(Aside) We could represent an image as a struct of arrays, where we have one array of every single R value in the image, one array of every single G value, and so forth, but this method is less common and only more efficient if you expect to read all the pixels in a given channel.