I'm the other developer. There is actually some pretty cool math behind this: The underlying algorithm has not been published. At its core is a color median fi

I'm the other developer. There is actually some pretty cool math behind this: The underlying algorithm has not been published. At its core is a color median filter, a block-L1-minimizing generalization of the well-known scalar median filter. Applying such a filter to an image rounds out corners and destroys fine features, so a separate recovery step is needed. Filtered pixels that differ significantly from their original values are iteratively replaced with the distance-minimizing convex combination of neighboring filtered pixels, digging out corners and other destroyed features without re-introducing noise.