RFC: Implement line_z and fill_z for GR

[daschw]

This allows to have lines with colors corresponding to a value, multi-colored lines, "heatmaps" with custom shapes, ... e.g.

using Plots; gr()

σs = 0.5:0.1:2
normal_x = linspace(-5, 5, 100)
normal_y = [exp.(-normal_x.^2 / (2σ^2)) / (2π * σ^2) for σ in σs]

xs, ys = Plots.unzip(Plots.partialcircle(0, 2π, 50, 0.5))
shapes = [Shape(xs + i, ys + j) for i in 1:10, j in 1:10]
zvals = [i * j for i in 1:10, j in 1:10]

y = cumsum(randn(1000))

plot(
    plot(normal_x, normal_y, line_z = σs'),
    plot(shapes, fill_z = zvals),
    plot(y, line_z = y),
    plot(y, fillrange = 0, fill_z = y, fillcolor = :grays),
    label = ""
    )

used to look like this:

Now it produces:

If this behaviour is desirable implementations for PyPlot and Plotly will follow.

[mkborregaard]

Sweet! 💯

[mkborregaard]

I think fill_z is implemented for PyPlot and Plotly, though, but to do something different - to custom color individual facets of surface plots (#641)

[daschw]

I think fill_z is implemented for PyPlot and Plotly, though, but to do something different - to custom color individual facets of surface plots (#641)

OK, thanks for the info! Then I should probably implement this for GR as well. I guess the behaviour of the last subplot is probably not required that often, but I can still implement the fill_z behaviour of the top right subplot.

[mkborregaard]

Cool. Might it create some confusion that fill_z means two different things then? (not sure)

[daschw]

Does it? I always understood the marker_z, line_z and fill_z attributes as possibilities to provide an additional value which is translated into colors (of a gradient) and can be interpreted via a colorbar. For markers it's marker_z, for lines line_z and for filled areas fill_z. Filled areas can appear in surfaces, shapes and fillranges. <- That's how I would read it ... Not sure if it's too confusing or not consistent enough (shapes also have lines/strokes, markers have lines/strokes and filled areas ...)

[mkborregaard]

No, I agree. But @tbreloff 's use of the term when used to color the facets in a surfaceplot is not consistent with that logic...

[mkborregaard]

But maybe it's consistent from the user's perspective, as you point out 👍

[mkborregaard]

I really love this enhancement.

[daschw]

Cool, I don't know how to implement Tom's facet color fill_z from PyPlot and Plotly for GR, though. I could not figure out how to color the facets of a surface in GR yet ...

[mkborregaard]

I guess that's why that is not implemented yet...