Skip to content

pyplotrs

A blazingly fast, publication-quality plotting library for Python, powered by Rust.

pyplotrs (pronounced py-plotters) pairs a clean, modern Python API with a Rust rendering core. It is built for beautiful static figures you can drop straight into a paper, a slide, or a web page — and, uniquely, PDF output whose text stays real, selectable, and editable in Illustrator (genuine embedded/subset fonts, never outlines).

import pyplotrs as pp

fig, ax = pp.subplots()
ax.line([0, 1, 2, 3], [0, 1, 4, 9], label="y = x²")
ax.scatter([0, 1, 2, 3], [0, 1, 4, 9])
ax.set(title="Hello, pyplotrs", xlabel="x", ylabel="y")
ax.legend()

fig.save("hello.pdf")   # editable vector text
fig.save("hello.png")   # 200 dpi by default

Get started Run the notebooks Browse the gallery

Why pyplotrs?

  • Editable-text PDF

    Text is embedded and subset, not outlined — open a saved PDF in Illustrator/Inkscape and select, re-type, or restyle every label. save(..., tagged=True) writes accessible, tagged PDF.

  • Publication-quality defaults

    A colorblind-safe palette, a sensible type scale, despined axes, and "nice-number" ticks out of the box, at journal-column size. No styling needed to look finished.

  • Fast

    The hot per-point and per-pixel loops live in Rust. A million-point line exports to PDF in ~0.5 s, a million-point scatter to PNG in ~0.15 s, the single-pass layout means the lead grows with panel count, and the GIL is released while rendering.

  • Portable output

    The chosen font is embedded into every saved file (PDF/SVG/PNG/HTML), so a figure looks identical on any machine regardless of installed fonts.

  • Real LaTeX math

    $...$ spans are typeset by a faithful, MathJax-grade engine driven by the math font's OpenType MATH table — and stay selectable text in the output.

  • No global state

    Every figure is an explicit object. No pyplot current-figure surprises; the same API works cleanly under threads.

What can it draw?

Area What you get
Lines & points line, scatter, step, stairs, stem, loglog/semilogx/semilogy
Bars & categories bar, barh, broken_barh, eventplot, plus automatic categorical axes from string data
Distributions hist, boxplot, violinplot (Rust KDE), pie
Uncertainty errorbar, fill_between, fill_betweenx, stackplot
Fields & images imshow + colorbar, matshow, spy, pcolormesh, hist2d, hexbin, contour, contourf, quiver, streamplot
Guides & shapes axhline/axvline, axhspan/axvspan, axline, hlines/vlines, rectangle, circle, ellipse, polygon, arrow
Polar plot, scatter on a configurable dial
3D scatter, plot, surface, bar3d, plot_wireframe, contour3d, plot_trisurf, quiver3d, voxels — projected to editable 2D vectors, with an interactive HTML viewer
Axes Linear, log, symlog, logit, date and categorical scales; 8 tick formatters; 4 color norms
Layout Grids with ratios, subplot_mosaic, GridSpec, twin axes, insets, secondary axes
Themes default, grayscale, dark, plus Theme.with_(...)
Color 127 exact colormaps, 25 categorical palettes, Oklab/CAM16-UCS conversion and CVD checks
Annotations text, annotate with callout arrows; LaTeX math anywhere
Animation animate(render, frames) → GIF / APNG
Formats PDF, SVG, PNG (with DPI), and self-contained HTML

See the gallery for the full set, each with runnable source.

Install

pip install pyplotrs

Pre-built wheels bundle the Rust core and fonts — no toolchain required, and no required runtime dependencies (not even NumPy). See installation for details and building from source.

Where to start