White point

import numpy as np

# Assuming `xyz` is your input ndarray of shape (600, 800, 3)
xyz = np.array(xyz, dtype=np.float32)  # Ensure the array is in float

# Normalize if not already normalized
xyz /= np.max(xyz)

# White point adaptation matrix from D65 to D50
M_D65_to_D50 = np.array([
    [ 0.9555766, -0.0230393,  0.0631636],
    [-0.0282895,  1.0099416,  0.0210077],
    [ 0.0122982, -0.0204830,  1.3299098]
])

# Reshape xyz to (n, 3) for matrix multiplication
xyz_flat = xyz.reshape(-1, 3)

# Apply the white point adaptation
xyz_d50_flat = np.dot(xyz_flat, M_D65_to_D50.T)

# Reshape back to original shape
xyz_d50 = xyz_d50_flat.reshape(xyz.shape)

# Transformation matrix from XYZ (D50) to linear RGB
M_XYZ_to_RGB = np.array([
    [ 3.2404542, -1.5371385, -0.4985314],
    [-0.9692660,  1.8760108,  0.0415560],
    [ 0.0556434, -0.2040259,  1.0572252]
])

# Apply the transformation to get linear RGB
rgb_linear_flat = np.dot(xyz_d50_flat, M_XYZ_to_RGB.T)

# Reshape back to original shape
rgb_linear = rgb_linear_flat.reshape(xyz_d50.shape)

# Gamma correction function
def gamma_correction(channel):
    return np.where(channel <= 0.0031308,
                    12.92 * channel,
                    1.055 * np.power(channel, 1/2.4) - 0.055)

# Apply gamma correction
rgb_srgb = gamma_correction(rgb_linear)

# Clip values to the range [0, 1]
rgb_srgb = np.clip(rgb_srgb, 0, 1)

# rgb_srgb now contains the sRGB color data