If the objective is to get the maximum 'visual' difference for two hashes, why not re-hash both of them first? That results in any small changes causing a huge difference in the final outputs.
Each visual is a representation of a single hash. If it was designed to compare two hashes, it could simply show a true/false status because the only purpose is to see if they match.
True if you did, but I don't think that's the case. Even a single bit difference would send the bishop in a different direction and therefore cause a major difference in the count of squares visited.
I did a quick test to verify this on OP's site. Just incrementing a sample hash by 1 does substantially change the visual.
It's possible two completely different hashes show similar visuals, but I don't think two similar hashes normally do. Hence no reason to re-hash.