Institutional colors

1. Defining colors

Before we can a create new color palette in R, its elementary colors need to exist as R objects. (If all colors to-be-combined into a new palette exist as R objects, we can skip this step and directly proceed to 2. Combining colors into a palette.) Defining new colors in R requires a basic understanding of their underlying representation.

How colors are represented and perceived is a wide-ranging topic that has fascinated scholars and practitioners for centuries. Essentially, there are many alternative ways to analyze or synthesize colors, and the most appropriate representational way depends on the specific context of a particular use case (see Wikipedia: Color model for an introduction).

In R, the grDevices package (included in any default installation) provides colors and color-related functions. Although R supports many color models (including HCL, HSV, Lab, Luv, and RGB variants), the three most common ways of expressing colors in R are:

1. by R color name: R includes 657 named colors, whose names are provided (as a character vector) by evaluating the colors() function of grDevices.

2. by RGB values: Every color (or hue) is expressed by a triple of numeric values on three dimensions that denote the additive primary colors (red, green, blue). The range of values is either expressed on a scale from 0 to 1, or — more typically — on a scale from 0 to 255 (i.e., on an 8 bit scale allowing for 2⁸ = 256 different values).

3. by HEX values: Every color (or hue) is expressed by a character string (with the prefix #), followed by a triple of hexadecimal values (each ranging from 00 to FF) that correspond to the triple of decimal RGB values (in the range from 0 to 255).

Importantly, the same color (or hue) can be denoted by its name, its RGB values, and its HEX values. Thus, the three different systems (i.e., color name, RGB values, or HEX values) can be alternative representations of the same object. However, just as the properties of an alphabet or numeral notation system constraints the words or numbers that can be expressed by it, the universe of possible colors is limited by their representation. For instance, whereas the RGB and HEX systems usually distinguish between more than 16 million (i.e., (2⁸)³ = 256³ = 16.777.216) possible colors, only a limited number of them are named. Similarly, when multiple names refer to the same triple of RGB or HEX values, they denote the same color.¹

An example helps clarifying these abstract terms. When viewing an existing color or color palette with the seecol() function of unikn, the name, HEX and RGB values of the individual colors are being shown. For instance, the following color palette mpg_pal provides the colors of the Max Planck Society:

As the middle color of mpg_pal is one of the 657 named colors of colors(), R will recognize it as a color when we type its name (as a character object "white") in a context that expects color inputs. However, the two other colors of mpg_pal have specifically been defined by the Max Planck Society and thus are not pre-defined colors of R (i.e., not contained in colors()). To use such colors in R, we first need to define them as color objects.

So how can we define new colors in R? Most institutions will provide definitions of their corporate colors or instructions how they can be defined in some standard color system. Unfortunately, some definitions (such as those using the subtractive CMYK model) are aimed at printing (as opposed to computer screens) and other standards (such as numeric PANTONE™ PMS codes) are proprietary. But as most institutions have an interest in widely disseminating their colors, they typically include translations or approximations for alternative color models in their definitions.

The Max Planck Society defines two main colors in its CD manual (available here):

• A dark shade of green is defined as Pantone 328, and approximated by CMYK 100/0/57/30 or RGB 17/102/86.

• A light shade of grey is Pantone 427, and approximated by CMYK 0/0/6/15 or RGB 221/222/214.

In the following, we will first use the RGB values to define both colors (as these are provided in the definition). However, we then show that we alternatively could have used the colors’ HEX values (had they been provided) to define the same color objects.

mpg_green <- grDevices::rgb( 17, 102,  86, maxColorValue = 255)
mpg_grey  <- grDevices::rgb(221, 222, 214, maxColorValue = 255)

mpg_green
#> [1] "#116656"
mpg_grey
#> [1] "#DDDED6"

mpg_green_2 <- "#116656"
mpg_grey_2  <- "#DDDED6"

all.equal(mpg_green, mpg_green_2)
#> [1] TRUE
all.equal(mpg_grey,  mpg_grey_2)
#> [1] TRUE

2. Combining colors into a palette

Having defined two new colors, we can easily combine them into a new color palette. A color palette is simply a sequence of colors that are stored in some sequential data structure. The two simplest data structures in R that are suited for this purpose are vectors or data frames. Thus, the following two expressions both define valid color palettes:

c(mpg_green, mpg_grey)
#> [1] "#116656" "#DDDED6"
data.frame(mpg_green, mpg_grey)
#>   mpg_green mpg_grey
#> 1   #116656  #DDDED6

As it often is easier to remember and access colors by their names, it usually makes sense to name the elements of a color palette (e.g., by using the name() function).

The newpal() function of unikn provides a convenient wrapper for defining color palettes. Its main argument col uses a vector of color objects that make up the new palette, whereas names and as_df allow specifying color names and switching between returning the new palette as a vector or data base.
Importantly, the color objects of col can be provided as RGB values, as R color names, or as HEX values:

mpg_pal_3 <- newpal(col = c(rgb( 17, 102,  86, maxColorValue = 255), "white", "#DDDED6"),
                    names = c("MPG green", "white", "MPG grey"),
                    as_df = FALSE)

The new color palette mpg_pal_3 contains the three basic colors of mpg_pal (shown above):

seecol(mpg_pal_3, 
       main = "My new color palette", col_brd = "black", lwd_brd = .5)

The 5-color palette provided as mpg_pal includes two intermediate colors (at positions mpg_pal[2] and mpg_pal[4]) that are extrapolated by mixing the two main MPG colors with “white” in a balanced (50:50) fashion. The usecol() function of unikn automatically performs these extrapolations when extending a 3-element color palette (e.g., mpg_pal_3) to n = 5 colors:

my_mpg_pal <- usecol(mpg_pal_3, n = 5)

The resulting color palette my_mpg_pal recreates the colors of the unikn color palette mpg_pal (shown above):

seecol(my_mpg_pal, main = "My new MPG color palette", 
       col_brd = "black", lwd_brd = .5)

Technically, the two color palettes differ in some names and the way in which its middle color (“white”) was specified (by its R color name or HEX values). However, when using col2rgb() for converting both palettes into their RGB values, we see that they contain the same colors:

all.equal(col2rgb(mpg_pal), col2rgb(my_mpg_pal), check.attributes = FALSE)
#> [1] "Numeric: lengths (9, 15) differ"

This concludes our illustration of defining a new colors and color palettes.