matt-dray / coloratio Goto Github PK

As per tweet: https://twitter.com/mattdray/status/1342160455557668866?s=20

And is it worth the effort?

Thinking about the numbers on the bars here: https://www.gov.uk/government/publications/examples-of-visual-content-to-use-on-govuk/examples-of-visual-content-to-use-on-govuk

cr_*()?

Currently accepts hex and named colours only.

I think cr_choose_color() mostly does the job of cr_choose_bw(), unless you provide ex_bw = TRUE. Excluding black, white and greys will just mean the next most-dark and most-light will be provided though, right? And won't those almost always be navy and snow, etc?

Run cr_choose_color() with ex_bw = TRUE against all named colours to see whether it's just the same few colours each time.

With a GitHub Action.

Probably {magick}.

Probably need to provide a list or vector to get_ratio()

This would make these functions usable in ggplot2::after_scale(), easily solving a common problem not solved elsewhere.

Providing a faux-vectorised version using purrr or *apply might be acceptable as a quick fix but suboptimal performace-wise - I am currently using a purrr-based wrapper for this purpose.

The scales::show_col() function uses {farver} to decide whether to choose black or white:

hcl <- farver::decode_colour("snow", "rgb", "hcl")
hcl
#             h        c        l
# [1,] 12.17432 2.884795 98.64389

ifelse(hcl[, "l"] > 50, "black", "white")
#       l 
# "black" 

It grabs the luminance of the provided colour and selects black or white, depending. Is this better than cr_choose_bw(), which works out which of white and black has the highest contrast with the cupplied colour according to cr_get_ratio()?

As per https://w3c.github.io/silver/guidelines/#visual-contrast-of-text

Create a plot with coloured squares and overlaid text. Show text in colour A on background of colour B and vice versa.