4

SA
M
PL
E
CH
AP
TE
R
MAKING SENSE
OF COLOR
MANAGEMENT

CRAIG HOCKENBERRY

FOREWORD BY
JOHN GRUBER
You just got a gig to design a website for an optometrist,
Dr. Eyeful. Sweet!
The doctor wants you to use his two favorite colors: red and
purple. (Bonus points if you include a “site” and “sight” pun in
the tagline.)
You’ve been developing sites since the days of web-safe color
palettes. By picking red, green, and blue (RGB) values of (255,
0, 0) and (102, 0, 204), you’ll satisfy the doctor’s branding
requirements while working with tried-and-true colors. Of
course, you’re careful to use those exact color specifications in
both your Photoshop document and CSS styles.
But when you load the page in your browser, you see a weird
color shift in the header (Fig 1.1).
Other browsers show a similar color shift, until you look at
the page on your tablet—and it’s perfect! What the heck?
After a bit of surfing, you find a site that recommends spec-
ifying something called sRGB in Photoshop. You don’t know
exactly what that means, but you tweak things and save a new
graphic file. It fixes the problem in Safari, but the color is still
off in Chrome. No matter what you do, you can’t get the CSS
and image colors to match.
This inconsistency bugs you. The doctor may have question-
able taste in colors, but his eyesight is perfect. He’s going to see
these shifts and ask what’s happening.
Problem is, you really have no clue.

WHAT IS COLOR MANAGEMENT?

At the most basic level, color management is a way to specify the
range of colors a device can represent.
Many designers and developers, including your author, have
found ways to wrangle color without truly understanding the
underlying technology. We’ve leaned on simple standards,
choosing three bytes of data for the primary colors of light:
red, green, and blue. When we specified R=255, G=0, and B=0,
we expected—and could expect—to see red on our displays. We
could pretty much ignore anything else about color, whether
we were designing for the web or a native app.

C o p i n g w i t h C o lo r s 1
Fig 1.1: The site header is an eyeful. What’s worse, the colors are shifting.

This worked for a time, but as you saw with Dr. Eyeful, our
basic RGB values don’t translate the same way across devices.
To find out why, let’s take a quick trip through history.

Coping with too many choices

Color management isn’t new. Photographers and print design-
ers have dealt with calibrating colors for decades. It all started
when people took images captured on film and output them
on paper with ink. Sounds straightforward, right? But you
had different kinds of film stock, along with paper weight and
finish. The variety of inks was astounding; the huge collection
of Pantone colors, for instance, got its start in the 1960s as
colored liquids.
Such technological diversity made it hard to keep imag-
ery consistent as it made its way from the camera lens to the

2 M a k i n g S e n s e o f C o lo r M a n ag e m e n t
printed page. Photographers, designers, and printers knew
they couldn’t rely solely on an RGB or a CMYK color space
to get consistent results. Their workflow included another
component, which profiled the color capabilities of their equip-
ment—and adjusted colors accordingly. (This component was a
complex mathematical model based on the human eye: a color
profile. We’ll get to this soon.)
We face the same scenario now, the same variety in tech-
nology. The digital sensor on a cheap cell phone camera can’t
capture as many colors as a high-end, pricey DSLR. Similarly,
an LCD display from only a few years ago looks absolutely hor-
rible when compared to the Retina display on an iMac. OLED
technology on wristwatches is a new animal entirely.
If you want your reds, greens, and blues to look right across
browsers or apps, you need to pay attention to how you specify
colors in your tools. While you often won’t have much choice
about how color is processed by a platform, you will get unpre-
dictable results if you ignore the requirements of whatever
environment you’re working in—which is what happened with
Dr. Eyeful’s website.
Understanding color management yields dividends. You’ll
ensure branding and other color elements in your interface look
the best they can. And think how many elements are involved
in a website—all the interface graphics, hero images, and other
bits—wouldn’t it be great if you could reuse those assets, with-
out modification, in a mobile app?
You can, once you know how your workflow manages color
and your target platforms present the images. As a first step, let’s
examine how colors got their start on our computers.

Old-fashioned color
In the late 1970s, color began to appear on computer displays.
These early machines had little memory—sometimes as small
as 4,096 characters total. Processors were also limited, and
accessed memory eight bits at a time, giving programmers a
minimum value of 0 and a maximum value of 255.

C o p i n g w i t h C o lo r s 3
Fig 1.2: Two green bulbs produce different colors.

Although machines have become much more powerful,

specifying color with eight bits has stayed with us: three RGB
channels together produce 16,777,216 unique colors, a range that
closely matches what our eyes can see. Some apps use 16-bit
color, but this extended range mostly benefits editing images.
We’ll use 8-bit values throughout this book; to get more
accurate color, we’ll add another type of data.

Green light bulbs

Imagine we have two green light bulbs, made by two
manufacturers.
When the bulbs are fully lit, even if you’re affected by certain
forms of color blindness, you can tell the colors differ slightly
(Fig 1.2). This is unsurprising, since the manufacturers used
separate processes and materials.
But your computer isn’t so smart. All it knows is that G=0
means the bulb is off, G=127 is at half brightness, and G=255
is fully on.
Now imagine two displays made with these light bulbs. I
have the one on the left, and you have the one on the right.
When I specify G=255, I see the color on the left. But when I
send you a file with that G=255, you see the color on the right.
Whoops. It might seem minor, but it wouldn’t be to a client
like Starbucks, Heineken, or John Deere, which takes green
very seriously. You could lose hours perfecting the green on
your screen and have the CEO see something else. That’s just
the tip of the iceberg: every customer who visits your client’s
site can have a completely different light bulb.

4 M a k i n g S e n s e o f C o lo r M a n ag e m e n t
 As photographers and print designers learned, we can’t
solely rely on the values for primary colors. We also need infor-
mation, or a color profile, from the manufacturer that explains
how they display the color green.

COLOR PROFILES
Anything that can capture or display an image—your camera,
your monitor, etc.—has a set of values that describes the range
of colors available for use. This set of data is the color profile,
and it comes in a format defined by the International Color
Consortium (ICC).
Creating a profile involves a lot of science. If you’re a man-
ufacturer of green light bulbs, you’ll take an expensive piece of
equipment, a color spectrophotometer, to measure the amount
of visible light produced at different color wavelengths. You
then use that data to generate a mathematical model—the pro-
file, which is specific to the device: my iMac display profile
would look terrible on your MacBook.
Some manufacturers choose to use a standard profile for
their bulbs. Two of the most popular are Adobe RGB and stan-
dard RGB (sRGB):

• Adobe introduced its profile in 1998 to represent CMYK

printer colors on an RGB display. It’s a favorite among pho-
tographers, because its wide range of colors produces better
images in post-processing.
• Made around the same time, sRGB mimics typical viewing
conditions in a home or office, with a smaller range of colors
(making it cheaper and easier to produce). Its ubiquity makes
it the color space for defining web standards.

Cameras often support Adobe RGB (or ProPhoto, the color

space developed by Kodak), while many displays (like that bar-
gain LCD at the electronics store) conform to sRGB.
Let’s see what these two standard profiles look like (Fig 1.3).

C o p i n g w i t h C o lo r s 5
Fig 1.3: The color ranges of two profiles: sRGB (left) and Adobe RGB (right).

The curved shape with a flat side represents all the colors
the human eye can see—it never changes. Each triangle rep-
resents a color profile; the X-corners are fully saturated red,
green, and blue.
The black circle marks the brightest green the profile can
produce. It’s like our green light bulbs: sRGB’s green has a
yellow tint, while Adobe’s is closer to the ideal color our eyes
see. (As you’ll find out in a moment, the weird shape springs
from our physiology!)
The size, or gamut, of each triangle reflects the range of pos-
sible colors. The larger triangle, Adobe RGB, has a wider gamut
(which is why the profile is so popular among photographers).
Keen observers will note black isn’t present. That’s because
we cheated a bit and showed only two dimensions. When we
add color saturation, the curved shape lies on its side and a fully
saturated white appears as a peak, with black as a base (Fig 1.4).
That pointy shape is everything our computers need to take
the raw numeric values in our files and display them so we all
see a similar color.

6 M a k i n g S e n s e o f C o lo r M a n ag e m e n t
 Fig 1.4: sRGB color profile with saturation
as a third dimension.

A profile to fit your eye

There’s no getting around it: color management is complex.
At its core, after all, is a numerical model for the human eye.
It doesn’t help that behind this heavy-duty science is a slew
of acronyms. Let me take a minute to break down a few names
and concepts that led to my first “aha!” moment in learning
about color management. First up, the Commission Interna-
tionale de l’Eclairage, the global authority on light and color.
If you don’t speak French, that’s International Commission on
Illumination—but everyone refers to this group as CIE.
Our retinas contain three types of cone cells that detect light.
Each type senses a different part of the color spectrum: blues
at short wavelengths, greens in the middle, and reds at longer
wavelengths. The curved boundary in the shapes you saw is a
plot of chromatic wavelengths visible to our eyes.
Our brains take stimuli from these cells and combine them
to give us color perception (and the official term, tristimulus.)
The scientists at CIE codified this biology into a mathematical
model called CIEXYZ in 1931.
Over the next 45 years, understanding of the human eye
improved. Researchers found that somewhere between the
optic nerve and the brain, stimuli are categorized according to
degrees of lightness, red/green, and blue/yellow. This led to a
second key color model, CIELAB.

C o p i n g w i t h C o lo r s 7
 Don’t let the old age of CIELAB and CIEXYZ trick you: both
are still the standards to put your eye’s behavior into numbers
our computers can process:

• Since it takes into account how our brains perceive color, a

LAB color space is better at reproducing tones and relative
color values. LAB is ideal for color created with inks.
• The XYZ color space is based on the eye’s response to stim-
ulation across the visible light spectrum. This works well
for specifying an exact color, even though it ignores that our
vision isn’t completely uniform. XYZ is ideal for devices that
emit light, like a computer display.

The eye’s color profile makes everything work

Phew: you’ve dipped your toe into the complicated waters of
color science! The important thing to know is both CIE color
spaces are device-independent: their only dependence is on our
bodies (which thankfully don’t change as fast as technology).
The eye’s freedom from hardware is crucial to moving colors
between profiles.
Converting directly between devices would be painful: every
device would need to know about every other device in the
world. Your display would need to know about all kinds of
cameras before it could work with photos. Same with your
printer and editing software. Think about how many models
of cameras, printers, and displays exist, and you’ll see this sit-
uation would soon become unwieldy.
Instead, we base conversion on the eye, so each device
only needs to specify how to transform color to and from the
eye’s color space (Fig 1.5). The raw numeric color values differ
depending on the document or device, but they all look the
same. Because of their central role, the CIE models are referred
to as profile connection spaces.
You’ll probably never use CIEXYZ or CIELAB directly, but
every time a pixel moves from one place to another, one of
the eye’s color spaces will play a part in making sure that pixel
looks its best.

8 M a k i n g S e n s e o f C o lo r M a n ag e m e n t
 Document Display

102, 0, 204 121, 1, 204

sRGB Your Profile

CIEXYZ / CIELAB
Camera Printer

101, 40, 184 0, 71, 74, 0

ProPhoto RGB Generic CMYK

Fig 1.5: Using color profiles, the eye perceives the same color for different raw values.

Profiles are everywhere

Color profiles are relatively small pieces of data, which makes
them easy to embed in image files and design documents. A file
with this profile information can be handled predictably by any
editor or presentation mechanism (including things as diverse
as printers and web browsers).
File formats that can embed profiles include popular ones
for web images (JPEG, PNG, SVG) and documents (PSD, PDF).
Digital negatives (DNG) allow profiles too, and color profiles
may even someday become part of your CSS.
The color profile information can also be stored in a separate
file. These files typically accompany hardware peripherals like
displays, printers, and cameras, where it’d be impractical to
embed the data.
Editors and other products that manipulate design use a
color profile that describes the range of colors available while
you’re creating artwork. This profile is often referred to as the

C o p i n g w i t h C o lo r s 9
 working space. Some apps, like Photoshop, let you choose the
working space. Other apps use a predefined color space, usually
a standard one like sRGB.
If you’re curious, you can explore all the color profiles
installed on your Mac using the ColorSync Utility in the Appli-
cations > Utilities folder. Under Profiles, you’ll see a long list,
sourced from Apple, Adobe, the display manufacturer, and
many others. When you select a profile, its graphic represen-
tation appears, along with other details.

Read the rest of this chapter and more when you buy the book!

10 M a k i n g S e n s e o f C o lo r M a n ag e m e n t