The vellum graphics ecosystem • vellumverse

vellumverse installs and loads the vellum graphics ecosystem in one step, the way tidyverse does for data science. Its packages carry a shared metaphor:

package	role	analogy
vellum	low-level graphics backend, a Rust scene graph, unit/layout engine, and PNG/SVG/PDF renderer	`grid`
vellumplot	pipe-first grammar of graphics that compiles a plot spec into a vellum scene	`ggplot2`
vellumwidget	client-side interactive HTML widgets for the scenes they produce	`plotly`/`htmlwidgets`

vellum is the parchment, vellumplot is the pen, and vellumwidget is the annotation revealed on the page.

The three share one seam, and it is what makes them worth using together. vellum measures text and solves layout without a device, so a compiled scene can report where every element landed: scene_model() gives one row per drawn element with its data key and resolved device-pixel box. vellumplot compiles a plot spec into such a scene, and vellumwidget hosts that same scene in a browser, reading the table instead of re-drawing anything. So one specification yields a static figure, a vector file, and an interactive widget that cannot disagree about geometry, because there is a single solved layout behind all three.

For the longer argument, see a critical appraisal of grid and vellum’s design principles.

Installation

The ecosystem compiles a Rust crate (in vellum), so you need a Rust toolchain (cargo/rustc) alongside R. Then:

# install.packages("pak")
pak::pak("r-vellum/vellumverse")

Usage

library(vellumverse)
#> ── Attaching packages ──────────────────────────────────── vellumverse 0.1.0 ──
#> ✔ vellum 0.1.0   ✔ vellumplot 0.1.0   ✔ vellumwidget 0.1.0

That single call attaches all three core packages, so you can go from a data frame to an interactive plot in one pipeline:

library(vellumverse)

df <- data.frame(wt = mtcars$wt, mpg = mtcars$mpg, model = rownames(mtcars))

vplot(df) |>                                     # vellumplot: grammar of graphics
  mark_point(x = wt, y = mpg, color = mpg,
             tooltip = model, data_id = model) |>
  scale_color_continuous() |>
  as_widget()                                    # vellumwidget: interactive widget

vplot(), mark_point(), and the scales come from vellumplot; as_widget() comes from vellumwidget; both compile down to a vellum scene.

Helpers

vellumverse_packages()    # the packages vellumverse bundles
vellumverse_conflicts()   # functions masked across the ecosystem

To load the packages without the startup banner, set options(vellumverse.quiet = TRUE) before attaching.