orbis

A layered grammar of graphics for R that is interactive by default, sharp at any resolution, and knows the world map.

orbis builds a plot from a grammar you already recognise, but instead of committing to one output it compiles the plot to a resolution-independent scene description and hands that to two renderers:

a self-contained SVG writer with embedded JavaScript — tooltips, hover highlighting, mouse-wheel zoom, drag to pan, and clickable legend keys, with no JavaScript library and no internet connection required;
R’s own graphics devices — publication-quality PNG, TIFF or PDF at any resolution you ask for.

Geography is not an add-on. A simplified world polygon dataset ships with the package, and six map projections — including Equal Earth and an orthographic globe — are built in.

Installation

# install.packages("remotes")
remotes::install_github("mqfarooqi1/orbis")

orbis imports only base R packages plus htmltools.

A first plot

library(orbis)

orb(mtcars, x = wt, y = mpg, colour = hp, size = disp) +
  orb_points() +
  orb_labs(title = "Fuel economy", x = "Weight (1000 lbs)",
           y = "Miles per gallon", colour = "Horsepower")

Scatter plot of fuel economy against weight, points coloured by horsepower and sized by displacement

The world, built in

orb_worldmap(values = gdp, value_col = "v", projection = "robinson",
             palette = "mako") +
  orb_labs(title = "World choropleth")

World choropleth in the Robinson projection, countries shaded by value

Switch one argument for a globe:

orb_worldmap(values = gdp, value_col = "v", projection = "orthographic",
             palette = "ember", ocean = "#0B1F33") +
  orb_coord_map("orthographic", centre = c(25, 20)) +
  orb_theme_dark()

Projections available: equirectangular, mercator, robinson, mollweide, equalearth (the modern equal-area projection of Savric, Patterson & Jenny) and orthographic. Points can be placed on any of them with orb_geo_points().

Small multiples

orb(mtcars, x = wt, y = mpg, colour = hp) +
  orb_points() +
  orb_facet(cyl) +
  orb_labs(title = "Small multiples by cylinder count")

Three small-multiple panels of fuel economy, one per cylinder count, sharing axes and a colour legend

Panels share their scales by default, so they are directly comparable; pass scales = "free" (or "free_x" / "free_y") to let each panel stretch to its own data. Colours and sizes are always resolved across the whole data set, so a value means the same thing in every panel.

Interactive, then print-ready — from one object

The same plot object produces both. Nothing is re-specified:

p <- orb(mtcars, x = wt, y = mpg, colour = factor(cyl)) + orb_points()

orb_interactive(p)                  # opens in the viewer: hover, zoom, pan
orb_save(p, "figure.png", dpi = 600) # 600 dpi raster for print
orb_save(p, "figure.pdf")            # vector
orb_save(p, "figure.svg")            # vector, and still interactive
orb_save(p, "figure.html")           # a self-contained interactive page

In the interactive output you can hover for a tooltip, scroll to zoom, drag to pan, double-click to reset, and click a legend key to hide or show that series.

What is in it

Function	Purpose
`orb()`	Start a plot: data plus a mapping from variables to channels
`orb_points()`, `orb_line()`, `orb_bars()`, `orb_area()`, `orb_text()`	Layers
`orb_map()`, `orb_geo_points()`, `orb_worldmap()`	Geographic layers
`orb_coord_map()`	Six map projections, including Equal Earth and an orthographic globe
`orb_facet()`	Small multiples: one panel per level, shared or free scales
`orb_scale_colour()`, `orb_scale_fill()`, `orb_scale_size()`, `orb_scale_x()`, `orb_scale_y()`	Scales, limits, log and square-root axes
`orb_theme_light()`, `orb_theme_dark()`, `orb_theme_minimal()`, `orb_theme_ink()`	Themes
`orb_save()`, `orb_svg()`, `orb_interactive()`, `orb_draw()`	Output
`orb_palettes()`	Perceptually uniform and qualitative palettes
`world_map`	Simplified world polygons, 229 regions

Honest limitations

orbis is young, and it is not a drop-in replacement for a mature plotting package. In particular:

No statistical layers. There is no smoother, no boxplot and no histogram geometry; summarise your data first.
Text metrics are approximated when laying out axes and legends, because the SVG writer does not query a font engine. Very long labels can be spaced imperfectly.
The world map is deliberately simplified for figure-sized output. For detailed cartography, or for anything requiring a coordinate reference system, use a dedicated spatial package.

References

Wickham, H. (2010) “A Layered Grammar of Graphics.” Journal of Computational and Graphical Statistics 19, 3-28. doi:10.1198/jcgs.2009.07098
Šavrič, B., Patterson, T. & Jenny, B. (2019) “The Equal Earth map projection.” International Journal of Geographical Information Science 33, 454-465. doi:10.1080/13658816.2018.1504949
Crameri, F., Shephard, G. E. & Heron, P. J. (2020) “The misuse of colour in science communication.” Nature Communications 11, 5444. doi:10.1038/s41467-020-19160-7
Snyder, J. P. (1987) Map Projections: A Working Manual. US Geological Survey Professional Paper 1395. doi:10.3133/pp1395
Douglas, D. H. & Peucker, T. K. (1973) “Algorithms for the reduction of the number of points required to represent a digitized line or its caricature.” Cartographica 10, 112-122. doi:10.3138/FM57-6770-U75U-7727

World boundaries derive from Natural Earth (public domain), via the maps package.

MIT licensed. Muhammad Farooqi.