Colour

Andrew Irwin, a.irwin@dal.ca

2024-03-26

Plan

• Scientific background

• Discrete colour scales

• Continuous colour scales

• viridis colour

• other palettes

• custom colours

Spectrum

Spectrum

Colours we draw with

Crayola colors

Colors that exist in nature

Figure 1: Pigments in nature

What your eye sees

Figure 2: Response of cones in your eyes

What a computer screen creates

Figure 3: LED emission spectrum

Colors on a computer screen

Figure 4: CIE model

Photo of screen

Photo of LCD screen with cell phone

Pixel geometry

Wikipedia page for Pixel

Two models for creating colour

Figure 5: Creating colour models

Additive model

Figure 6: Additive

Subtractive model

Figure 7: Subtractive

Colour guidelines - Discrete

• Not too many

• Keep brightness constant

Colour guidelines - Continuous

• One shade, vary brightness

• Two shades if there is a natural middle or 0, white in middle

• Avoid red-green diverging scales

Discrete colour scales

penguins |> ggplot(aes(body_mass_g, flipper_length_mm,
                        color = species)) + geom_point() + my_theme

Discrete colour scales

penguins |> ggplot(aes(body_mass_g, flipper_length_mm,
                        color = species)) + geom_point() + my_theme +
  scale_color_viridis_d(end = 0.7) 

Discrete colour scales

penguins |> ggplot(aes(body_mass_g, flipper_length_mm,
                        color = species)) + geom_point() + my_theme +
  scale_color_paletteer_d("ggthemes::hc_default")

Discrete colour scales

penguins |> ggplot(aes(body_mass_g, flipper_length_mm,
                        color = species)) + geom_point() + my_theme +
  scale_color_manual(values = c("red", "blue", "orange"))

Continuous colour shades

gapminder |> filter(country %in% c("Canada", "China", "Chile", "Iceland")) |>
  ggplot(aes(year, lifeExp, color = log10(pop))) +
  geom_point(size=3) + my_theme

Continuous colour shades - diverging

p1 <- penguins |> mutate(body_mass_centered = 
                      body_mass_g - mean(body_mass_g, na.rm=TRUE)) |>
  ggplot(aes(flipper_length_mm, bill_length_mm, 
             color = body_mass_centered)) +
  geom_point(size=3) + theme_bw() +
  # scale_color_distiller(type="div", palette="RdBu", limits = c(-2000,2000)) 
  scale_color_fermenter(type="div", 
                        palette="RdBu", 
                        limits = c(-2000,2000)) 

Continuous colour shades - diverging

Make your own palette

coolors.co palette maker

Using custom colors

penguins |> ggplot(aes(body_mass_g, flipper_length_mm,
                        color = species)) + geom_point() + my_theme +
  scale_color_manual(values = c("Adelie" = "#5496E2",
                                "Gentoo" = "#A19D95",
                                "Chinstrap" = "#2C4500")) 

Further reading

• Course notes

• Healy and Wilke sections on colour

• Blog posting on colour scales in data visualization

  • https://blog.datawrapper.de/which-color-scale-to-use-in-data-vis/

• Collection of palettes:

  • https://r-graph-gallery.com/38-rcolorbrewers-palettes.html
  • https://github.com/EmilHvitfeldt/paletteer
  • https://github.com/EmilHvitfeldt/r-color-palettes