Week4 Color
Week4 Color
• The red, green, and blue axes of the RGB color space are not
useful as separable channels.
HSL
• The hue-saturation-lightness system is heavily used by
artists and designers.
• Hue: what we normally think of as pure colors that are
not mixed with white or black
• Saturation: the amount of white mixed with that pure
color (e.g., pink as a partially desaturated red)
• Lightness: the amount of black mixed with a color
• Very similar to Hue-Saturation-Value (HSV) space.
HSL
• Pseudo-perceptual: It does not truly
reflect how we perceive color.
• Lightness (L) is different from how
we perceive luminance.
• Example: six different hues
arranged in order of their luminance
(The true luminance is a somewhat
better match with our perceptual
experience.)
Spectral Sensitivity
• The spectral sensitivity of our eyes to
luminance depends on the
wavelength of the incoming light.
Colormaps
Color Schemes
• When choosing color schemes, we will want mappings from data to color that are
not just numerically but also perceptually uniform.
• Our perception is not uniform across the space of possible colors.
• The range of chroma we are able to see depends strongly on luminance.
• Numerically equal gaps between a sequences of reds (say), are perceived differently from
the same gaps mapped to blues
Color Schemes
When we want to use different
expressions of a preattentive
attribute (e.g., hue) to separate
objects into different groups, we
should select expressions of that
attribute that vary significantly
from one another.
Color Schemes
• Different color spaces have been defined and standardized in ways that account
for these uneven or nonlinear aspects of human color perception.
• The body responsible for this is Commission Internationale de l’Eclairage
(International Commission on Illumination).
• R and ggplot make these features available to us.
Color Schemes
• R provides color models and color spaces that try to
make the mappings from data to color numerically and
perceptually uniform.
• Example: A series of sequential gradients using the HCL
(Hue-Chroma-Luminance) color model (palettes
generated from R’s colorspace library) ►
• Goal: to generate a perceptually uniform scheme, where hops
from one level to the next are seen as having the same
magnitude.
Colormaps
• A colormap specified a mapping between colors and data values (a visual
encoding with color)
• Colormapping: also called Pseudocoloring
• Pseudocoloring is the technique of representing continuously varying map values using a
sequence of colors. Choropleth map
• Used widely for astronomical radiation charts, medical imaging, and many other
scientific applications.
Colormaps
• The major design choice for colormap construction: whether the intent is to
distinguish between categorical attributes or to encode ordered attributes.
• Sequential ordered colormaps show a progression of an attribute from a minimum to a
maximum value
• Diverging ordered colormaps have a visual indication of a zero point in the center where the
attribute values diverge to negative on one side and positive on the other.
• Bivariate colormaps are designed to show two attributes simultaneously.
Colormaps
• Taxonomy of colormaps mirrors the taxonomy of data types.
• Categorical colormaps
• Ordered colormaps
• Sequential
• Diverging
• For categorical data, they are the next best channel after spatial position.
• Typically designed by using color as an integral identity channel to encode a
single attribute.
• Easily nameable colors are desirable, both for memorability and ability to discuss
them using words.
Categorical Colormaps
• For some uses, the colors should be close in luminance to avoid major differences
in salience and to ensure that all can be seen against the same background.
• For other uses, colors should be sufficiently different in luminance that they can be
distinguished even in black and while.
• Colormaps for small regions such as lines should be highly saturated, but large
regions such as areas should have low saturation.
• ColorBrewer: A good resource for creating colormaps
Categorical Colormaps
• Two design choices when faced
with the problem of discriminability
mismatch:
1. Reduce the number of bins
1) Through a deliberate data
transformation that takes into
account the nature of the data and
task
2) Filter the attributes to only encode a
small set of the most important ones
with color, and aggregate all of the
rest into a new category of other
Categorical Colormaps
• Two design choices when faced with the problem
of discriminability mismatch:
2. Use a different visual encoding idiom that uses other
visual channels instead of, or in addition to, the color
channel alone
Ordered Colormaps
• An ordered colormap is appropriate for encoding ordinal or quantitative
attributes.
• A sequential colormap ranges from a minimum value to a maximum value.
• A diverging colormap has two hues at the endpoints and a neutral color as a midpoint (e.g.,
white, gray, or black, or a high-luminance color such as yellow)
• How many unique hues to use in continuous colormaps depends on what level of
structure should be emphasized: the high-level structure, the middle range of local
neighborhoods, or fine-grained detail at the lowest level.
Ordered Colormaps
• Rainbow colormap
• Advantage: people can easily discuss specific subranges because the differences are easily
nameable
• Problems:
• Hue is used to indicate order
• The scale is not perceptually linear
• Fine detail cannot be perceived with the hue channel
Ordered Colormaps
• Monotonically increasing luminance colormaps: where the multiple hues are
ordered according to their luminance from lowest to highest (the figure on right)
• The varying hues allow easy segmentation into categorical regions.
• Luminance supports both high-level distinctions and low-level structure perception.
Ordered Colormaps
• The standard rainbow colormap is perceptually nonlinear.
• A segmented rainbow colormap is a fine choice for categorical data with a small
number of categories.
• Deliberately bin the data explicitly rather than relying on the eye to create bins
Bivariate Colormaps
• Univariate colormaps visually encode a single
attribute.
• Bivariate colormaps encode two separate
attributes.
• When one of the two attributes is binary: create a
colormap with two families of colors by fixing a base
set of hues and varying their saturation
• When both attributes are categorical with multiple
levels, results will be poor.
References
• Few, S. (2009). Now you see it: simple visualization techniques for quantitative
analysis.
• Munzner, T. (2014). Visualization analysis and design. CRC press.
• Stone, M. (2003). A field guide to digital color. CRC Press.
• Tufte, E. R. (1990). Envisioning information. Graphics Press LLC.
• Ware, C. (2019). Information visualization: perception for design.