Automated colourisation algorithms typically involve a two-step process. First, a grayscale image is passed through a neural network that estimates the colour information (base palette). Then, a post-processing step is applied to refine the colourisation and enhance the visual quality.