Colour Science for Python

The colour.continuous module provides tools for working with continuous signals. It includes classes for representing continuous functions and multi-signal data structures. Key components include:

• AbstractContinuousFunction: An abstract base class for continuous functions.
• Signal: A class representing a continuous signal.
• MultiSignals: A class for handling multiple signals simultaneously.

The Colour API uses common prefixes to bundle related objects, which facilitates easier introspection and auto-completion in environments like IPython or Jupyter.

Key categorisation patterns include:

• Spectral Distribution functions: Prefixed with sd_ (e.g., colour.sd_blackbody()).
• Spectral Distribution attributes/methods: Grouped under the colour.SD namespace (e.g., SD_GAUSSIAN_METHODS, SDS_ILLUMINANTS).
• RGB Colourspaces: Accessed via the colour.models namespace with the prefix RGB_COLOURSPACE_ (e.g., colour.models.RGB_COLOURSPACE_BT709).

The colour library is organized into specialized sub-packages. Most of the public API is accessible directly from the root colour namespace, but specific domains are grouped into sub-packages such as:

• adaptation: Chromatic adaptation models.
• appearance: Colour appearance models.
• blindness: Colour vision deficiency models.
• characterisation: Colour correction and camera/display characterisation.
• colorimetry: Core objects for colour computations (e.g., spectral distributions).
• difference: Colour difference computations.
• io: Input/output for reading and writing data.
• models: Colour models (e.g., Lab, Luv, LCH).
• notation: Colour notation systems.
• plotting: Diagrams and figures.
• temperature: Colour temperature computations.

You can access the public API of the root namespace or specific sub-packages using standard Python imports.

In Colour, spectral distributions are represented by the colour.SpectralDistribution class (or its subclasses). These are instances of the colour.continuous.Signal class, which implements an interpolating function. This allows you to evaluate the spectral distribution at any given wavelength within its domain, even if that wavelength was not part of the original discrete data.

Multi-spectral distributions are handled by the colour.MultiSpectralDistributions class, which acts as a container for multiple colour.continuous.Signal instances.

Most plotting functions are available in the colour.plotting namespace. You can use colour_style() to apply the default colour science plotting styles to your environment.

from colour.plotting import *

colour_style()

The .wavelengths and .values properties return copies of the underlying discretized data. Modifying these properties directly (e.g., sd.values[0] = 0) will not update the actual SpectralDistribution object.

To update the values, you must re-assign the modified array back to the property.

import colour

data = {
    500: 0.0651,
    520: 0.0705,
    540: 0.0772,
    560: 0.0870,
    580: 0.1128,
    600: 0.1360,
}
sd = colour.SpectralDistribution(data)

# INCORRECT: This does not change the distribution
sd.values[0] = 0

# CORRECT: Re-assign the modified array
vals = sd.values
vals[0] = 0
sd.values = vals

To compute CIE xy chromaticity coordinates from CIE XYZ values, use colour.XYZ_to_xy(XYZ). This returns a 2-element array [x, y].

# Computing *CIE xy* chromaticity coordinates for the *neutral 5 (.70 D)* patch.
xy = colour.XYZ_to_xy(XYZ)
print(xy)

You can install the core Colour package and its primary dependencies using pip or conda.

Using pip:

pip install --user colour-science

Using conda:

conda install -c conda-forge colour-science

For detailed installation procedures regarding secondary dependencies, refer to the official Installation Guide.