Plotting with Plots.jl and Makie

Choose a backend, build a figure from series and attributes, lay out subplots, save reliably, and know when Makie is worth the extra weight.

Backends

using Plots
gr()                       # chose the backend for this session
default(fmt = :png, size = (800, 500), dpi = 150)

x = range(0, 2pi, length = 200)

p = plot(x, sin.(x);
         label = "sin", lw = 2, color = :steelblue,
         xlabel = "angle (rad)", ylabel = "value",
         title = "Trigonometric functions")

plot!(p, x, cos.(x); label = "cos", ls = :dash, lw = 2)
hline!(p, [0.0]; label = "", color = :grey, alpha = 0.4)

savefig(p, "trig.png")
savefig(p, "trig.svg")

• plot creates a figure, plot! adds to the current one. The exclamation mark convention means the call mutates rather than returns something new.
• Always pass an explicit label or label = ""; otherwise the legend fills up with inferred names.
• The backend is chosen once per session and cannot be swapped after a plot exists.

Layout and series

p1 = scatter(rand(50), rand(50); title = "scatter", markerstrokewidth = 0)
p2 = histogram(randn(1000); bins = 30, title = "histogram", legend = false)
p3 = bar(["a", "b", "c"], [3, 7, 2]; title = "bar")

grid = plot(p1, p2, p3; layout = (2, 2), size = (900, 700))
savefig(grid, "grid.png")

# a 3D surface
xs = ys = range(-2, 2, length = 50)
p4 = surface(xs, ys, (x, y) -> exp(-(x^2 + y^2)))
savefig(p4, "surface.png")

Layouts accept tuples such as (2, 2) or a custom grid with mixed spans through @layout. Build each subplot as a value first, then combine them, rather than chaining everything into one call.

When to use Makie

using CairoMakie        # static, no GPU needed
using GLMakie          # interactive window or browser

fig = Figure(size = (800, 500))
ax = Axis(fig[1, 1], xlabel = "angle (rad)", ylabel = "value", title = "sin")
lines!(ax, x, sin.(x); color = :steelblue, linewidth = 2)
scatter!(ax, x[1:20:end], sin.(x[1:20:end]); color = :tomato)

fig[2, 1] = Legend(fig, ax, "series")
save("makie.png", fig)

FAQ

Related

DataFrames, CSV and tabular data in Julia Debugging, testing and benchmarking Julia code

Last refreshed 2026-09-18.