ggplot2 Themes

Custom Color and Fill Scales

xaringanthemer includes monotone color and fill scales to match your ggplot2 theme. The scale functions all follow the naming pattern scale_xaringan_<aes>_<data_type>(), where <aes> is replaced with either color or fill and <data_type> is one of discrete or continuous.

These scales use colorspace::sequential_hcl() to create a sequential, monotone color scale based on the primary accent color in your slides. Color scales matching the inverse slides are possible by setting the argument inverse = TRUE.

ggplot(diamonds, aes(x = cut)) +
  geom_bar(aes(fill = ..count..), show.legend = FALSE) +
  labs(x = NULL, y = "Count", title = "Diamond Cut Quality") +
  theme_xaringan() +
  scale_xaringan_fill_continuous()

ggplot(mpg, aes(x = hwy, y = cty)) +
  geom_count(aes(color = ..n..), show.legend = FALSE) +
  labs(x = "Highway MPG", y = "City MPG", title = "Fuel Efficiency") +
  theme_xaringan_inverse() +
  scale_xaringan_color_continuous(breaks = seq(3, 12, 2), inverse = TRUE, begin = 1, end = 0)

In general, these color scales aren’t great at representing the underlying data. In both examples above, the color and fill scales duplicate information displayed via other aesthetics (the height of the bar or the size of the point). I recommend using these scales primarily for style, although the scales can be more or less effective depending on your color scheme.

The scales come with a few more options:

• Choose a different primary color using the color argument.

• Use the inverse color slide theme color with inverse = TRUE (only applies when color is not supplied).

• Invert the direction of the discrete scales with direction = -1.

• Control the range of the continuous color scale used with begin and end. You can also invert the direction of the continuous color scale by setting begin = 1 and end = 0.

Automatically match slide and plot fonts

xaringanthemer uses the showtext and sysfonts packages by Yixuan Qiu to automatically download and register Google Fonts for use with your ggplot2 plots.

In your slide theme, use the <type>_font_google argument with the google_font("<font name>") helper (or the default xaringanthemer fonts) and theme_xaringan() will handle the rest. In our demo theme, we used style_mono_accent() with

• text_font_google = google_font("Kelley Slab") and
• header_font_google = google_font("Oleo Script").

g_diamonds_with_text <-
  g_diamonds +
  geom_text(aes(y = ..count.., label = format(..count.., big.mark = ",")),
    vjust = -0.30, size = 8, stat = "count") +
  labs(x = "Cut", y = "Count")

g_diamonds_with_text + theme_xaringan()

theme_xaringan() applies the header font to the plot and axis titles and the text font to the axis ticks labels and any text geoms or annotations.

Manually specify plot fonts

You can also specify specific fonts for your plot theme. Both text_font and title_font in theme_xaringan() and theme_xaringan_inverse() accept google_font()s directly.

g_diamonds_with_text +
  theme_xaringan(
    text_font = google_font("Ranga"),
    title_font = google_font("Holtwood One SC")
  )

Using fonts not in Google Fonts

If you want to use a font that isn’t in the Google Fonts collection, you need to manually register the font with sysfonts so that it can be used in your plots.

I found a nice open source font called Glacial Indifference by Alfredo Marco Pradil available at fontlibrary.org. In my theme style function, I would use

style_mono_accent(
  text_font_family = "GlacialIndifferenceRegular",
  text_font_url = "https://fontlibrary.org/face/glacial-indifference"
)

but sysfonts won’t know where to find the TTF font files for this font.

To register the font with sysfonts, we use sysfonts::font_add(), but first we need to download the font file — the sysfonts::font_add() function requires the font file to be local.

By inspecting the CSS file at the link I used in text_font_url, I found a direct URL for the .ttf files for GlacialIndifferenceRegular. I’ve included the code I used to download the font to a temporary file below, but in case the URL breaks, I’ve included Glacial Indifference in the xaringanthemer package.

font_url <- file.path(
  "https://fontlibrary.org/assets/fonts/glacial-indifference/",
  "5f2cf277506e19ec77729122f27b1faf/0820b3c58fed35de298219f314635982",
  "GlacialIndifferenceRegular.ttf"
)

font_temp <- tempfile()
download.file(font_url, font_temp)

# Path to the local custom font file
font_temp <- system.file(
  "fonts/GlacialIndifferenceRegular.ttf", package = "xaringanthemer"
)

# Register the font with sysfonts/showtext
sysfonts::font_add(family = "GlacialIndifferenceRegular", regular = font_temp)

# Now it's available for use!
g_diamonds_with_text +
  theme_xaringan(
    text_font = "GlacialIndifferenceRegular",
    title_font = "GlacialIndifferenceRegular"
  )

R Markdown

To use the showtext package in R Markdown, knitr requires that the fig.showtext chunk option be set to TRUE, either in the chunk producing the figure or globally in the document.

xaringanthemer tries to set this chunk option for you, but in some circumstances it’s possible to call theme_xaringan() in a way that xaringanthemer can’t set this option for you. When this happens, xaringanthemer will produce an error:

Error in verify_fig_showtext(fn) : 
  To use theme_xaringan_base() with knitr, you need to set the chunk
  option `fig.showtext = TRUE` for this chunk. Or you can set this option
  globally with `knitr::opts_chunk$set(fig.showtext = TRUE)`.

If you find yourself facing this error, follow the instructions and choose one of the two suggestions:

1. Add fig.showtext = TRUE to the chunk producing the figure

2. Or set the option globally in your setup chunk with knitr::opts_chunk$set(fig.showtext = TRUE).

MacOS

On MacOS, you’ll need to have xquartz installed for sysfonts to work properly. If you use homebrew, you can install xquartz with

brew cask install xquartz

In RStudio

showtext and RStudio’s graphic device don’t always work well together. Depending on your version of RStudio, if you try to preview plots that use theme_xaringan(), the fonts in the preview will still be the default sans font or you may not see a plot at all.

To work around this, open a new quartz() (MacOS) or x11() (Windows/Unix) plot device. The plots will then render in a separate window. I usually create a quartz() device with a similar size ratio to my slides.

## On Windows
# x11(width = 16 * 2/3, height = 9 * 2/3)

## On MacOS
quartz(width = 16 * 2/3, height = 9 * 2/3)
## run plot code to preview in separate window
dev.off() # call when done to close the device