Tired of using Helvetica in your R graphics? Here’s how to use the fonts you like

Although R has vast graphical functionality I've lamented the lack of support for additional fonts. You can spend an incredible amount of time fine-tuning a ggplot2 graphic, fiddling with the length of the tick marks, getting the legend just right but then the Helvetica text detracts from the beauty of what you've created.

My understanding is that support for 'non-standard' fonts is extremely difficult given the large number of computing setups, graphics devices etc. Nevertheless we often need to prepare Postscript and PDF plots for scientific papers and reports and the standard fonts often don’t cut it. Luckily there is support for additional fonts particularly if you’re creating PDFs or postscript files. I'm going to show an example using the package extrafont. There is also a relatively new package called showtext that I got to work for me, but I didn't find that it offered any functionality beyond extrafont (and it crashed my R session twice) so I won't cover that package.

1. Import fonts (and some quick data setup)

For the extrafont package you'll need to make sure that you have GhostScript on your system in order to embed the fonts (you will also need to tell R where it's located – see below). In addition, you will need to import the fonts you need. Luckily the package comes with a function that does this for you without much fuss – it takes a couple of minutes, depending on how many fonts you have. Here is an example of the code to import and then review the fonts:

library(extrafont)
font_import() # import all your fonts
fonts() #get a list of fonts
fonttable()
fonttable()[40:45,] # very useful table listing the family name, font name etc
##   package         afmfile                         fontfile       FullName
## 1      NA  AGENCYB.afm.gz  C:\\Windows\\Fonts\\AGENCYB.TTF Agency FB Bold
## 2      NA  AGENCYR.afm.gz  C:\\Windows\\Fonts\\AGENCYR.TTF      Agency FB
## 3      NA  ahronbd.afm.gz  C:\\Windows\\Fonts\\ahronbd.ttf   Aharoni Bold
## 4      NA    ALGER.afm.gz    C:\\Windows\\Fonts\\ALGER.TTF       Algerian
## 5      NA Aller_Rg.afm.gz C:\\Windows\\Fonts\\Aller_Rg.ttf          Aller
## 6      NA    Aller.afm.gz    C:\\Windows\\Fonts\\Aller.ttf     Aller Bold
##   FamilyName      FontName  Bold Italic Symbol afmsymfile
## 1  Agency FB AgencyFB-Bold  TRUE  FALSE  FALSE         NA
## 2  Agency FB  AgencyFB-Reg FALSE  FALSE  FALSE         NA
## 3    Aharoni  Aharoni-Bold  TRUE  FALSE  FALSE         NA
## 4   Algerian      Algerian FALSE  FALSE  FALSE         NA
## 5      Aller         Aller FALSE  FALSE  FALSE         NA
## 6      Aller    Aller-Bold  TRUE  FALSE  FALSE         NA

Now we're ready to use the fonts in an actual plot. Based on my review of Winston Chang's GitHub repository for the package all computer systems can use extrafont to embed fonts in PDF/PS files but extra fonts are only available in bitmap output on Windows machines.

# a little setup, I'm using data from the amazing NMMAPS air pollution study
# background here http://bit.ly/1pqboBG
library(ggplot2)
nmmaps<-read.csv("chicago-nmmaps.csv", as.is=T)
nmmaps$date<-as.Date(nmmaps$date)
nmmaps<-nmmaps[nmmaps$date>as.Date("1996-01-01"),]

2. Use your new fonts – Bauhaus 93 TrueType font anyone?

Below I'm saving directly to an image and the non-default font shows up both on a screen device and in the saved PNG file. But on a Mac you may be limited to saving to PDF (see next setp)

.

# default font
ggplot(nmmaps, aes(x = date, y = temp)) + geom_point(color="red")+
  ggtitle("This is a default font")+
  theme(plot.title = element_text(size=30, face="bold", vjust=1))

plot of chunk unnamed-chunk-5

# new font (note the family argument)
ggplot(nmmaps, aes(x = date, y = temp)) + geom_point(color="red")+
  ggtitle("This is a non-default font")+
  theme(plot.title = element_text(size=30, 
                                  vjust=1, family="Bauhaus 93"))

plot of chunk unnamed-chunk-5

3. Create a PDF (and don't forget to embed your fonts)

You can use ggplot2's great ggsave function to save the plot to PDF.

# note that this step will get warnings that you can ignore
ggsave("newfont.pdf")

If you open the PDF now, though, you will be sorely disappointed! The new font will not appear in the PDF because we have not embedded the font in the PDF yet. In order to do this you have two last steps. Tell R where GhostScript is located and then embed the fonts.

Sys.setenv(R_GSCMD = "C:/Program Files (x86)/gs/gs9.02/bin/gswin32.exe")
embed_fonts("newfont.pdf")

And you're ready to send that PDF directly to your journal of choice.

4. Last comment

I found that most of the fonts worked perfectly both on my image devices as well as PDF but in a few cases something went wrong with the PDF. For example, I played around with the Algerian TrueType font and got output like this:

NO

when it should look like this:

YES

Some digging would be required to track down the issue and, fortunately, Algerian TrueType is not high on my list of fonts to use.

Leave a Reply

Your email address will not be published. Required fields are marked *