Save PDFs with custom fonts

{camcorder} doesn’t only work with raster images such as PNG and JPEG but also PDF files. There are some benefits of using the PDF format when saving ggplot output: vector graphics are lossless, can be converted easily in raster formats of any resolution and also be manipulated afterwards in a vector design tool. Furthermore, the PDF format often supports the use of custom fonts (when using the Cairo device)1.

To automatically save your graphics in PDF format, just change the device setting to cairo_pdf2 when recording your ggplot code:

library(ggplot2)
library(camcorder)

gg_record(
  dir = file.path(tempdir(), "recording"), 
  device = cairo_pdf, # we need to set the Cairo device
  width = 8,
  height = 5
)

Mac users should ensure that XQuartz is installed which is needed to use the cairo pdf device.

To supply custom fonts in R, the respective font needs to be installed locally on the system.

You can make sure the font file is installed by using the system_fonts() from the latest standard, the {systemfont} package:

systemfonts::system_fonts()
#> # A tibble: 148 × 9
#>    path                   index name  family style weight width italic monospace
#>    <chr>                  <int> <chr> <chr>  <chr> <ord>  <ord> <lgl>  <lgl>    
#>  1 /usr/share/fonts/type…     0 URWG… URW G… Book… normal norm… FALSE  FALSE    
#>  2 /usr/share/fonts/true…     0 Libe… Liber… Bold  bold   norm… FALSE  FALSE    
#>  3 /usr/share/fonts/type…     0 Nimb… Nimbu… Bold… bold   norm… TRUE   TRUE     
#>  4 /usr/share/fonts/true…     0 Libe… Liber… Bold  bold   norm… FALSE  TRUE     
#>  5 /usr/share/fonts/open…     0 Nimb… Nimbu… Bold… bold   norm… TRUE   FALSE    
#>  6 /usr/share/fonts/type…     0 Nimb… Nimbu… Regu… normal norm… FALSE  FALSE    
#>  7 /usr/share/fonts/type…     0 Nimb… Nimbu… Ital… normal norm… TRUE   FALSE    
#>  8 /usr/share/fonts/type…     0 Nimb… Nimbu… Bold… bold   norm… FALSE  FALSE    
#>  9 /usr/share/fonts/type…     0 D050… D0500… Regu… normal norm… FALSE  FALSE    
#> 10 /usr/share/fonts/X11/…     0 P052… P052   Bold… bold   norm… TRUE   FALSE    
#> # ℹ 138 more rows

You can simply filter this tibble for any font:3

systemfonts::system_fonts() |> 
  dplyr::filter(grepl("Dyna", family)) |>
  dplyr::pull(name) |> 
  sort()
#> character(0)

Now let’s create a graphic with the DynaPuff Condensed typeface as the base_family of our theme:

g <- 
  ggplot(diamonds, aes(x = cut)) + 
  geom_bar(fill = "grey65") +
  theme_minimal(
    base_family = "DynaPuff Condensed",
    base_size = 24
  )

g

And now let’s add a non-condensed, bold title:

g +
  ggtitle("PDFs are a font lovers best friend") +
  theme(
    plot.title.position = "plot",
    plot.title = element_text(family = "DynaPuff", face = "bold")
  )

That’s it. If you want to know more about good practices how to handle and customize fonts in {ggplot2} check this blog post by June Choe.

  1. Thanks to the {ragg} package custom font support is working quite well for raster images now—but it’s not a vector graphic 🙃↩︎

  2. Cairo is an open-source graphics library that is known to work very well with custom fonts.↩︎

  3. To make the difference obvious we use the fun, quirky font DynaPuff which is freely available.↩︎