Codecov test coverage CRAN status Lifecycle: stable R-CMD-check

This package provides graphic devices for R based on the AGG library developed by the late Maxim Shemanarev. AGG provides both higher performance and higher quality than the standard raster devices provided by grDevices. For a comparison with the default devices, see the performance and quality vignettes.


The package can be installed from CRAN with install.packages('ragg') or, if the development version is desired, directly from github:

# install.packages('devtools')


ragg provides drop-in replacements for the png, jpeg, and tiff graphic devices provided by default from the grDevices packages and can both produce png, jpeg and tiff files. Notable features, that sets itself apart from the build-in devices, includes:

You can use it like any other device. The main functions are agg_png(), agg_jpeg() and agg_tiff(), all of which have arguments that closely match those of the png(), jpeg() and tiff() functions, so switching over should be easy.


file <- knitr::fig_path('.png')

on_linux <- tolower(Sys.info()[['sysname']]) == 'linux'
fancy_font <- if (on_linux) 'URW Chancery L' else 'Papyrus'

agg_png(file, width = 1000, height = 500, res = 144)
ggplot(mtcars) + 
  geom_point(aes(mpg, disp, colour = hp)) + 
  labs(title = 'System fonts — Oh My! 😱') + 
  theme(text = element_text(family = fancy_font))


Further, it provides an agg_capture() device that lets you access the device buffer directly from your R session.

cap <- agg_capture(width = 1000, height = 500, res = 144)
plot(1:10, 1:10)

scatter <- cap()

# Remove margins from raster plotting
par(mai = c(0, 0, 0, 0))

Use ragg with knitr

knitr supports png output from ragg by setting dev = "ragg_png" in the chunk settings or globally with knitr::opts_chunk$set(dev = "ragg_png").

Use ragg in RStudio

ragg can be used as the graphic back-end to the RStudio device (for RStudio >= 1.4) by choosing AGG as the backend in the graphics pane in general options (see screenshot)

Setting ragg as backend in RStudio

Code of Conduct

Please note that the ‘ragg’ project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.