Add lightmap mask to only run encoding on pixels actually used in the lightmap (large lightmaps have to be allocated to make sure there's room, so that means large parts of the lightmap usually go unused)