Gulf of Mexico (GoM) subsalt imaging often suffers from poor illumination due to salt-related wavefield distortion, even with full-azimuth (FAZ) acquisition. In order to image the weakly illuminated subsalt plays, isolating the signal from the noise is a crucial component of many depth-imaging practices. Reverse time migration (RTM) subsurface 3D dip-azimuth gathers, which separate seismic data into different dip/azimuth components, are utilized to address illumination problems in structure-oriented imaging. A weighting scheme using RTM 3D dip gathers for image enhancement is proposed; it is based on a priori structure information and targets noise that has conflicting dips with the structure. The sensitivity of the method to the uncertainty of the initial dipping information of the structure is further discussed and evaluated. The method is applied on subsalt structures on both a synthetic data set and a real data set with staggered-acquisition FAZ data. The tests demonstrate the necessity of signal-to-noise ratio enhancement in the imaging of FAZ, long-offset data, and the effectiveness of RTM 3D dip gathers in unmasking poorly illuminated zones.