I find the mismatch between the cover and the spine somewhat dissonant. It would be great if an approximation of the background and text colors, and perhaps typeface as well, could be extracted from the cover image (perhaps with something like Tensorflow.js).