When the spatial dynamic of the imaged source is large (i.e. when the ratio of the largest source structure over the synthesized resolution is large), the basic CLEAN algorithms may (rarely) turn smooth area of the source into a serie of ridges and stripes. Indeed, when the dirty beam pattern is subtracted from a smooth feature of the dirty map, the sidelobes patterns appear in the residual map. The search for the next clean component will then pick first the pixels in the sidelobes pattern amplifying this pattern. Several variants of CLEAN has been devised to answer this problem.