DIGITAL COLOR: THEORY AND PRACTICE
Report on course "A Field Guide to Digital Color" organized
Maureen Stone and Pauline Ts'o
by Ben Wyrick
24 July 2002
Color is a powerful, subtle and sometimes fickle tool. Add a little
too much red to your blue, and suddenly you're looking at something
entirely different than what you started with. How does color work,
and what can be done to keep it consistent through digital and mechanical
processes? Maureen Stone and Pauline Ts'o
shared their knowledge of color in a course entitled "A Field
Guide to Digital Color."
Stone began the course by explaining how the human visual system
reads color. Light enters the eye, strikes the retina, and is absorbed
by cone cells. The light is then transmitted to the brain via the
optic nerve, where it is interpreted. An important lesson to draw
from this model is that an observer may not actually see the true
color of an object--the observer's cones may not respond to the
exact wavelength. Furthermore, the brain interprets the
signal received from the cones, and that interpretation could be
misleading, hence optical illusions.
Stone explained the three types of cones: long, medium and short.
Long cones are sensitive to red colors (which have long wavelengths),
medium cones are sensitive to green, and short cones are sensitive
to the shortest visible wavelengths, blue colors.
Cone cells are found mostly at the center of the eye and are unevenly
distributed. Blue cones are rarer than red and green. Rod cells,
which are not sensitive to color, are located around the periphery
of the eye, and are responsible for night vision. According to Stone,
rods and cones usually do not function at the same time.
The eye has a blind spot where the optic nerve meets the retina.
However, this blind spot is seldom noticed since our eyes are continuously
scanning the world and integrating between snapshots.
Understanding the theory behind the human visual system is useful
in developing practical color models and standards for computer
graphics. For example, the phosphors inside televisions and computer
monitors are red, green and blue--just the colors our cones are
Humans are capable of detecting millions of colors--more than desktop
printers, monitors, projectors and printing presses are capable
of producing. To complicate matters, different devices produce different
ranges of colors. Monitors mix light and printers mix ink, and results
often vary. For example, your monitor might be capable of displaying
a color like "fire engine red," but to your printer it
looks more like "dry rust red."
To combat the problem of accurate color reproduction, sets of device-independent
standards are being created. One such standard was created by the
International Color Consortium (ICC) and has been widely accepted,
Stone explained. The ICC standard is based on experiments conducted
by the Commission Internationale de l'Eclairage (CIE), which tested
how people perceive color.
Once a standard color space like ICC is agreed upon, devices can
convert their colors to colors within the space. Problems arise
when a device's colors are exotic flavors of the standard set. These
color mismatches are resolved in a process called gamut mapping,
which uses algorithms to convert extreme colors to standard colors.
When everything works, what you see on your monitor is what you
get in print, thanks to an understanding of the human visual system
and an agreed-upon color space.
* * *
A thorough knowledge and exacting use of color is critical to a
special effects production house like Rhythm & Hues Studios.
Pauline Ts'o of Rhythm & Hues talked about how color can lead
the eye and strike a mood. Preserving the richness and subtlety
of color is taken very seriously at special effects houses, which
use high-end hardware and software to get the right look. Ts'o spoke
on the particulars of Rhythm & Hues' color pipeline for both
video/broadcast work and film work.
First, live action footage is scanned into a proprietary file format
called Rll, which uses 16-bit color channels. Most consumer image
editing devices use 8-bit color channels and as a result are not
able to store and manipulate a rich spectrum of colors. 8-bit schemes
can also introduce banding artifacts due to their limitations. The
Rll format stores colors on a logarithmic scale as opposed to the
more commonly used linear scale. According to Ts'o, a logarithmic
representation of color allows more gradations in dark colors, which
film is more sensitive to. Rendered frames are created in a linear
space and then special software is used to correctly map them to
the logarithmic color space used by monitors.
Calibrating monitors is a routine task at Rhythm & Hues and
Ts'o joked about users being severely warned not to change the settings.
Employees work in darkened rooms to avoid bright light reflecting
on monitor glass. Ts'o admitted to calibrating her home television
set by watching Rhythm & Hues commercials when they are broadcast.
Working with film stock requires extra steps to ensure color correctness,
Ts'o explained. Positive match clips and wedge clips are special
referencing tests made to keep color consistent between the client's
live-action footage and the studio's special effects footage. Ts'o
also says her studio works closely with the film processing lab
to make sure chemicals are fresh and their equipment is properly
When everything works, the end result of understanding the human
visual system and maintaining a pure digital color through the many
steps of a production pipeline is a faithful representation
of the director's vision.Links: The International
Color Consortium: www.color.org