Graphical user interface to analyse palettes. c4a_table
shows a table that can be opened in the browser. c4a_gui
is a graphical user interface (shiny app) around this table.
Usage
c4a_gui(type = "cat", n = NA, series = "all")
c4a_table(
type = c("cat", "seq", "div", "cyc", "bivs", "bivc", "bivd", "bivg"),
n = NULL,
m = NULL,
continuous = FALSE,
filters = character(0),
cvd.sim = c("none", "deutan", "protan", "tritan"),
sort = "name",
text.format = "hex",
text.col = "same",
series = "all",
range = NA,
colorsort = "orig",
include.na = FALSE,
show.scores = FALSE,
columns = NA,
verbose = TRUE
)
Arguments
- type
type of palette. Run
c4a_types
to see the implemented types and their description. Forc4a_gui
it only determines which type is shown initially.- n, m
n
is the number of displayed colors. For bivariate palettes"biv"
,n
andm
are the number of columns and rows respectively. If omitted: for"cat"
the full palette is displayed, for"seq"
,"div"
and"cyc"
, 7, 9, and 9 colors respectively, and for"bivs"
/"bivc"
/"bivd"
/"bivg"
4 columns and rows. Forc4a_gui
it only determines which number of colors initially.- series
Series of palettes to show. See
c4a_series
for options. By default,"all"
, which means all series. Forc4a_gui
it only determines which series are shown initially.- continuous
should the palettes as continuous instead of discrete. Only applicable for
"seq"
,"div"
, and"cyc"
.- filters
filters to be applied. A character vector with a subset from:
"nmax"
(only palettes wheren = nmax
, which is only applicable for categorical palettes),"cbf"
(colorblind-friendly),"fair"
(fairness),"naming"
(nameability),"crW"
(sufficient contrast ratio with white), and"crB"
(sufficient contrast ratio with black). By default an empty vector, so no filters are applied.- cvd.sim
color vision deficiency simulation: one of
"none"
,"deutan"
,"protan"
,"tritan"
- sort
column name to sort the data. The available column names depend on the arguments
type
andshow.scores
. They are listed in the warning message. Use a"-"
prefix to reverse the order.- text.format
The format of the text of the colors. One of
"hex"
,"RGB"
or"HCL"
.- text.col
The text color of the colors. By default
"same"
, which means that they are the same as the colors themselves (so invisible, but available for selection)."auto"
means automatic: black for light colors and white for dark colors.- range
vector of two numbers that determine the range that is used for sequential and diverging palettes. Both numbers should be between 0 and 1. The first number determines where the palette begins, and the second number where it ends. For sequential palettes, 0 means the leftmost (normally lightest) color, and 1 the rightmost (often darkest) color. For diverging palettes, 0 means the middle color, and 1 both extremes. If only one number is provided, this number is interpreted as the endpoint (with 0 taken as the start). By default, it is set automatically, based on
n
.- colorsort
Sort the colors (
"cat"
only). Options:"orig"
(original order),"Hx"
(hue, where x is a starting number from 0 to 360),"C"
(chroma),"L"
(luminance)- include.na
should color for missing values be shown?
FALSE
by default- show.scores
should scores of the quality indicators be printed? See details for a description of those indicators.
- columns
number of columns. By default equal to
n
or, if not specified, 12. Cannot be higher than the palette lengths.- verbose
should messages and warnings be printed?
Details
See vignette how the properties are calculated. Parameters, such as threshold values which determined when palettes are classified as "colorblind-friendly", can be specified via c4a_options
. Also the nameability score function (which is in development) can be specified there. See the examples of c4a_options
for both use cases.
See also
References of the palettes: cols4all-package
.
Examples
if (requireNamespace("shiny") &&
requireNamespace("shinyjs") &&
requireNamespace("kableExtra") &&
requireNamespace("colorblindcheck") &&
interactive()) {
c4a_gui()
# categorical palettes with maximum number of colors
c4a_table(type = "cat")
# sort sequential palettes by hue
c4a_table(type = "seq", n = 7, sort = "H")
# sort sequential palettes by hue type (how many hues are used)
c4a_table(type = "seq", n = 5, sort = "hues")
}