ggsketch — hand-drawn geoms for ggplot2 • ggsketch

Grammar-native, hand-drawn geoms for ggplot2 - the rough.js sketch aesthetic in pure R, with no JavaScript and no browser.

Sketchy hachure-filled violin plots of highway mpg by vehicle class, with a handwriting-font title

ggsketch gives you ggplot2 geoms (geom_sketch_col(), geom_sketch_line(), geom_sketch_point(), …) that render with a wobbly, hand-drawn look: roughened double-stroke outlines and hachure / cross-hatch / zigzag / dots / dashed fills. Because the geoms are real grid grobs wrapped in ggproto, they compose with aes(), stats, scales, facets, and coords, and draw correctly on every R graphics device — screen, PNG, PDF, and SVG.

Why another sketch package?

	ggsketch	ggrough
Approach	Native ggplot2 geoms (grid grobs)	Post-hoc convert a finished plot to SVG, redraw in HTML Canvas
Output	Any device: screen / PNG / PDF / SVG	HTML widget only (breaks static PDF/PNG)
Composes with `aes()` / stats / scales / facets	Yes	No (operates on the rendered plot)
JavaScript / browser	None	Requires rough.js in a browser
Status	Active	Maintainer marks it dormant (“doesn’t work with recent releases of ggplot2”)

The ggrough maintainer himself notes that “a nice way to create sketchy visualisations would be a neat addition to the {ggplot2} ecosystem.” ggsketch fills that gap with native geoms.

Installation

# install.packages("pak")
pak::pak("orijitghosh/ggsketch")

ggsketch is pure R (NeedsCompilation: no); its only hard dependencies are ggplot2, grid, rlang, scales, cli, and withr.

Full documentation and a gallery of every geom: https://orijitghosh.github.io/ggsketch/.

Quick start

library(ggplot2)
library(ggsketch)

df <- data.frame(product = c("Alpha","Bravo","Charlie","Delta"),
                 units   = c(34, 51, 22, 47))

ggplot(df, aes(product, units)) +
  geom_sketch_col(fill = "#7BAFD4", seed = 1L) +
  labs(title = "Units sold") +
  theme_sketch()

Every randomized routine is seeded, so a given seed always produces the same wobble — your plots are reproducible. Set a session-wide default with options(ggsketch.seed = 1L).

Showcase

A scatter with a hand-drawn frame (theme_sketch(rough_frame = TRUE)), the sketch colour palette, and a roughened linear fit:

Sketchy scatter of fuel economy with hand-drawn gridlines, roughened points, and linear fits coloured by drivetrain

Overlapping densities and the new "scribble" fill, both on the sketch palette:

Overlapping sketch density curves of highway mpg by drivetrain Scribble-filled sketch bar chart

A filled area and line over time, again with a roughened frame:

Sketchy hachure-filled area and line chart of US unemployment over time

The geoms

Family	Geoms
Lines & points	`geom_sketch_line()`, `geom_sketch_path()`, `geom_sketch_point()`
Bars & tiles	`geom_sketch_col()`, `geom_sketch_bar()`, `geom_sketch_rect()`, `geom_sketch_tile()`
Areas & curves	`geom_sketch_polygon()`, `geom_sketch_ribbon()`, `geom_sketch_area()`, `geom_sketch_density()`, `geom_sketch_smooth()`
Circular & composed	`geom_sketch_circle()`, `geom_sketch_ellipse()`, `geom_sketch_segment()`, `geom_sketch_step()`, `geom_sketch_boxplot()`
Helpers	`annotate_sketch()`, `theme_sketch()`
Frame & scales	`element_sketch_line()`, `element_sketch_rect()` (via `theme_sketch(rough_frame = TRUE)`), `scale_colour_sketch()`, `scale_fill_sketch()`, `register_sketch_font()`

Shared sketch parameters

Parameter	Meaning
`roughness`	How far points are jittered (0 = ruler-straight, ~1 default, >3 loose)
`bowing`	How much segments bow outward
`n_passes`	Overlaid strokes (2 = the classic “double stroke”)
`seed`	Integer for reproducible wobble
`fill_style`	`"hachure"`, `"cross_hatch"`, `"zigzag"`, `"zigzag_line"`, `"scribble"`, `"dots"`, `"dashed"`, `"solid"`
`hachure_angle`, `hachure_gap`, `fill_weight`	Fill line angle, spacing, and weight

How it works

Three layers, kept strictly separate:

Layer 1 — pure geometry (R/core-*.R): numbers → numbers. Seeded roughening, ellipse/Bézier sampling, and an Active-Edge-Table scan-line hachure filler that handles concave polygons. No grid, no ggplot2.
Layer 2 — grid grobs (R/grob-*.R): makeContent() converts to device inches and re-roughens at the real render size, so resizing re-draws cleanly.
Layer 3 — ggproto geoms (R/geom-*.R): standard ggplot2 extension API.

Roughening always happens in inch space, so the look is consistent across aspect ratios and devices.

Credits & non-affiliation

The algorithms are reimplemented in original R from the published descriptions of the rough.js algorithms (Preet Shihn, 2020) and the hachure approach of Wood et al. No rough.js source is vendored, copied, or translated, and no JavaScript ships in this package. See inst/NOTICE.

ggsketch is an independent R package reimplementing the hand-drawn sketch aesthetic from first principles. It is not affiliated with, derived from, or endorsed by the rough.js project, ggrough, or any related JavaScript libraries.