You might argue that work like Mondrian's implies additional information, in the form of implied commentary on centuries of art history (arguably necessary for his work to be considered interesting or "good"). You might then argue that this makes up for the lower explicit information density.
