Conventional 3D streamer seismic processing generally ignores any azimuth component in the data. We are used to the convenience of acquiring overlapping shot gathers in straight lines, as we can sort the data into common midpoint gathers, and then exploit the power of stack in processing to attenuate random noise. The higher the fold, the more random noise we attenuate. Conventional 3D surveys are acquired using “swath”or “racetrack” vessel shooting, wherein the survey has a single line orientation (or “survey azimuth”), and a long, narrow spread of streamers are towed by a single vessel. Apart from the front of the streamers (short source-receiver offsets), most source-receiver combinations have a relatively common azimuth (the angle between their particular vector and the survey orientation, Figure 1). Thus, the subsurface geology is illuminated only from one particular shooting direction. We assume that most coherent noise types are well behaved and we can remove them in processing. We assume that the target illumination is acceptably uniform, and we can produce clean seismic images. Most of the time these assumptions are in the ballpark of truth and our resultant seismic data allow us to achieve our exploration and appraisal objectives.