Slab fill alpha combines poorly with outlines

[Ax3man]

Hi Matthew,

Thanks for making my favorite ggplot extension package!

slab_alpha applies to both the fill of the slab, as well as the outline. While the alpha can work nicely for slabs with either a fill or an outline, it looks quite poor whenever both are present:

Consider e.g.:

dist_df <- data.frame(mean = 0, sd = 1)

ggplot(dist_df, aes(xdist = dist_normal(mean, sd))) +
  stat_slab(fill = 'firebrick', color = 'black', slab_alpha = 0.6, slab_linewidth = 3)

Half of the outline overlaps with the fill, while the other half does not, which makes the outline look like two outlines.

The desired result can be obtained by manually assigning a color with an alpha value:

ggplot(dist_df, aes(xdist = dist_normal(mean, sd))) +
  stat_slab(fill = scales::alpha('firebrick', 0.6), color = 'black', slab_linewidth = 3)

But perhaps it would be nicer if the alpha of the outline and the fill can be controlled separately, or, if ggdist follows the ggplot convention of applying the alpha only to fills (as in e.g. geom_density()).

[mjskay]

Yeah good point, the current approach is definitely not the best. Though, I'm not convinced having alpha target only the fill is right either, as that can already be done by using aes(fill = after_scale(alpha(...))).

Another option would be to have alpha set an alpha mask on the combined line and fill, so that they are first rendered together and then alpha-blended with the background. That would eliminate the artifacts caused by the outline and fill being alpha blended with each other. Though that could be unintuitive in some cases since it can only apply when alpha is constant at the slab level --- when it isn't, the behavior would have to be either the current behavior or for alpha only to apply to fill.