Fix img axes

[t-bltg]

Fix #4087.

using Plots

png(plot(Gray.(randn(10,15))), "foo")

[t-bltg]

@sefffal, I assumed that the labels will suit you, the alternatives are (horizontal axis as example).

• label vertices 0:15 (PR, distance from border)
• labels vertices 1:16
• labels cells 1:15

Labelling n cells centers is strictly equivalent to labelling n+1 vertices, so this was the easiest fix. But if that doesn't suit you, we should find a way to label cell centers as with plotly (to me this is only a matter of personal taste), if that's even possible (not sure it's doable easily).

[sefffal]

Thanks @t-bltg for looking at this!
My preference would definitely be to label the cell centres, 1:15. At least for most applications in astronomy we like to think of the indices as pointing to the centres of the pixels.
But if there is another convention that is more widely used, I can work around whatever you choose. Thank you!

[t-bltg]

Thanks for your reply.

What we could do right now is fix the current which is obviously wrong with 0:15, but keep #4087 opened until someone finds a more decent solution or a hack for centering ticks in GR.

[t-bltg]

@BeastyBlacksmith, whats is your opinion on this ? There is no point opening a PR in https://github.com/JuliaPlots/PlotReferenceImages.jl if this is not the fix which is going to get in.

[sefffal]

A second reason to prefer labeling cells starting at 1 is that it matches nicely with Julia indexing. For example, you can access the cell at position 1,1 just by accessing arr[1,1]. Since Julia uses 1 based indexing I think this is more natural.

As an aside, how does this work for heatmaps? Can the tick positioning code from there be used for images as well?

[t-bltg]

As an aside, how does this work for heatmaps? Can the tick positioning code from there be used for images as well?

Looks like heatmap is centered at least for GR: https://docs.juliaplots.org/stable/generated/gr/#gr-ref43

[BeastyBlacksmith]

I think we should make this consistent with heatmap

[t-bltg]

using Plots
img = randn(10, 15)
png(plot(Gray.(img), xticks=1:15, yticks=1:10), "foo")

png(plot(Gray.(img)), "bar")

@sefffal, @BeastyBlacksmith, ok now ?

[sefffal]

Beautiful!

[codecov]

Codecov Report

Merging #4088 (a7ea05b) into master (88cd04b) will not change coverage.
The diff coverage is 50.00%.

@@           Coverage Diff           @@
##           master    #4088   +/-   ##
=======================================
  Coverage   62.20%   62.20%           
=======================================
  Files          29       29           
  Lines        7315     7315           
=======================================
  Hits         4550     4550           
  Misses       2765     2765           

Impacted Files	Coverage Δ
src/axes.jl	84.37% <ø> (ø)
src/examples.jl	63.63% <ø> (ø)
src/recipes.jl	58.49% <50.00%> (ø)
src/backends/gr.jl	88.27% <0.00%> (-0.09%)	⬇️
src/types.jl	52.17% <0.00%> (+4.34%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 88cd04b...a7ea05b. Read the comment docs.