Colour 0.4.0 - Alpha Milestone

Over a year in the making, this release integrates most of the GSoC 2021 work from Cédric (@villirion), all the code from Geetansh (@SGeetansh) and the remaining GSoC 2020 code from Nishant (@njwardhan). We would like to thank them again for their great contributions!

Python 2.7 support has been dropped and the minimal version is Python 3.8 as per https://scientific-python.org/. The following minimal dependency versions are also required:

• imageio >= 2, < 3
• numpy >= 1.19, < 2
• scipy >= 1.5, < 2

The highlights of this release are as follows:

• Colour now runs on iOS and iPadOS with Pyto.
• The import of colour is now 3.6 times faster.
• Typing annotations have been added and the codebase is checked with Mypy.
• The documentation has been updated and uses the pydata-sphinx-theme and has better compliance with PEP257.
• The code formatter is now Black,
• Many Python 3 features such as f-Strings or the dataclass decorator have been adopted.
• The plotting API has been improved to be more consistent when setting the colours of some figures, e.g. spectral or planckian locus.
• New colour appearance models:
  • Zhai and Luo (2018) chromatic adaptation model.
  • Kim, Weyrich and Kautz (2009) colour appearance model.
  • ZCAM colour appearance model.
  • Helmholtz-Kohlrausch effect estimation.
• New colour models:
  • Oklab colour model.
  • Hanbury (2003) IHLS (Improved HLS) colourspace.
  • DIN99b, DIN99c, and DIN99d refined formulas.
  • ProLab colourspace.
  • Sarifuddin and Missaoui (2005) HCL colourspace.
• New RGB colourspaces and transfer functions:
  • Nikon N-Gamut colourspace and the N-Log log encoding.
  • Blackmagic Wide Gamut colourspace and the associated Blackmagic Film Generation 5 OETF.
  • DaVinci Intermediate OETF.
  • RED Log3G10 encoding and decoding curves with linear extension.
• Other notable features:
  • Huang et al. (2015) power-functions.
  • LUT 1D, LUT 3x1D and LUT 3D inversion.
  • UPRTek and Sekonic spectral data parsers.
  • SPImtx LUT input and output.
  • R'G'B' to Y'CbCr matrices computation.
  • Gamut ring/section plotting.
  • Rösch-MacAdam colour solid hue lines.
  • Support for OpenColorIO processor.
    • Note that the optional dependency is not specified in the pyproject.toml, see AcademySoftwareFoundation/OpenColorIO#1573 for more information.

Thanks again to all the contributors to this release!

• @aforsythe
• @fredsavoir
• @ilia3101
• @jedypod
• @JGoldstone
• @nick-shaw
• @njwardhan
• @Paul-Sims
• @ramparvathaneni
• @romanovar
• @saransh
• @SGeetansh
• @sobotka
• @villirion
• @zachlewis

Features

Typing

• The API has been fully annotated with typing annotations, a new colour.hints sub-package exporting all the hints has been created.

Performance

• Import time has been reduced to ~1.5secs from 5.5secs on @KelSolaar's MacBook Pro (Retina, 13-inch, Mid 2014) by using a lazy load mechanism for the spectral data and delaying various imports in the colour.plotting.tm3018 module. See PR #840 for more information.
• Repetitive spectral computations are now cached and a small cache management API has been created, see colour.utilities.CACHE_REGISTRY attribute for more information.
• The colour.sd_to_XYZ and colour.msds_to_XYZ definitions now use the same n-dimensional code under the hood. Some minor numerical differences are expected as the explicit summations and multiplications have been replaced with np.dot.

colour.adaptation

GSoC - 2021

• Implement support for Zhai and Luo (2018) chromatic adaptation model with colour.adaptation.chromatic_adaptation_Zhai2018 definition. (@villirion, @KelSolaar)

colour.appearance

GSoC - 2021

• Implement support for Kim, Weyrich and Kautz (2009) colour appearance model with colour.XYZ_to_Kim2009 and colour.Kim2009_to_XYZ definitions. (@villirion, @KelSolaar)

• Implement support for ZCAM colour appearance model with the colour.XYZ_to_ZCAM and colour.ZCAM_to_XYZ definitions. (@KelSolaar)
• Implement support for Helmholtz-Kohlrausch effect estimation with colour.HelmholtzKohlrausch_effect_object_Nayatani1997 and colour.HelmholtzKohlrausch_effect_luminous_Nayatani1997 definitions. (@ilia3101, @KelSolaar)
• Colour appearance models now use a dataclass-like class output instead of the namedtuple and thus are mutable.

colour.characterisation

• Various ACES Input Device Transform computation objects have been updated to generate additional data and support more features for the Academy Input Device Transform (IDT) calculator. (@KelSolaar, @aforsythe)
• Add colour.camera_RGB_to_ACES2065_1 definition. (@KelSolaar, @aforsythe)

colour.colorimetry

• Implement support for spectral uniformity computation with colour.spectral_uniformity definition. (@KelSolaar)
• Add colour.TVS_ILLUMINANTS attribute providing a reference for the CIE XYZ tristimulus values of the CIE illuminants. (@KelSolaar)
• Ensure that colour.SpectralShape and dict KeysView class instances can be passed as domain to colour.SpectralDistribution and colour.MutliSpectralDistributions classes. (@KelSolaar)
• The colour.colorimetry.yellowness_ASTME313 definition has been updated to use the recommended Yellowness Index equation as given by ASTME313. The alternative method has been renamed to colour.colorimetry.yellowness_ASTME313_alternative. (@KelSolaar, @romanovar)

colour.difference

• Implement support for STRESS index computation according to García et al. (2007) method with colour.index_stress definition and colour.INDEX_STRESS_METHODS attribute. (@KelSolaar)
• Implement support for Huang et al. (2015) power-functions improving colour-difference formulas with colour.difference.power_function_Huang2015 definition. (@KelSolaar)

colour.io

GSoC - 2020

• Implement support for LUT 1D, LUT 3x1D and LUT 3D inversion with the colour.LUT1D.invert, colour.LUT3x1D.invert and colour.LUT3D.invert methods. (@njwardhan, @KelSolaar)

GSoC - 2021

• Implement support for UPRTek and Sekonic spectral data parsers with the colour.SpectralDistribution_UPRTek and colour.SpectralDistribution_Sekonic classes. (@SGeetansh, @KelSolaar)

• Implement support for OpenColorIO processor with colour.io.process_image_OpenColorIO definition. (@KelSolaar)
• Implement support for SPImtx LUT input and output with new colour.io.read_LUT_SonySPImtx and colour.io.write_LUT_SonySPImtx definitions and colour.io.LUTOperatorMatrix support class. (@nick-shaw, @KelSolaar, @zachlewis)
• The colour.io.tm2714.Header_IESTM2714 class can now be hashed and compared for equality. (@JGoldstone)

colour.models

GSoC - 2021

• Implement support for Hanbury (2003) IHLS (Improved HLS) colourspace with colour.RGB_to_IHLS and colour.IHLS_to_RGB definitions. (@SGeetansh, @KelSolaar)
• Implement support for DIN99b, DIN99c, and DIN99d refined formulas in colour.Lab_to_DIN99 and colour.DIN99_to_Lab definitions. (@SGeetansh)
• Implement support for ProLab colourspace with colour.ProLab_to_XYZ and colour.XYZ_to_ProLab definitions. (@SGeetansh, @KelSolaar)

• Implement support for Sarifuddin and Missaoui (2005) HCL colourspace with colour.RGB_to_HCL and colour.HCL_to_RGB definitions. (@Saransh-cpp, @KelSolaar)
• Implement wrapper colour.XYZ_to_ICTCP and colour.ICTCP_to_XYZ definitions. (@KelSolaar)
• Implement wrapper colour.XYZ_to_CAM02LCD, colour.CAM02LCD_to_XYZ, colour.XYZ_to_CAM02SCD, colour.CAM02SCD_to_XYZ, colour.XYZ_to_CAM02UCS and
  colour.CAM02UCS_to_XYZ definitions. (@KelSolaar)
• Implement wrapper colour.XYZ_to_CAM16LCD, colour.CAM16LCD_to_XYZ, colour.XYZ_to_CAM16SCD, colour.CAM16SCD_to_XYZ, colour.XYZ_to_CAM16UCS and
  colour.CAM16UCS_to_XYZ definitions. (@KelSolaar)
• Implement support for R'G'B' to Y'CbCr matrices computation with colour.matrix_YCbCr and colour.offset_YCbCr definitions. (@KelSolaar, @nick-shaw)
• Implement support for Oklab colour model with colour.XYZ_to_Oklab and colour.Oklab_to_XYZ definitions. (@KelSolaar)
• Implement support for Nikon N-Gamut colourspace and the N-Log log encoding and decoding curves with colour.models.RGB_COLOURSPACE_N_GAMUT class and colour.models.log_encoding_NLOG and colour.models.log_decoding_NLOG definitions. (@sobotka, @KelSolaar)
• Implement support for RED Log3G10 encoding and decoding curves that uses a linear extension as given in the final version of White Paper on REDWideGamutRGB and Log3G10. (@jedypod)
• Implement support for Blackmagic Wide Gamut colourspace and the associated Blackmagic Film Generation 5 OETF and its inverse with colour.models.RGB_COLOURSPACE_BLACKMAGIC_WIDE_GAMUT attribute, colour.models.oetf_BlackmagicFilmGeneration5 and colour.models.oetf_inverse_BlackmagicFilmGeneration5 definitions. (@KelSolaar)
• Implement support for DaVinci Intermediate OETF and its inverse with colour.models.oetf_DaVinciIntermediate and colour.models.oetf_inverse_DaVinciIntermediate definitions. (@fredsavoir, @KelSolaar)

colour.plotting

• Implement support for gamut ring/section plotting with the colour.plotting.plot_visible_spectrum_section and colour.plotting.plot_RGB_colourspace_section definitions. (@KelSolaar)

figure, axes = plot_visible_spectrum_section(
    model='DIN99', origin=0.5, section_opacity=0.15, standalone=False)

bounding_box = [
    axes.get_xlim()[0],
    axes.get_xlim()[1],
    axes.get_ylim()[0],
    axes.get_ylim()[1]
]

section_colours = colour.notation.HEX_to_RGB(
    colour.plotting.CONSTANTS_COLOUR_STYLE.colour.cycle[:4])

origins = []
legend_lines = []
for i, RGB in zip(np.arange(0.5, 0.9, 0.1), section_colours):
    origins.append(i * 100)
    plot_RGB_colourspace_section(
        colourspace='sRGB',
        model='DIN99',
        origin=i,
        section_colours=RGB,
        section_opacity=0.15,
        contour_colours=RGB,
        axes=axes,
        standalone=False)
    legend_lines.append(
        Line2D([0], [0], color=RGB, label='{0}%'.format(i * 100)))

axes.legend(handles=legend_lines)

colour.plotting.render(
    title='Visible Spectrum - 50% - sRGB Sections - {0}% -  DIN99'.format(
        origins),
    axes=axes,
    bounding_box=bounding_box)

• Improve zorder across the colour.plotting sub-package and implement various new parameters to customize better figures such as the Chromaticity Diagram.

colour.volume

• Implement support to generate the Rösch-MacAdam colour solid hue lines with the colour.volume.XYZ_outer_surface/colour.volume.solid_RoschMacAdam definition. (@KelSolaar)

import colour

figure, axes = colour.plotting.plot_chromaticity_diagram_CIE1931(
    standalone=False)

bins = len(colour.volume.spectrum.SPECTRAL_SHAPE_OUTER_SURFACE_XYZ.range())

xy = colour.XYZ_to_xy(
    colour.volume.solid_RoschMacAdam(
        point_order='Pulse Wave Width', filter_jagged_points=True)[1:-1])

for i, j in enumerate(range(0, xy.shape[0], (bins - 1) // 2)):
    hue_lines = np.vstack([xy[j:j + (bins - 1) // 2, :], xy[-1]])
    axes.plot(hue_lines[:, 0], hue_lines[:, 1], c='k')

colour.plotting.render()

Fixes

colour

• Ensure that colour runs on iOS and iPadOS. (@KelSolaar)

colour.colorimetry

• Ensure that the cache used by colour.sd_to_XYZ definition returns a copy of the data. (@KelSolaar)

colour.plotting

• Fix IES TM-30-18 Colour Rendition Report Local Color Fidelity sub-plot values. (@KelSolaar, @Paul-Sims)

Changes

colour.algebra

colour.biochemistry

colour.characterisation

colour.colorimetry

• colour.SpectralShape class must always be initialised with the start, end and interval parameters. (@KelSolaar)

colour.geometry

colour.io

colour.models

colour.plotting

colour.recovery

colour.temperature

colour.utilities

colour.volume