Colour Science for Python

0.4.7 · active · verified Wed Apr 15

Colour Science is a comprehensive open-source Python package providing a collection of algorithms and datasets for colour science applications, including colourspace conversions, colour appearance models, and spectral computations. It's currently at version 0.4.7 and maintains an active development cycle with frequent alpha milestones and releases, often introducing new features and compatibility updates.

Warnings

breaking Python version support has evolved. As of v0.4.7, the library requires Python >=3.11 and <3.15. Older Python versions (e.g., 3.5, 3.10) are no longer supported.
Fix: Ensure your Python environment is >=3.11 and <3.15. Upgrade Python if necessary, or use a compatible older version of 'colour-science' if specific Python versions are required.
gotcha Several key dependencies like `scipy` and `imageio` became optional in v0.4.7. If your code relies on functionalities provided by these packages, you must explicitly install them, potentially using `pip install "colour-science[all]"` or `pip install colour-science[scipy,imageio]`.
Fix: For existing projects, verify all required optional dependencies are explicitly listed in your project's requirements or installed via `pip install "colour-science[all]"`.
deprecated The `pygraphviz` dependency, previously used for graph visualization, was replaced by `pydot` in v0.4.5 due to `pygraphviz`'s installation difficulties.
Fix: If you rely on graph visualization features, ensure `pydot` is installed (`pip install pydot`) and update any code that might have directly interacted with `pygraphviz` utilities, though the library generally abstracts this.
gotcha Support for `Numpy 2` was implemented in v0.4.5. While this is a feature, users with very old `Numpy` versions (e.g., pre-1.19.x) might encounter compatibility issues, as the library evolved to support newer `Numpy` APIs.
Fix: Ensure `Numpy` is updated to a modern, supported version. The library aims for broad `Numpy` compatibility but always benefits from recent `Numpy` releases.
gotcha In v0.4.6, `trimesh` was briefly and incorrectly added as a required dependency, leading to unexpected installations for some users. This was quickly fixed.
Fix: If you encountered this issue, upgrading to v0.4.7 or later will remove the unintended `trimesh` dependency unless explicitly requested as an optional extra.

Install

pip install colour-science Base Installation
pip install "colour-science[all]" With All Optional Dependencies

Imports

colour
```
import colour
```
The PyPI package is `colour-science`, but the importable module is `colour`.
RGB_to_XYZ
```
import colour
colour.RGB_to_XYZ(...)
```
Most functionalities are directly accessible via the top-level `colour` module or its submodules (e.g., `colour.models`, `colour.spaces`).

Quickstart

This quickstart demonstrates a basic colourspace conversion from sRGB to XYZ and then to CIE Lab using the `colour` library. It showcases how to import the library and use its primary conversion functions with NumPy arrays.

import colour
import numpy as np

# Define an sRGB colour in the range [0, 1]
srgb_colour = np.array([0.4, 0.6, 0.8])

# Convert sRGB to XYZ D65
x_y_z_colour = colour.sRGB_to_XYZ(srgb_colour)

# Convert XYZ D65 to Lab D65
lab_colour = colour.XYZ_to_Lab(x_y_z_colour)

print(f"sRGB Colour: {srgb_colour}")
print(f"XYZ D65 Colour: {x_y_z_colour}")
print(f"Lab D65 Colour: {lab_colour}")

view raw JSON →