The meshimage recipe

GeoMakie presents a meshimage recipe to render images in projected spaces, without having to reproject.

You can generally think of these as discretized versions of the true continuous reprojection.

meshimage works best with the GL-backends (GLMakie/WGLMakie) because they have the best inbuilt support for meshes. CairoMakie also supports meshimage, but the quality of the image will be worse unless your mesh is very fine (npoints > 1000 usually).

using GeoMakie, GLMakie

Affine transform (no nonlinearity)

fig = Figure()
ax1 = Axis(fig[1, 1])
mi = meshimage!(ax1, (Makie.logo()))
fig

save("simple_scaling.png", fig)

Nonlinear but separable (log scaling)

fig = Figure()
ax1 = Axis(fig[1, 1]; xscale = log, yscale = log10)
lg = (Makie.logo())
mi = meshimage!(ax1, 1 .. size(lg, 1), 1 .. size(lg, 2), lg)
fig

save("log_scaling.png", fig)

Nonlinear and non-separable (geographic transformation)

fig = Figure()
ax1 = GeoAxis(fig[1, 1]; dest = "+proj=igh")
mi = meshimage!(ax1, -169..169, -90..90, (Makie.logo()); npoints = 400)
fig

This is peak graphic art right here! save("geo_scaling.png", fig)

Logo!

Finally, we create a kludged together GeoMakie logo with the Earth in the background, and the Makie logo at the fore!

fig = Figure()
ax1 = GeoAxis(fig[1, 1]; dest = "+proj=igh")
s1 = surface!(ax1, (-180f0)..(180f0), -90f0..90f0, zeros(size(GeoMakie.earth() |> rotr90)); shading = NoShading, color = GeoMakie.earth() |> rotr90)
fig

m1 = meshimage!(ax1, -100..100, -40..40, (Makie.logo()); npoints = 300)

m1 = meshimage!(ax1, -166..169, -90..90, (Makie.logo()); npoints = 400)
hidedecorations!(ax1)
fig

---

This page was generated using Literate.jl.