Movatterモバイル変換


[0]ホーム

URL:


Wayback Machine
356 captures
13 Oct 2002 - 16 Jan 2026
NovDECJan
21
201020112012
success
fail
COLLECTED BY
Organization:Alexa Crawls
Starting in 1996,Alexa Internet has been donating their crawl data to the Internet Archive. Flowing in every day, these data are added to theWayback Machine after an embargo period.
Collection:Alexa Crawls
Starting in 1996,Alexa Internet has been donating their crawl data to the Internet Archive. Flowing in every day, these data are added to theWayback Machine after an embargo period.
TIMESTAMPS
loading
The Wayback Machine - https://web.archive.org/web/20111221111848/http://www.normankoren.com:80/color_management.html
















In Association with Amazon.com






Google




















CarlosE. Mora's two-part Spanish translation: Losmodelos y espacios de color (II) Gestión del color y laciencia del color:Introducción  | LasBases de la gestión del color.


Obtainingpredictable color reproduction in the digital darkroom can bea challenge because each device-- digital camera, scanner, monitor, orprinter-- responds to or produces color differently. If you limit youroperations to your own well-characterized equipment and follow theproceduresoutlined inMonitorcalibration andPrintercalibration, you can get reasonably good results withoutcolormanagement.(The operating system performs a certain amount covertly, in thebackground.)But you may want to go further-- to push the envelope. You may want toTo meet these goals you'll need to get involved with the setof softwarepackages and procedures known as color management. There's a learningcurve,but it will ultimately simplify your workflow.

The series begins with anIntroductionto color management and color science (this page).Implementationpart 1 describes how to set up color management andinterpret thecontents of(filesthat describe the color response of a device or a). It featuresPictureWindow Pro, but includes information on Photoshop.Implementationpart 2 discusses monitor profiling and workflowdetails. The seriescontinues withObtainingICC profilesand building them with MonacoEZcolor andEvaluatingprinters and ICC profiles.

Backgroundreading:

RealWorld Color Management: Industrial-strengthproduction techniques,by Bruce Fraser, Fred Bunting, and Chris Murphy. Paperback, 560 pages.The closest thing to a Color Management Bible. Buy it if you want to gointo real depth. But as its subtitle "Industrial-strength productiontechniques"indicates, much of the material covers prepress applications (CMYK,etc.),which are of only indirect interest to photographers.
MasteringDigital Printing: The Photographer's and Artist'sGuide to High-QualityDigital Output, by Harald Johnson. Paperback, 400 pages. Perhaps moreforartists than photographers, it has a good introduction to colormanagement.
ColorScience: Concepts and Methods, Quantitative Dataand Formulae,by Günther Wyszecki and W. S. Stiles. Paperback, 968 pages. Acollectionof scientific review papers-- not for the casual reader. This is thereferenceI use to verify questionable statements on color and color managementinpopular literature. It's thesource.

Colormanagement equipment/supplies

ColorHQ.com |  ColorManaged.com


Spectral sensitivity of the human eyeTheretina of the human eye has two categories of light receptor:,which are active in dim light and have no color sensitivity, and,which are active in bright light and provide us with our ability todiscriminatecolor. You probably learned that the three types of cone are sensitiveto red, green, and blue (R, G, and B). The relative sensitivityof the three receptorsfor the "normal" human eye, designated by Greek letters beta, gamma andrho (β,γ,andρ), is illustrated by the blue, green,and red curves on the right.Althoughthe beta and gamma sensors correspond closely to blue and green, therhosensor (the red curve) isn't even close to red. An ink with the samereflectivityspectrum would appear yellow-orange.

The eye/brain discriminates color by processing therelativestimuli in the three sensors. R, G, and B are used as additive primarycolors because their distribution across the visible spectrum producesa wide-gamut color image,because theymatch the eye's response. Fewer than three colors is insufficient.Additionalcolors offers some advantage-- that's why recent inkjet photo printershave 6 to 8 colors. Combining three colors-- even monochromatic(spectrallypure) colors produced by lasers-- can produce most,thoughnot,of the colors the eye can see.

How is this known? A set of experiments is run usinga split screen.Half is illuminated by a monochromatic light source with variablewavelength.The other half is illuminated by an adjustable combination of red,greenand blue, which can be produced by lasers or by filters, which have abroaderspectrum than lasers. If the two halves of the screen can be matchedwithsome combination of the R, G, and B lights, then the color of the puremonochromatic light is within theof colors defined by the three light sources. If no match is possible--if white light must be added to the monochromatic source to provide amatch--then the color of the monochromatic source is. This experiment shows thatnocombinationof three real light sources can duplicate the full gamut of humanvision.

Monitors with three phosphor colors (RGB) havelimited color gamuts;printers with four ink colors (CMYK = cyan, magenta, yellow, black)haveeven smaller gamuts. Printers with additional colors-- 6 to 8 are notuncommon--have larger gamuts. The eye's peculiar response has consequences forthediscipline called,which has arisen to quantify human vision. A detailed exposition wouldoverwhelming, but a few aspects, which are widely discussed but poorlyunderstood, are important for color management.






CIE imaginary primariesIn order to quantify humancolor vision, theCIE(CommissionInternationale de L'Éclairage)has established a set of imaginary "red," "blue," and "green" primarycolorsthat, when combined, cover the full gamut of human color vision, i.e.,a combination of the three can match any monochromatic light source.Thesecolor primaries (shown on the left) have a curious property-- they havenegative energy in portions of their spectra, i.e., they are notphysicallyrealizable.

CIE Imaginary primaries

CIE color matching functions, courtesy Earl F. Glynn
CIE color matchingfunctions,courtesy Earl F. Glynn
reproduced from
efg's(Earl F. Glynn's) Chromaticity Diagrams Lab Report.
The combinations of these three "light" sources required to matchmonochromatic(spectral) light are determined experimentally (by performingmathematicalmatrix transformations of the results of the split screen matchingexperiment). The curves for these combinations, shown on the right, are called the.They are designated x,y,and z,and they never have negative values. TheColor Matching Functionsare used to derive thethat uniquely define an object's;two objects with the same tristimulus values have identical colorappearancewhen viewed under the same conditions. The X, Y, and Z tristimulusvaluesare calculated by integrating (summing) the product of the spectralreflectivity,the illuminant (the light source), and the corresponding color matchingfunction from 380 to 700 nanometers (nm). Although the tristimulusvaluesuniquely define an object's color, they do not define the eye'sresponseto the color, which depends on the environment and the eye'sadaptation.

If two objects with different spectral reflectivities have thesamecolor appearance (tristimulus values) under one light source, they aresaid to bematched.If they exhibit a marked difference under another light source (as withtwo pieces of cloth that look identical in a shop but differentoutdoors),they suffer from,frequently called.(This definition is misleading, but it has become so prevalent I won'ttry to fight it.) This phenomenon is most visible with neutral colors(grays).It is a problem with many color printers, particularly with Black&White prints; it was particularly severe with the Epson 2000P, butnewer printers (2005 and later) have largely solved the problem. Inactuality,metamerism, properly defined, is not a bad thing: it enables acombinationof just a few inks to take on the colors of a wide range of objects.

CIE 1931 xy chromaticity diagram from Gamutvision
CIE1931 xy chromaticity diagram,showing sRGB (smaller)
  and Adobe RGB (1998) (larger) color spaces. This diagram
iswidelyused  but not perceptually uniform;it overemphasizes greens.
 Image produced byGamutvision.

XYZ tristimulus values are important because they form thebasisof the familiar but often misunderstoodCIE 1931 chromaticitydiagram,shown on the right. This diagram is based onnormalizedtristimulusvalues, x, y, and z, where, x = X/(X+Y+Z), y = Y/(X+Y+Z), and z =Z/(X+Y+Z).This normalization (division by X+Y+Z) removes the brightness from thediagram so that only two coordinates, x and y, are needed to definechromaticity.z = 1-x-y would be redundant. Since Y is closely related to perceivedluminance,colors are sometimes expressed as xyY tristimulus values.

The colors in thediagram are not accurate, but the representation produced byGamutvision isabout as good ascan be obtained with the limited color gamut of a computer monitor.

The horseshoe line starting at 400 nm on the lowerleft and wrappingaround the top to 700 nm on the right is called the. It represents the pure spectralcolors-- the beautiful,intense colors produced by a prism in clear sunlight. The screen imageis but a pale approximation. The straight lineconnecting the endpoints of the horseshoe is called the. The full gamut of humanvision lies within thisfigure. The vertical axis gives an approximate indication of theproportionof green; the horizontal axis moves from blue on the left to red on theright. The location of white depends on the illuminant colortemperature.Some typical values (fromefg):







The 1931chromaticity diagram is not without itsflaws. The distance betweenjust noticeable color differences (called ΔE) ismuch greater in thegreen region on the top than at the bottom; it is not perceptuallyuniform.For this reason, CIE has defined additional color spaces, particularly(1976) and(1976), whichrepresent colors more uniformly. But the 1931 diagram persists becauseof historical inertia. All the CIE color spaces encompass the fullgamutof human vision and all are device-independent.BruceLindbloom presents theequationsfor converting between them. But unlike RGB color spaces, they aren'tintuitive.The precise meaning of their coordinates is difficult to visualize andthey contain values outside the gamut of human vision. Hence theyaren'tused as working color spaces for image editing. They play a vital butinvisibleroll in color management; you don't need to understand their details tomanage color effectively.
sRGB vs. Adobe RB (1998), shown in L*a*b* space
sRGB vs.Adobe RGB (1998) color gamuts
shown in L*a*b* color(showsthat there is
more than meets the eye in the CIE 1931 xydiagram)
Image produced byGamutvision.

The CIE 1931 chromaticity diagram is used to illustrate theof a device or a--therange of reproducible colors generated by a specific set of primaries.(Color space profiless also include a specification for gamma and whitepoint.) For three-color additive devices, gamut is defined by thetriangleformed by the x, y chromaticity values of the Red, Green and Blue (RGB)primary colors. Colors inside the triangle can be reproduced; colorsoutsidecan't; they are.The above figure on the right illustrates the gamut for thesRGBcolorspace. sRGB was designed to match the gamut ot a typical monitor withgammaof 2.2 and white point set to 6500K (D65). It is thede facto standardfor the Internet and the Windows operating system.

The gamut of sRGB, shown on the right inside the gamut ofAdobe RGB (1998), is quite limited. It maintains a distancefrom the line of purples and is weak in green and cyan, although thisweaknessis greatly exaggerated by the distortion of the 1931 diagram. Inmonitors,the gamut is limited by the phosphors, which are chosen for brightness,longevity, low cost, and low toxicity. Ideal phosphors-- with colorslocatednear 450, 520 and 650 nm on the spectrum locus-- don't exist. Mostmonitorswith P22 phosphors have similar gamuts.

Printers, whose colors are based on variants of CMYK(cyan, magenta,yellow, black) subtractive primaries, have gamuts whose shape is morecomplexthan a simple triangle-- often somewhat hexagonal with additionalvorticesat the Cyan, Magenta, and Yellow primaries. Because inks have imperfectspectraand are somewhat opaque, dark or light colors tend to have smallergamutsthan middle tones. Most four-color CMYK printers have smaller gamutsthanmonitors, but high quality inkjet printers with more than four colors(typicallywith the addition of light C and light M) may have larger gamuts. Colorslidefilms have considerably larger gamuts.

Now we approach one of the key issues in colormanagement: How is colorrepresented in computers? We focus on RGB color spaces because they areused for digital image files, and there are a good many of them. Othercolor models (CMYK, HSV and HSL) are discussedelsewhereon this site and inMicrosoft'sMSDN library.

In 24-bit RGB color spaces, color is described bythree 8-bit bytes,each of which can take on values 0 through 255. Pure red is (255,0,0),green is (0,255,0), blue is (0,0,255), black is (0,0,0), and white is(255,255,255).Some programs likePictureWindow Pro andPhotoshop CS can utilize 48-bit color that allows values of 0 through65,536for each primary color. 48-bit color eliminates banding that can occurwhen colors are adjusted in the 24-bit mode.


But whatexactly is meant by"pure" R, G and B? To accommodate the widerange of gamuts indifferent devices-- digital cameras, film, scanners, monitors, andprinters,a variety of color spaces has been developed. The de facto standard forthe Internet and Windows, sRGB, has a limited gamut corresponding to atypical CRT monitor. Other color spaces have larger gamuts--correspondingroughly to high quality printers or to film. Some have extremely largegamuts, covering most of the colors the eye can see and some it can't.Maintaining consistent color appearance in the translation betweendifferentdevices and color spaces is no easy task; color management provides areasonablysane and practical solution. But it's is no panacea. Even the mostsophisticatedsystem can't make two devices with different gamuts displayexactlythe same colors; it can't make a monitor or CMYK printer displayVelvia-saturatedcolors.



Thediagram below, taken from Jonathan Sachs' excellent tutorial onColorManagement for Picture Window Pro, is a simplifiedillustration ofa. (SeeImplementationpart 2 for adetaileddiagram.) Programs that utilize color management aredescribed as being,where ICM stands for Image Color Management. ICM standards aremaintainedby theInternational Color Consortium.
Basic color management image flow, courtesy Jonathan Sachs

In a color-managed workflow, the color response of eachdevice, eachimage file, and each image in the computer's active memory ischaracterizedby a file called an.ICC profiles have the extension ".icm" and are stored in specificlocationson Windows computers.



Youmay need to do somebookkeeping in these folders because problems canarise (especially with pre-XP versions of Windows) when more than about30 profiles are present. Since profiles are loaded by profile creationprograms such as MonacoEZcolor, Windows itself, image editors, anddevicedrivers, they can easily proliferate. I recommend creating a foldercalled"Unused profiles" for profiles you don't use. If nothing else, it willshorten the drop-down lists. You can obtain a printout of profiledescriptionsfrom Picture Window Pro by clicking on, to bring up the ColorManagement dialog box,then clicking ICC profiles can also be embedded aswithinimage files:TIFF, JPEG,PNG,and BMP are supported by most ICC-aware image editors.

ICCProfilesconsist primarily of tables that relate numeric data, for example, RGB(222,34,12), to colors expressed in a device-independent CIE colorspacecalled a CIE-XYZ or CIELAB. The colors may be the objectssensed bya scanner or produced by a printer or monitor. They can also refer tooneof the numerous.Monitor profiles have the same format as color space profiles. Profilesmay contain additional data, such as a preferred rendering intent andgamma,Monitor profiles often contain instructions for loading video cardlookuptables, i.e., forcalibrating the monitor.

The heart of color management is the translation orgamutmapping between devices with different color gamutsand files withdifferent color spaces. Mapping functions are shown in the yellow boxesin the illustration, above. They are performed by a or(),also called a, using data in the profiles.The CMM combines theinput and output profiles, both of which are referenced to a PCS, toperforma direct conversion between the devices or color spaces. Itinterpolatesdata in printer profile tables, which would be prohibitively large ifallpossible color values were included.

Picture Window Pro uses either the Windows defaultcolor engine,ICM2.0, or an alternative engine,LittleCMS. AdobePhotoshop has its own color engine, ACE. Colorengine mappingsmay be called from ICM-aware programs or device drivers. You must beawareof where the translation takes place in your environment. If you arecareless,mapping can take place twice (or not at all), with undesirable results.We will show examples.

For reference,ColorSyncis Apple's color engine.ColorMatchRGB is Apple's default color space, with gamma =1.8 andgamutbetween sRGB and Adobe RGB (1998).is an older color space with a narrower gamut. Easy to get confused.

is performed withone of the fourrenderingintents (gamut mapping algorithms) recognized bythe ICC standardand by Windows ICM 2.0. The rendering intent determines how colors arehandled that are present in the source but out of gamut in thedestination.Since there are several nomenclatures for gamut mapping, I use a colorcode to distinguish the sources:,,.I'll generally stick with thenomenclature.

  1. Gamut mapping diagram,also calledor. Thisis PW Pro's default, and is generally recommended for photographicimages.The color gamut is expanded or compressed when moving between colorspacesto maintain consistent overall appearance. Low saturation colors arechangedvery little. More saturated colors within the gamuts of both spaces maybe altered to differentiate them from saturated colors outside thesmallergamut space. In the diagram on the right, the left and right of thecolorspace blocks represents saturated colors; the middle represents neutralgray. Perceptual rendering applies the sameBruceFraser points out that for an image with unsaturated colors,e.g.,
  2. , also calledor. Reproduces in-gamut colors exactly and clips out-of-gamut colors to thenearest reproducible hue. Not reversible. See diagram.BruceFraser says, "
  3. , also calledor. Reproduces in-gamut colors exactly and clips out-of-gamut colors to thenearest reproducible hue, sacrificing saturation and possiblylightness.On tinted papers, whites may be darkened to keep the hue identical totheoriginal. For example, cyan may be added to the white of acream-coloredpaper, effectively darkening the image. Rarely of interest tophotographers.

  4. ,also calledor. Mapsthe saturated primary colors in the source to saturated primary colorsin the destination, neglecting differences in hue, saturation, orlightness.For block graphics; rarely of interest to photographers.
Rendering intents don't always perform according to thetextbook description. You can view their actual performance withGamutvision—a powerful utility written by the author of this tutorial.

ICCprofiles reference the CIE device-independent colorspaces-- CIEXYZand CIELAB. When gamut mapping is performed, tables from the source anddestination profiles are combined; gamuts are mapped directly. Thisminimizesloss of image quality. More on rendering intents can be found inarticlesfromMicrosoft,ProfileCity.comandBruceFraser.
 
Gamut mapping is acomplextopic, particularly for Perceptual rendering intent. The details of theare contained within the profile. There are a great ways of performingPerceptual rendering. Here are some highly technical papers, guaranteedto generate more questions than they answer. Like what techniques areactuallyused in your profiles? This doesn't soundanythinglike the familiar"you push the button, we do the rest" marketing hype.


Colormanagement has two key features:A usesICC-aware programs to do two things: (1) recognize color spaces anddeviceprofiles, and (2) apply the appropriate gamut mapping when transferringimages. That's all! But with color management, "the devil is in thedetails,"and what details! They includeWe'll deal with these issues in the remainder of thisseries.
 



[8]ページ先頭

©2009-2026 Movatter.jp