Reading this just as I fiddled with georeferencing some images with JOSM's PicLayer plugin. It uses maximum 3 control points to define an affine transform, howe

Reading this just as I fiddled with georeferencing some images with JOSM's PicLayer plugin. It uses maximum 3 control points to define an affine transform, however for flat images (not taken with camera) it's more beneficial to use only 2 points (we do not expect shear). I wish I discovered this earlier when georeferencing local park plans.