It is a pitfall to judge a supposed noncosmetic process by a cosmetic effect. I once got caught. The process was migration before stack. The feature that was deemed desirable was the relative strength of the steepest clear event on the record, a fault-plane reflection. But even gain control can affect dip spectra! I hoped the process was working by correctly eliminating some of the rejection of steep dips by CDP stack. Perhaps it was, but how could I know whether this was happening or whether dips were being accidentally enhanced by spatial filtering?