Search code examples
pythonarraysnumpyindexingrgb

Converting an array of floats into RGBA values in an efficient way

I am trying to create a system to take an array of floats which range from 0.0 to 1.0 and convert them into RGBA values based on a lookup table. The output should be an array that is one dimension larger that the input with the last dimension being size 4 and consisting of the RGBA values.

Currently I have only been able to do this via loops. Dose anyone know of any numpy indexing methods that could achieve this same result more efficiently.

import numpy as np
import matplotlib.pyplot as plt

cyan = np.array([(x*0,x*1,x*1,255) for x in range(256)])

input_array = np.arange(0,0.8,0.05).reshape(4,4)

input_array = input_array*256

colour_array = []
for x in range(input_array.shape[0]):
    for y in range(input_array.shape[1]):
        colour_array.append(cyan[int(input_array[x,y])])
        
colour_array = np.array(colour_array).reshape(4,4,4)

plt.imshow(colour_array)

enter image description here

Solution

  • Use the following:

    shape = input_array.shape
index = input_array[*np.indices(shape).reshape(2, -1)].astype(int)
colour_array1 = cyan[index].reshape(4, *shape)

    Confirm the two are equal:

    np.allclose(colour_array, colour_array1,atol=0)
Out[62]: True

    USE THE OTHER SOLUTION!!!