In the hclwizard() shiny app the Python export now
leverages the Python colorspace package available from PyPI at
https://pypi.org/project/colorspace/. For more details
see the Python package documentation at https://retostauffer.github.io/python-colorspace/.
Improvement in desaturate(): In addition to previous
input color specifications (hex strings, named colors, RGB matrices),
formal S4 color objects are now also supported. In this case, the
desaturated colors are also returned as color objects of the same class
as the input. With this improvement, desaturate() behaves
exactly analogous to the color vision deficiency emulation functions
which had been improved in the same way in version 2.1-0 (see
below).
Try harder to preserve the names in named color vectors when
transformed with lighten(), darken(),
adjust_transparency(), simulate_cvd(), etc.
(suggested by Matias Andina).
Make the fixed color vision deficiency simulation in linear RGB
coordinates (introduced in version 2.1-0) the default in the
cvd_emulator() shiny app as well.
Minor bug fixes in hclplot(), the automatically
generated figure titles only use the H/L coordinates of the colors
visualized. Minor fix on how right (index of colors on the
right hand side of a diverging palette) is calculated.
Support n = 1 color in diverging_hcl()
and divergingx_hcl() which then both return the neutral
color in the center of the respective palette.
Bug fix for color vision deficiency simulations in
simulate_cvd() based on the work of Machado et al.
(2009): Following some illustrations from the supplementary
materials, the transformations in previous versions of the package had
been applied to gamma-corrected sRGB coordinates directly. However, the
Machado et al. paper implicitly relies on a linear RGB space
(see page 1294, column 1) where their linear matrix transformations for
simulating color vision deficiencies are applied. Therefore, a new
argument linear = TRUE has been added to
simulate_cvd() (and hence in deutan(),
protan(), and tritan()) that first maps the
provided colors to linearized RGB coordinates, applies the color vision
deficiency transformation, and then maps back to gamma-corrected sRGB
coordinates. Optionally, linear = FALSE can be used to
restore the behavior from previous versions where the transformations
are applied directly to the sRGB coordinates. For most colors the
difference between the two strategies is negligible but for some
highly-saturated colors it becomes more noticable, e.g., for red,
purple, or orange. Thanks to Matthew Petroff for reporting this issue
and to Kenneth Knoblauch for advice and guidance.
Improvement in simulate_cvd() (and hence in
deutan(), protan(), and
tritan()): When colors are specified as hex strings or
named colors, the sRGB coordinates after transformation are rounded
appropriately to integers in 0-255. Previous versions implicitly took
the floor rather than round of the coordinates.
Support formal S4 color objects in simulate_cvd()
(and hence in deutan(), protan(), and
tritan()). In that case colors are transformed internally
to sRGB coordinates, color vision deficiency is simulated, and then a
formal S4 color object (of the same class as the input) is returned.
This has the advantage that no rounding is applied as when going through
hex color strings. Similarly, for an RGB matrix input, the output is
also not rounded anymore.
Bug fix in scale_colour_discrete_qualitative() (and
analogously for sequential, diverging, and divergingx) to make sure that
the scale functions can also be called within a function whose arguments
are controlled by an outer function (reported by Marcelo S.
Perlin).
Added the accessible perceptual contrast algorithm (APCA) that is
under development for the next major revision of the WCAG as an
alternative algorithm in contrast_ratio(). See
https://github.com/Myndex/SAPC-APCA.
Fixed a bug in lighten() and darken()
for the case when some colors were named (e.g., "blue")
rather than hex strings only. In that case linear RGB rather than sRGB
was incorrectly used internally (reported by Colin Goodall).
contrast_ratio() for computing and
visualizing contrast ratios of pairs of colors as defined in the Web
Content Accessibility Guidelines (WCAG) (https://www.w3.org/TR/WCAG21/) of the World Wide Web
Consortium (W3C).Added two new palettes from the viridis family
(viridis/viridisLite), namely "Rocket"
and "Mako" in sequential_hcl().
Assure that all do.call() for the
ggplot2 scale functions are made with
envir = parent.frame() in order to avoid problems with
non-standard evaluation, as reported in: https://stackoverflow.com/questions/66049336/paste-name-in-scale-fill-color-in-colorspace-does-not-work-in-a-loop
New function extract_transparency() for extracting
alpha transparency from color specifications, either as numeric,
integer, character or hexmode vectors. This accompanies the previously
available function adjust_transparency() allowing for more
flexible modifications of alpha transparencies in colors.
Release of version 2.0-0 accompanying the publication of the paper “colorspace: A Toolbox for Manipulating and Assessing Colors and Palettes” in the Journal of Statistcal Software at https://doi.org/10.18637/jss.v096.i01. The paper is a compact version of the package web page (https://colorspace.R-Forge.R-project.org/) including some more discussion of related methods and software along with some more references to the scientific literature.
New function adjust_transparency() that adjusts the
alpha transparency of some color specification and returns a hexadecimal
string where the alpha component is modified, added, or
removed.
Added three more palettes from Fabio Crameri’s scientific color
maps (scico), namely "Hawaii" and
"Batlow" in sequential_hcl() and
"Roma" in divergingx_hcl().
Additional option cvd in swatchplot()
so that color vision deficiency emulation can be included on the
fly.
Added binned ggplot2 color scales (in addition to discrete and continuous).
Changed the default colors in specplot() from
rainbow_hcl(3) to qualitative_hcl(3) which is
darker and has more chroma. Also, the order of the legend has been
reversed (Luminance / Chroma / Hue) so that the legend labels are closer
to the axis that they pertain to.
Improve the documentation of the LUV() and
LAB() color spaces, pointing out that the
U/V and A/B
coordinates are typically in [-100, 100] but can actually be
exceeded by highly saturated colors.
Bug fix in diverging_hcl() and
divergingx_hcl() where the central color of the diverging
palette could sometimes collapse to transparent white due to numerical
inaccuracies on 32-bit platforms.
Improve the documentation of the HSV() and
HLS() color spaces. These do NOT necessarily correspond to
sRGB; we can convert from these to any RGB space that we choose (linear
or non-linear, any whitepoint). In particular, if we want to convert
from one of these color spaces to XYZ or beyond, we must first convert
to a specific RGB color space.
New article/vignette “Somewhere over the Rainbow” with published examples of RGB rainbow palettes (or similar highly saturated and non-monotonic palettes).
Bug fix in divergingx_hcl(n) with even
n where the two central colors were erroneously duplicated.
Also, partial matching of palette names has been fixed.
New sequential multi-hue palette: Purple-Yellow. This is a slightly improved version (i.e., with higher luminance contrast) of the palette used in Figure 4 of Stauffer et al. (2015, BAMS).
New flexible diverging palette Zissou 1 in
divergingx_hcl(). This closely matches the palette of the
same name in wesanderson. Note that this is rather unbalanced,
has relatively low luminance contrasts and uses very high chroma
throughout.
New palette Cividis in divergingx_hcl()
approximating the palette of the same name from the viridis family.
While luminance increases monotonically from dark to light in the
palette (thus indicating a sequential and not a
diverging palette), the hue and chroma trajectories resemble a
diverging pattern. Therefore, the flexibility of
divergingx_hcl() is needed and the palette could not be
approximated by sequential_hcl().
Limits of hue axis are improved in specplot().
Previously, the hues were always matched to [0, 100] on the
chroma/luminance axis. Now they are matched to [0, maximum
chroma].
Major update of the package that enhances many of its capabilities, e.g., more refined palettes, named palettes, ggplot2 color scales, visualizations for assessing palettes, more and enhanced shiny and Tcl/Tk apps, color vision deficiency emulation, and much more. See below for further details. A new web site presenting and documenting the package has been launched at https://colorspace.R-Forge.R-project.org/
Claus O. Wilke and Claire D. McWhite joined the colorspace team, adding and enhancing various features, including (but not limited to) especially the color vision deficiency emulation, the ggplot2 palettes, and new shiny apps.
New function simulate_cvd() for simulating color
vision deficiencies with convenience interfaces deutan(),
protan(), and tritan().
New function hcl_palettes() to query pre-defined
HCL-based palettes: qualitative, sequential (single-hue), sequential
(multi-hue), diverging. The corresponding print(),
plot(), and summary() methods can help to
explore the palettes.
Pre-defined HCL palettes are taken from previous publications about colorspace as well as approximations from other packages (ColorBrewer.org, CARTO, viridis, scico).
Users can also register their own custom color palettes for
subsequent usage (within the same session) in
qualitative_hcl(), sequential_hcl(), and
diverging_hcl() using the register = "..."
argument. To generally make such custom palettes available, a
registration R code a la
colorspace::qualitative_hcl(..., register = "myname") can
be placed in the .Rprofile or similar startup scripts. Also
the choose_color()/hclwizard() app allows to
register palettes in the current session.
New and more flexible qualitative_hcl() palette
function. This is similar to the old rainbow_hcl() but
allows to use the pre-defined palettes and change the parameters more
easily.
Palette function sequential_hcl() is now
substantially more flexible: encompasses both single-hue and multi-hue
palettes; gained a new parameter cmax for non-monotonic
chroma paths. Parameters h1, h2,
c1, c2, l1, l2,
p1, p2, cmax allow to easily
modify existing palettes in just a few HCL parameters.
Function diverging_hcl() is introduced as a copy of
diverge_hcl() for a more consistent naming of the *_hcl
palettes where * is one of the adjectives “qualitative”, “sequential”,
and “diverging”. Both diverging_hcl() and
diverge_hcl() now also gained a cmax argument
just like sequential_hcl(). Individual parameters
h1, h2, c1, l1,
l2, p1, p2 can also be easily
modified.
New functions
divergingx_hcl()/divergex_hcl() have been
added for fully fle_x_ible diverging palettes (as opposed to the more
restricted balanced palettes in
diverging_hcl()/diverge_hcl()). These support
parameters h1, h2, h3,
c1, c2, c3, l1,
l2, l3, p1, p2,
p3, p4, cmax1,
cmax2.
Many new predefined palettes that facilitate close approximation of almost all palettes from ColorBrewer.org/RColorBrewer, CARTO/rcartocolor, and viridis. Additionally, approximations to a few of Fabio Crameri’s scientific color maps (scico) are available as well.
New interactive shiny app hcl_color_picker() - or
equivalently, choose_color() - for exploring HCL colors,
and manually assembling individual colors or palettes. Douglas C. Wu
(@wckdouglas)
provided the original implementation for the color palette
feature.
New functions lighten() and darken()
for programatically lightening and darkening colors.
New convenience function swatchplot() that
facilitates displaying color swatches to display and compare collections
of color palettes.
specplot() gained an argument y=NULL to
optionally display a second palette and compare their trajectories. By
default, specplot() now only shows the HCL spectrum but not
the RGB spectrum (rgb = FALSE) because it is mainly used
for illustrating and comparing properties of HCL-based
palettes.
New function hclplot() for visualizing trajectories
of color palettes in two-dimensional HCL space projections.
New function demoplot() that makes the demonstration
plots (map, heatmap, pie, lines, etc.) from the
choose_palette()/hclwizard() app available
outside the GUI on the command line.
Added a new function max_chroma() that
(approximately) computes the maximum chroma possible for a given hue and
luminance combination in HCL space.
Registration of C routines.
In LAB_to_XYZ conversion, replace decimal
approximations with exact rational numbers (reported by Glenn Davis).
Follows Bruce Lindbloom: http://brucelindbloom.com/index.html?LContinuity.html
New function whitepoint() that can both query the
current whitepoint and set it to a different value. By default CIE D65
with XYZ coordinates 95.047, 100.000, 108.883 is used. But it is
possible to set another global whitepoint now, used for all conversions
in the package (suggested by Glenn Davis).
Fixed a bug in desaturate() for named colors (such
as "gray92") where erroneously the RGB()
rather than sRGB() model was used internally.
Added argument desaturate(..., amount = 1) for
optional partial desaturation.
as_HLS(), which was passing
ans rather than color as the colour to convert
(and that was producing not only wrong results, but random results
because the values in ans were not initialized). Thanks to
Thomas Julou for the report.return rather than
return() in
choose_palette()/hclwizard().In addition to the Tcl/Tk-based GUI for
choose_palette() there is now a shiny-based GUI.
choose_palette() by default still uses the Tcl/Tk version
while hclwizard() is a new wrapper that by default calls
the new shiny version.
New function specplot() that converts a given
palette in hex codes to RGB and HCL coordinates and visualizes their
spectrum as a line plot.
hex2RGB() now omits the alpha channel (if any) in
the hex colors provided.
Extended choose_palette() for sequential palettes
with multiple hues: Now two palettes are included in the examples that
are very close to “viridis” and “magma” from matplotlib in Python (also
available in R via package viridis)
Changed Depends/Imports/Suggests to conform with current R CMD check.
choose_palette() when using palette
functions with optional alpha channels.Alpha channel is preserved in desaturate for named colors
(especially "transparent" and NA). (Reported
by Simon Potter.)
Added alpha argument for all palette functions (see
?rainbow_hcl).
Small fixups for R CMD check.
hex() and
hex2RGB() now. (Reported by Richard Cotton.)choose_palette().New Tcl/Tk-based GUI for choosing different types of palettes:
qualitative (rainbow_hcl), single-hue sequential
(sequential_hcl), multi-hue sequential
(heat_hcl), and diverging (diverge_hcl). The
GUI provides a wide collection of pre-stored palettes, easy manipulation
of the corresponding arguments, illustration through a broad range of
plot types (maps, heatmaps, variations of bar plots, scatter plots, and
many more), emulation of desaturation and dichromatic vision,
loading/saving palettes, etc.
Bug fix in polarLAB_to_LAB conversion.
All .Call() calls now with
PACKAGE = "colorspace".
Added some simple tests based on the examples and vignette.
Added desaturate() function for removal of chroma in
a given vector of colors.
Bug fix in HLS_to_RGB conversion for
s == 0.
Added sRGB colorspace. (Existing RGB
colorspace is linearized “sRGB”.)
Conversions to and from HSV and HSL can
only occur from or to RGB or sRGB (because
both HSV and HSL are relative colorspaces,
meaning relative to a particular RGB colorspace). (Converting to or from
RGB gives a different result compared to conversion to or
from sRGB.)
All gamma parameters in all R-level functions have
been deprecated. (The sRGB colorspace has implicit
gamma.)
citation("colorspace")Moved color palettes from vcd to colorspace, including
vignette("hcl-colors")
Added infrastructure for HLS color space
New CITATION file