basic-colormath

basic_colormath

Everything I wanted to salvage from the python-colormath library ... with no numpy deps and 14x speed.

• Perceptual (DeltaE CIE 2000) and Euclidean distance between colors
• Conversion between RGB, HSV, HSL, and 8-bit hex colors
• Simple, one-way conversion to Lab
• Some convenience functions for RGB tuples and 8-bit hex color strings
• Vectorized functions for numpy arrays
• Proximity and cross-proximity (rectangular) matrices for numpy arrays

Lab color format is exciting because it can cover a larger colorspace than RGB. But don't get too excited yet. If you convert an RGB tuple to Lab with no additional information, then the result will—of course—not contain more information than the RGB tuple you converted from. Other parameters are necessary to get anything out of these elaborate formats. I don't know how to do that, and most likely neither do you, so why not drop all of that complexity?

I've installed python-colormath on a lot of projects. The library does many interesting things, but most of what I wanted was perceptual color distance. This requires Lab colors, which have more parameters than an RGB tuple provides. Colormath didn't use those parameters, so the result didn't require the elaborate classes and methods provided by Colormath.

The color distance I provide here is DeltaE CIE 2000. Aside from (presumably) some specialized applications, this is the best of the multiple color distances provided by python-colormath. Tuples in, float out, and with a lot more speed. It doesn't use all of those expert parameters, but neither did Colormath. This is the same result you'll get from any of the online DeltaE calculators you're likely to find.

This library is more or less specialized for working with "upscaled" RGB tuples ([0, 255], [0, 255], [0, 255]). Functions will take floats or ints in that range and return floats. If you want ints, use float_tuple_to_8bit_int_tuple. This is dramatically better int conversion than int(float) or int(round(float)), so use it insead of those.

distance functions

Rgb = Annotated[tuple[float, float, float], ([0, 255], [0, 255], [0, 255])]
Hsv = Annotated[tuple[float, float, float], ([0, 365), [0, 100], [0, 100])]
Hsl = Annotated[tuple[float, float, float], ([0, 365), [0, 100], [0, 100])]
Lab = Annotated[tuple[float, float, float], ([0, 100], [-128, 127], [-128, 127])]
Hex = Annotated[str, "#000fff"]

rgb_to_lab(rgb: Rgb) -> Lab:
    # Converts RGB to Lab. To optionally cache for get_delta_e_lab

hex_to_lab(hex: Hex) -> Lab:
    # Converts hex to Lab. To optionally cache for get_delta_e_lab

get_delta_e(rgb_a: Rgb, rgb_b: Rgb) -> float:
    # Calculate the Delta E (CIE 2000) between two RGB colors.
    # This is the one you'll usually want.

get_delta_e_hex(hex_a: Hex, hex_b: Hex) -> float:
    # Calculate the Delta E (CIE 2000) between two hex colors.
    # Takes hex colorstrings.

get_delta_e_lab(lab_a: Lab, lab_b: Lab) -> float:
    # Calculate the Delta E (CIE2000) of two Lab colors.
    # To call with cached Lab values.

get_sqeuclidean(rgb_a: Rgb, rgb_b: Rgb) -> float:
    # Calculate the squared Euclidean distance between two RGB colors.

get_sqeuclidean_hex(hex_a: Hex, hex_b: Hex) -> float:
    # Calculate the squared Euclidean distance between two HEX colors.

get_euclidean(rgb_a: Rgb, rgb_b: Rgb) -> float:
    # Calculate the Euclidean distance between two RGB colors.

get_euclidean_hex(hex_a: Hex, hex_b: Hex) -> float:
    # Calculate the Euclidean distance between two HEX colors.

other conversions

Converts to other simple formats.

def rgb_to_hsv(rgb: Rgb) -> Hsv: ...

def hsv_to_rgb(hsv: Hsv) -> Rgb: ...

def rgb_to_hsl(rgb: Rgb) -> Hsl: ...

def hsl_to_rgb(hsl: Hsl) -> Rgb: ...

def rgb_to_hex(rgb: Rgb) -> Hex: ...

def hex_to_rgb(hex_: Hex) -> Rgb: ...

convenience functions

_Ratio = float | tuple[float, ...] | None

scale_rgb(rgb: Rgb, scalar: float) -> Rgb:
    # Scale an rgb tuple by a scalar.

mix_rgb(*rgb_args: Rgb, ratio: _Ratio=None) -> Rgb:
    # Mix any number of rgb tuples.

    :param rgb_args: rgb tuples ([0, 255], [0, 255], [0, 255])
    :param ratio: 0.0 to 1.0 for the weight of the first rgb_arg or a tuple of floats
        to distribute across rgb_args or None for equal ratios. Ratios will be
        normalized and (if fewer ratios than colors are provided) the remaining
        ratios will be equal.
    :return: rgb tuple ([0, 255], [0, 255], [0, 255])

scale_hex(hex_: Hex, scalar: float)-> Hex
    # Scale a hex color by a scalar.

mix_hex(*hex_args: Hex, ratio: _Ratio=None) -> Hex
    # Mix any number of hex colors.

better float to int conversion

def float_to_8bit_int(float_: float | int) -> int: ...

def float_tuple_to_8bit_int_tuple(rgb: Rgb) -> tuple[int, int, int]: ...

vectorized functions

If you have numpy installed in your env, basic_colormath will provide vectorized versions of most functions along with proximity matrices and cross-proximity matrices.

Function	Vectorized Function	(Cross-) Proximity Matrix
float_to_8bit_int	floats_to_uint8
get_delta_e	get_deltas_e	get_delta_e_matrix
get_delta_e_hex	get_deltas_e_hex	get_delta_e_matrix_hex
get_delta_e_lab	get_deltas_e_lab	get_delta_e_matrix_lab
get_euclidean	get_euclideans	get_euclidean_matrix
get_euclidean_hex	get_euclideans_hex	get_euclidean_matrix_hex
get_sqeuclidean	get_sqeuclideans	get_squeclidean_matrix
get_sqeuclidean_hex	get_sqeuclideans_hex	get_sqeuclinean_matrix_hex
hex_to_rgb	hexs_to_rgb
hsl_to_rgb	hsls_to_rgb
hsv_to_rgb	hsvs_to_rgb
rgb_to_hex	rgbs_to_hex
rgb_to_hsl	rgbs_to_hsl
rgb_to_hsv	rgbs_to_hsv
rgb_to_lab	rgbs_to_lab
mix_hex
mix_rgb
scale_hex
scale_rgb

proximity matrices

(Cross-)proximity matrix functions take a (-1, 3) array of color values or (-1,) array of hex color strings and return a proximity matrix. This is a square matrix with the same number of rows and columns as the number of colors provided. The value at row i and column j is the distance between the color at index i and the color at index j. The diagonal is always zero.

An optional second argument creates a cross-proximity matrix. This is a matrix with the same number of rows as the first argument and the same number of columns as the second argument. The value at row i and column j is the distance between the color at index i in the first argument and the color at index j in the second argument.

If you need more

Sadly, python-colormath has been abandoned, long enough now that a numpy function on which it relies has been not only deprecated but removed. If you still need to use python-colormath, patch np.asscalar:

import numpy as np
import numpy.typing as npt

def _patch_asscalar(a: npt.NDArray[np.float64]) -> float:
    """Alias for np.item(). Patch np.asscalar for colormath.

    :param a: numpy array
    :return: input array as scalar
    """
    return a.item()

np.asscalar = _patch_asscalar  # type: ignore