As I understand it: this treats image generation as constraint satisfaction. The constraints are that each NxN patch appears in the source image. The satisfaction method is arc consistency https://en.wikipedia.org/wiki/AC-3_algorithm, except, when that settles down prematurely, pick the least-constrained patch and make a random valid choice, then continue. (If this leads to getting stuck, then give up instead of backtracking.)
Is that the idea? The description wasn't completely clear to me.
Is that the idea? The description wasn't completely clear to me.