- and 24-bit data use the middle one for everything. I'll be using several of my DCam spaces for my own digital camera work.

Note that because the five DCam spaces share the same elegant tone curve, converting images between them will cause significantly less damage than conversions between many working spaces, when the data have 8-bits per channel. The quintet can function as a flexible, yet manageable way to optimize the character of spaces for a full range of digital camera work. I believe this system of 5 masters and 5 variant sets is an optimal way to work with the full range of digital captures while minimizing quantization errors. For the very best results, each image should be converted directly into one of the master spaces, but this will not be possible when using Adobe's Camera Raw 3 or Canon's Digital Photo Professional 2.1, for example. Capture One, Raw Shooter, RAW Developer and DxO Optics Pro are better in this regard because they allow your choice of working spaces into which to convert from the camera profile. With ACR or Canon's DPP, it will be necessary to convert into either Adobe 1998, ProPhoto RGB or Wide Gamut RGB, typically, and from them convert into my space of your choosing, a regrettable processing hit (unless the data are in 16-bit form during the conversion) and extra bit of work. All RAW converters should support the user's choice of destination profiles, and perhaps in the future they will. Alternatively, you can use my Adobe 1998 or ProPhoto variant sets, and perhaps I'll make one for Wide Gamut RGB too at some point.

Profiles and Sets:

(Jump to Available Products page for pricing or purchase.)

(Compare master profile gamut volumes here — 64KB JPEG)

(Find many more gamut diagrams in Sections I, II and IV of this page.)

1) Chrome Space 100, J. Holmes: this is an overhauled version of my first working space, Ektachrome Space, J. Holmes and replaces it, as of mid-2005. Chrome Space 100 is backward compatible with the original space and cleans up a few technical issues that I was unable to address when I first created my master archiving space for scanned film. I have used Ektachrome Space for all my scanning since 1997, and will scan into Chrome Space 100 from now on, as a rule. The gamuts and white points of Chrome Space 100 and its sister space Ekta Space PS 5, J. Holmes (aka Ekta Space) are identical. Only the tone curves are different. Chrome Space 100 uses a custom 1,024-point tone curve very carefully designed to match the tonality of perceptually linear printers. This curve is shared by my DCam spaces. Ekta Space instead uses a simple Gamma 2.2 curve, as do Adobe RGB and some other spaces (most working spaces use gamma curves). Gamma curves are too flat near black. I built Ekta Space because Adobe® Photoshop® 5 did not support RGB spaces unless they had simple gamma curves, but Ekta Space went on to become the more commonly used of the two because I gave it away from the beginning. My transparency spaces were conceived around the actual gamut of the dyes in transparency films, as mapped by a proper scanner profile, so they are most appropriate for use with scans of positive films, however as a practical matter they work well with scans of color negatives and digital camera captures.

Note that the difference between the tone curve of Chrome Space 100 and that of Ekta Space PS 5 below is one that can be very useful to impart extra deep shadow detail to scans that are in Ekta Space by simply assigning Chrome Space 100 or its variants. This is commonly useful w. scans of transparencies, which themselves tend to crush shadow detail because the film often has insufficient latitude for the subject matter. Most transparency scans fail to restore as much of this crushed detail as they could and should. Please check out this page to show what happens to an image with dark shadows that is in Ekta Space when you assign Chrome Space 100 to it.

gamut diagram

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


2) Ekta Space PS 5, J. Holmes aka "Ekta Space": download this industry-standard free profile and my latest Read Me file about it and a lot of related topics, here on the Products page. This profile is a variation on my original master working space, Ektachrome Space, J. Holmes, which I developed to convert my scans into when I was faced with the need for an optimal RGB space for storing my archive of scanned transparency film, as I began to make fine drum scans of it in 1997. The only difference between it and my original profile is its tone curve, which is a simple gamma 2.2 curve instead of the superior, perceptually uniform tone curve of my original space.

gamut diagram (same as Chrome Space 100)

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


(Compare the five DCam master profile gamut volumes here.)

3) DCam 1, J. Holmes: The first of my five new spaces designed in 2006 specifically for use with digital cameras, by far the smallest, and intended solely for images without very saturated colors and with relatively neutral highlights. Here is an example. This space is intended as a special-purpose supplement to the other spaces. Light colors of modest saturation are likely to be clipped, but this space offers the possibility of taking extra good care of low-gamut images that might otherwise wind up occupying a tiny fraction of the volume of the working space, by keeping image quantization to a very low level. It has a gamut volume nearly 11 percent smaller than sRGB and over 38% smaller than Adobe 1998. It is exactly half the volume of DCam 2. It covers a minimal color range but it is carefully balanced around the core colors of digital capture. Its headroom is too low for most images with moderately colorful lighter tones to avoid clipping entirely.

gamut diagram


Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


4) DCam 2, J. Holmes: With fully double the volume of DCam 1, this is the second of my five new spaces designed in 2006 specifically for use with digital cameras. As the smallest/most conservative of my three general-purpose DCam spaces, DCam 2 is nevertheless a fairly large space. It is ideal for images of moderate to low saturation and/or having moderate contrast in saturated colors. It has a gamut volume 23 percent larger than Adobe 1998 and about 79% larger than sRGB, plus it is much more efficiently shaped than Adobe 1998 for containing likely actual image colors. I estimate that it contains about 50% more of the most useful subject color volume than Adobe 1998 does. It can contain all the colors that we can see on nearly all displays available today, except for a few of their very strongest blues, and a great deal more besides. I will use this space for many of my relatively muted digital captures, and I expect it to mainly be used by other people in conjunction with at least one of the next two larger DCam spaces.

Because the images captured with digital cameras vary greatly in their gamuts, no single space can be entirely ideal for digital capture workflows, although if the data are always kept in 16-bit per channel form, the very large spaces, including DCam 4, ProPhoto and DCam 5 work very well despite their great size. Unlike working with transparency film, I believe that it makes relatively more sense to have a tiered system of working spaces which makes it possible for each image to have a storage vessel which is neither too small nor quite oversized, thereby optimizing quality, at least when the capture is converted to 8-bits per channel at any point while still in the working space.

I advise using Adobe 1998 as the color space for in-camera JPEGs, so long as it and sRGB remain the only two choices. When you are only concerned with the RAW captures from your camera (and not JPEGs), setting the camera to sRGB may help to make the blinking highlight warning more sensitive to blown colorful highlights.

gamut diagram

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


5) DCam 3, J. Holmes: The third of my five new spaces specifically for digital cameras, DCam 3 is a more spacious and forgiving, general-purpose, very large gamut space for digital camera captures. This space is 133% larger than sRGB, 61% larger than Adobe RGB, 30% larger than DCam 2, and 20% larger than Chrome Space 100. It very efficiently covers the range of colors likely to be captured by a digital camera, with substantially more green to red range, meaning especially more yellow and nearby yellow-green and orange, than Chrome Space 100 and somewhat fewer purple-blues. I will use this space frequently when converting my DCam files as I work with my new digital camera system. If you are working with digital capture and you're considering your first of my working spaces with chroma variants, this is the one I would recommend as the your first choice without knowing more about how you work.

gamut diagram

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


6) DCam 4, J. Holmes: The fourth of my five new spaces specifically for digital capture and the second largest of the five, DCam 4 is a general-purpose, extremely large gamut space for digital camera captures, with a gamut volume and shape similar to that of Kodak's ProPhoto RGB, but like most of my spaces it uses my proprietary tone curve for better perceptual linearity and therefore better-protected output quality on ideal printers, and it is a little more efficiently designed than ProPhoto in my opinion. This space is 204% larger than sRGB, 110% bigger than Adobe RGB, 70% larger than DCam 2, 56% bigger than Chrome Space 100, and 31% bigger than DCam 3. This space, like ProPhoto, should generally only be employed with 16-bit per channel image data, given its great size, however these should not be considered hard and fast rules. DCam 4 provides a good measure of insurance against gamut-busting colors in case they don't fit within DCam 3 and substantial further headroom to avoid clipping some bright or dark colors as well (no additional headroom in the pure cyan to blue to pure magenta range, however). It is a viable choice for a single space for all digital captures to the same extent that ProPhoto is. If you are comfortable with ProPhoto's very large size, then this is a somewhat better alternative, with a better tone curve, especially when combined with its matching variant set. Note that I am offering a variant set for ProPhoto as well, which may be preferable if you are stuck with a RAW converter which won't let you choose any working space to convert into, or if you already have finished work in ProPhoto.

gamut diagram

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


7) DCam 5, J. Holmes: The fifth and largest of my five new spaces specifically for digital capture, DCam 5 is useful as a special-purpose, extremely large gamut space for digital camera captures and other RGB imaging work. It has a gamut volume and shape just sufficient to perfectly encompass the entire human visual gamut. Like most of my spaces it uses my proprietary tone curve for optimal perceptual linearity. The net, Lab-constricted gamut volume of this space is 324% larger than DCam 1, 278% larger than sRGB, 161% bigger than Adobe RGB, 112% larger than DCam 2, 95% bigger than Chrome Space 100, 63% bigger than DCam 3, and 25% bigger than DCam 4. Its non-Lab-constricted gamut is about 8% larger still. This space, like DCam 4 and ProPhoto, should generally only be employed with 16-bit per channel image data, given its great size. DCam 5 provides a lot more headroom than DCam 4, though this range of added headroom is sparsely populated by typical image colors, so under most circumstances it should not be particularly useful. It may prove useful, however, as a space into which to do batch RAW conversion, so as to defer the handling of clipping issues until later on. After years of looking at the way RGB spaces are limited in their ability to cover the CIE chromaticity diagram's visual gamut range by the inability of the blue coordinate to be placed below the x-axis, I just decided that I wanted to make such a profile, and the exact coordinates necessary to just fully encompass the visual gamut were the most obvious and appealing design for such a giant space. Perhaps others will find uses for it which have not occurred to me.

gamut diagram

Chroma Variant Set — 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


8) Chroma Variant Set #8 [for use with Adobe
® RGB (1998)]†: A new and very handy option for controlling color better with images that are already in Adobe RGB.

gamut diagram

Set of 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


9) Chroma Variant Set #9 (for use with ProPhoto RGB)†: A new and very handy option for controlling color better with images that are already in Kodak's ProPhoto RGB (aka ROMM RGB, aka RGB Master). This is also an economical, all-purpose complete set for images that will remain in 48-bit form at least until converted into an output space.

gamut diagram

Set of 29 variants: +04, +08, +12, +16, +20, +25, +31, +37, +44, +52, +60, +70, +80, +90, +99, -04, -08, -12, -18, -24, -30, -37, -44, -52, -60, -70, -80, -90, -99


† These two sets of chroma variants share no content or information whatsoever from the original profiles, which are the intellectual property of Adobe Systems, Inc. and Eastman Kodak Company, except inasmuch as they utilize the same generic gamma curves and white points as the original profiles. Nevertheless, the chroma variants perfectly complement these spaces just as my other sets perfectly complement my own master profiles. The proprietary tone curve of my profiles is not generic and is therefore protected intellectual property, as is the unique combination of primaries in every published RGB working space. All published profiles are also themselves protected by copyright.

Any and all trademarks referenced in these pages are the properties of their respective owners.
Adobe® Photoshop®, Adobe®, Adobe® RGB and Adobe® RGB (1998) are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States and/or other countries.

(Find many more gamut diagrams in Sections I, II and IV of this page.)

(Jump to Available Products page for pricing or purchase.)