Add support for user-defined color checker charts.

[NoRoKr]

From my investigations, it does not seem that colour allows creation of COLOURCHECKERS or COLOURCHECKERS_SDS objects defined by the user.
Preferably, a user who has access to a colorimeter or spectroradiometer could measure their own Macbeth charts either to evaluate the measurement equipment or the quality of charts and display them using colour.plotting.plot_multi_colour_checkers.

Currently, I am able to plot each individual patch using the plot_multi_colour_swatches function, but there are a number of occasions where having the entire color checker chart in one go would be nice.

[KelSolaar]

Hi @NoRoKr,

It should be doable as follows:

# %%
import numpy as np
from collections import OrderedDict

import colour
import colour.plotting

CRC = np.array([
    [0.4002, 0.3504, 0.1005],
    [0.3773, 0.3446, 0.3582],
    [0.2470, 0.2514, 0.1933],
    [0.3372, 0.4220, 0.1329],
    [0.2651, 0.2400, 0.2427],
    [0.2608, 0.3430, 0.4306],
    [0.5060, 0.4070, 0.3005],
    [0.2110, 0.1750, 0.1200],
    [0.4533, 0.3058, 0.1977],
    [0.2845, 0.2020, 0.0656],
    [0.3800, 0.4887, 0.4429],
    [0.4729, 0.4375, 0.4306],
    [0.1866, 0.1285, 0.0611],
    [0.3046, 0.4782, 0.2339],
    [0.5385, 0.3129, 0.1200],
    [0.4480, 0.4703, 0.5910],
    [0.3635, 0.2325, 0.1977],
    [0.1958, 0.2519, 0.1977],
    [0.3101, 0.3163, 0.9001],
    [0.3101, 0.3163, 0.5910],
    [0.3101, 0.3163, 0.3620],
    [0.3101, 0.3163, 0.1977],
    [0.3101, 0.3163, 0.0900],
    [0.3101, 0.3163, 0.0313]
])
CRC += (np.random.random(CRC.shape) - 0.5) * 0.075

# CRC plots.
colour_checker_t = colour.characterisation.ColourChecker(
    'Test', OrderedDict([(str(i), x) for i, x in enumerate(CRC)]),
    colour.ILLUMINANTS['CIE 1931 2 Degree Standard Observer']['C'])

colour.plotting.plot_multi_colour_checkers(['ColorChecker 1976', colour_checker_t])

Did you have anything else in mind?

[NoRoKr]

Hi @KelSolaar,
I was ultimately able to get it to work this way. Thank you.

I do have a few concerns that I believe could be improved upon here. I could move these into a separate issue if necessary.

from colour import plotting

ill = colour.ILLUMINANTS['CIE 1931 2 Degree Standard Observer']['D65']
ref = colour.COLOURCHECKERS_SDS['babelcolor average']

sd_to_xyY = lambda sd : colour.XYZ_to_xyY( colour.sd_to_XYZ(sd) / 100,  ill) 

xyYs = map(sd_to_xyY, ref.values())

CC = colour.characterisation.ColourChecker(
        'Test 1', OrderedDict([(str(i), x) for i, x in enumerate(xyYs)]),
        ill)

colour.plotting.plot_single_colour_checker(CC)
colour.plotting.plot_multi_colour_checkers(['babelcolor average', CC])

First, the requirement to rescale the data seems like it shouldn't be necessary by default.

Second, this outputs the following image:

There's a clear issue with the color in the converted one. Perhaps I am missing a necessary parameter to have CC rendered correctly in sRGB?

[NoRoKr]

With regards to my last comment, I forgot the basics of color for a second and see my mistake (not incorporating the light source correctly)!

[KelSolaar]

Hi @NoRoKr,

That is what I was about to say! You need to pass the illuminant to the spectral conversion function :]

In respect to the division by 100, there is a good reason why it is not the default: https://colour.readthedocs.io/en/develop/basics.html#domain-range-scales

By default, the conversion from spectral to tristimulus values generates percentages, i.e. reflectance, transmittance or absorptance percentages.

[NoRoKr]

sd_to_xyY = lambda sd : colour.XYZ_to_xyY(colour.sd_to_XYZ(sd, illuminant=ls) / 100 ) 

This fixes my previous problem