Restricting our purview to black and white digital artworks on a grid, we developed a lower-power version of Kolmogorov complexity, and then we found the complexity of every piece of 3x3 art. We also asked people to compare two artworks and decide which one was more visually complex as they understood the term. We used these comparisons to assign every artwork a strength rating (similar to a chess rating), and we found that the human-generated ratings were well correlated with the formula complexity of the artworks. Therefore, computers and humans largely agree on the complexity of small artworks!
Our paper (currently under submission) may be seen in the paper/ directory, and a presentation given at the Spuyten Duyvil Undergraduate mathematics conference may be found in the presentation/ directory.