By default when parsing a scene, we consider diagonally adjacent pixels to be part of the same object.
If a rule set isn’t found using that approach, we try parsing the scene again whereby we consider diagonally adjacent pixels to be part of different objects.