Detecting small subsurface features such as faults, fracture swarms, steep reef edges, and channel edges is a routine task in seismic interpretation. Common seismic attributes used for this task, such as curvature and incoherence, are derived from conventional images containing both reflections and diffractions. An emerging alternative approach is to interpret these edge features using the diffraction-imaging technique, which images diffracted energy separately. A diffraction-imaging volume is generated by separating specular reflections from diffractions in an unmigrated volume followed by migration of those diffractions. Different methods can be used to separate diffractions from reflections on unmigrated data. In a case history, they are separated with a plane-wave destruction (PWD) filter. An evaluation of diffraction-imaging performance is applied to a 3D seismic data set from the Cooper Basin of Australia. Diffraction-imaging horizon maps are compared with corresponding maps made from incoherence and curvature volumes. The resolution and the ability to detect faults, fractures, and channel-edge features are compared. The results show that diffraction imaging provides superior vertical and spatial resolution over conventional incoherence and curvature attributes for mapping faults and stratigraphy.