|
|
Portrait of Paul Centore by Lynn Anderson
|
Colour Theory for Painters
by Paul Centore
© March 16, 2012 (launched Oct. 20. 2010)
Introduction
This webpage is devoted to applying colour theory to everyday problems that
painters face. The goal is to provide, as far as possible, a rigorous understanding
of the use of colour in painting. Currently, the main result is a method by
which a painter can determine the colours of an object in shadow, relative to
the colours of that same object in light. A PDF file with an explanation and
examples,
Shadow Colours For Painters, can
be downloaded; more details are available below. It is hoped that other
results will follow.
Although
colour science has a well-established, relevant body of knowledge, most painters
are unfamiliar with it. An important facet of colour science is the Munsell
colour system, devised by the painter Albert Munsell at the turn of the 20th
century. The Munsell system's concepts of hue, value, and chroma, are basic
to painting, and provide a convenient framework for colour decisions in painting.
Despite its usefulness, the Munsell system is also not widely known among
painters. The conclusions in the articles cited here draw on the data and concepts
of colour science, for a firm foundation, rooted in human colour vision.
It should be emphasized that the results presented here are starting points, from
which painters can and probably should deviate. While it is helpful to know how to
produce a mechanically correct shadow series, it is more common to modify shadows
to "incorrect" colours for expressive ends. Nevertheless, these modifications can
be made intelligently only if the correct version is thoroughly understood. In a
similar way, Michelangelo's distortion of human proportions are effective only
because he was intimately familiar with correct proportions.
Articles
1. The article
Shadow Colours For Painters is intended
for artists. It gives a detailed,
non-technical explanation of shadow colours.
2. The technical basis for the above article appears in a second
article,
Shadow Series in the Munsell
System, which has been accepted for publication in the journal Color Research
and Application.
Abstract:
Using an inversion of the Munsell renotation, this paper calculates that a colour's
shadow series is approximately a straight line in the Munsell system. The line
starts at the colour's Munsell specification and ends about one value step below N0,
on the neutral axis. The colour's hue in shadow shifts slightly towards the yellow
part of the spectrum. The calculations suggest that ideal black belongs at about
N(-1) in the Munsell system, rather than at N0, if equality of perceptual steps is
to be maintained. Similarly, ideal white should be slightly lighter than N10.
Here is the set of 182 shadow series, along with approximating least squares lines,
plotted in terms of their values and chromas, but disregarding hue, in the Munsell
system (Fig. 4 of the article):
Note that the bulk of the extrapolated shadow series cross the neutral axis between
0 and 2 value steps below N0. The data used to generate this figure is contained
in the file
ShadowSeriesInMunsellSystem.txt, which others are free to
use for further research.
3. To perform shadow calculations, it was first necessary to invert the Munsell
renotation. This task is accomplished in the article,
An Open-Source Inversion Algorithm
for the Munsell Renotation, which has been accepted for publication in the
journal Color Research and Application.
Abstract:
The 1943 Munsell renotation includes a table that converts 2,734 Munsell
specifications into xyY coordinates, along with a graphical interpolation method,
and a graphical inversion method, that converts xyY coordinates back into Munsell
specifications. The current paper presents open-source computer code, running in
Matlab or Octave, that both interpolates and inverts the Munsell renotation
automatically. The steps in both algorithms are described in detail. Like
previous inversion algorithms, it relies on interpolations between entries
in the 1943 table. For colours near the MacAdam limits, the inversion also
requires extrapolations beyond the 1943 entries. The outputs of the current
implementation do not differ significantly from the outputs of other inversion
algorithms. The main distinguishing feature of the current algorithm is that
both the algorithm and code implementation are publicly available.
This second article is of more interest to colour scientists than to painters.
An Inverse Table
for the Munsell
Renotation gives a table of inverse renotation values. The main routine for the
code implementation is
xyYtoMunsell.m, which is written in Matlab/Octave. This routine runs in the
ColorLab project; supporting routines and more information are available below.
4. As a byproduct of the previous articles, a helpful description of colour space
was found:
A Zonohedral Approach to
Optimal Colours. This article has been accepted for publication in the journal
Color Research and Application.
Abstract:
This paper demonstrates that the CIE XYZ colour solid is a zonoid. An
approximating zonohedral colour solid is constructed explicitly from a set of
generating vectors, which are integrals of colour-matching functions over
narrow intervals of the visible spectrum. The zonohedral approach yields an intuitive,
constructive proof of the Optimal Colour Theorem: the reflectance function of an
optimal colour takes on only the values 0 or 1, with at most two transition
wavelengths. In addition, zonohedral techniques can simplify computations: for
example, optimal colours can be found without calculating transition
wavelengths. Finally, zonohedra provide a simple, unified approach to colour
space, and eliminate much of the confusion arising from chromaticity diagrams.
This third article is also of more interest to colour scientists than to painters. Here
is the colour solid constructed as a zonohedron (Fig. 9 of the article):
5. The previous article led to a joint paper with Michael H. Brill,
Extensible
Multi-Primary Control Sequences.
Abstract:
In a display with more than three primaries (called a multi-primary display), a
color can be expressed as multiple combinations (called control sequences) of
primaries. This paper presents an algorithm for assigning control sequences, that
preserves current assignments when further primaries are added. We call these control
sequences extensible. It is shown that the gamut of any number of primaries is
a zonohedron, which can be dissected into parallelepipeds. Control sequences
are assigned within each parallelepiped. The current parallelepipeds remain
when more primaries are added, so the current assignments are preserved. Multi-primary
displays can also cause unwanted metamerism, and make continuous color scales
appear discontinuous. The algorithm avoids these problems. When viewed through
natural filters, such as yellowed ocular lenses, multi-primary displays can
sometimes make two different colors appear identical. If the primaries satisfy
the Binet-Cauchy criterion, which is always the case when all primaries are
monochromatic, then these spurious matches are avoided.
This article has been published in the Journal of the Society for Information
Displays. The article is of more interest to display engineers than to painters. Here
is a dissection of the zonohedral gamut, resulting from four primaries, into
parallelepipeds (from Fig. 4 of the article):
|
6. A further paper (accepted for publication in the Journal of the Society
for Information Displays) on metamerism in multi-primary displays:
Non-Metamerism of Boundary
Colours in Multi-Primary Displays.
Abstract:
A control sequence gives the intensities of the primaries for a pixel of a display
device. The display gamut, i.e. the set of all the colours that a display can produce,
is a zonohedral subset of CIE XYZ space, and contains both boundary and interior
colours. Displays with four primaries or more exhibit metamerism, in which different
control sequences produce colours that appear identical to an observer. This paper
shows mathematically that, provided no three primaries are linearly dependent, metamerism
can only occur for interior colours. When there are four or more primaries, metamers
can always be found for interior colours. A colour on the gamut boundary, by contrast, is
only produced by a unique control sequence. The proof used for displays can be extended
to object-colour solids, to show that optimal colours, which are on the boundary of
an object-colour solid, have unique reflectance functions.
7. Further development along the same lines:
Invariants Under Illuminant
Transformations.
Abstract:
An object colour's CIE $XYZ$ coordinates can change when it is viewed under different
illuminants. The set of $XYZ$ coordinates for all object colours, which is called the
object-colour solid, likewise varies under different illuminants. This paper shows that,
despite these changes, some properties are invariant under illuminant transformations.
In particular, as long as the illuminant is nowhere zero in the visible spectrum, optimal
colours take the same Schr\"odinger form, and no two optimal colours are metameric.
Furthermore, all object-colour solids have the same shape at the origin: they all fit
perfectly into the convex cone (which we will call the spectrum cone) generated by the
spectrum locus. The spectrum cone, itself, does not vary when the illuminant changes.
The object-colour solid for one illuminant can be transformed into the solid for another
illuminant, by an easily visualized sequence of expansions and contractions along
irregular rings, called zones.
This article is of more interest to colour scientists than to painters. Here
is an example of a zone around the object-colour solid for a simplified
illuminant (from Fig. 4 of the article):
|
ColorLab Contributions
The ColorLab open source software project, which can be downloaded
here, was
used to perform calculations when needed. The ColorLab project can be run from
the computational software package MATLAB, or from Octave (download
here), which is a
freely available open source clone of MATLAB. In the course of these investigations,
29 new ColorLab routines, as well as about a dozen data files suitable for
ColorLab, were written, and contributed to the ColorLab project. The tables
at the end of this page list these files, and their text can be
accessed by clicking on the highlighted links. One table lists the primary computational
routines, and another table lists secondary routines, which are not of much interest by
themselves, but which are needed by the primary routines. There is also a table of .mat
files, which contain data structures in a binary format, that MATLAB or Octave can
read. All the data files and routines have
been archived and compressed into the file
MunsellConversions.zip, which can be downloaded and unzipped. The code
in these files supersedes, and in some cases is a revision of, similar code that was
posted in October 2010. Each file contains a revision date at the end of the initial
comments, for comparison with earlier versions if desired. Other
investigators are encouraged to check and modify the routines
as needed, and use them for further research.
Conversions between Munsell and CIE (xyY or XYZ) coordinates
In addition to the Octave/Matlab code presented here, other tools are available for
converting between Munsell specifications and CIE coordinates:
1. Here
are some open-source C/C++ conversions routines.
2. Here is a Munsell package for R, the open source
statistics program.
3. Psychtoolbox-3 is a set of open source routines that run in Matlab,
so they should work in Octave as well. This toolbox is developed mainly at the University of
Pennsylvania, where it is used for vision research. Currently, it has code to convert from xyY to
Munsell, but not from Munsell to xyY. I submitted my routines, in MunsellConversions.zip, but have not
heard back.
4. Color2drop and
Drop2Color are two
Matlab programs written by Zsolt Kovacs, of the University of Brescia. They are available in both
English and Italian versions. They output paint mixtures (using artist's paints from several different
brands) that produce an input Munsell colour, and vice versa. The source code is available upon
request. I have not looked at the conversion method in detail, but the documentation mentions
cubic interpolation.
5. Wallkill color produces commercial software that
can be used interactively to make Munsell conversions. For $9.95 (as of Jan. 8, 2012), one can buy
a one-year license for the basic Wallkill program. In the basic
program, Munsell specifications can be entered manually, one
at a time, and converted to xyY coordinates, and vice versa. One glitch is that the program
will terminate, and must be restarted, if the entered specification is outside the MacAdam limits. The
Wallkill program runs only on Windows systems.
6. BabelColor's Patch Tool is also commercial software
that performs Munsell conversions.
7.
Here is a helpful online discussion about the related problem of converting Munsell
specifications to RGB coordinates.
8. The discussion above suggested this
free software for Munsell conversions. It is very helpful for visualization, but only runs on PCs.
9. A 1990 paper (N. S. Smith, T. W. A. Whitfield, & T. J. Wiltshire, "A Colour Notation Conversion
Program," COLOR Research and Application, Vol. 15, Number 6, December
1990, pp. 338-343) mentions conversion code in Pascal, that is free for non-commercial use. This
code could not be located.
10. A 1987 paper (Frederick T. Simon & Judith A. Frost. "A New Method for the Conversion of
CIE Colorimetric Data to Munsell Notations," COLOR Research and Application,
Vol. 12, Number 5, 1987, pp. 256-260) implemented Munsell conversion code in Fortran, as part
of a thesis project. This code could also not be located.
Contact Information
Comments, suggestions, and questions are welcomed. I can be reached by email at
{centore ATSIGN 99main.com}.