The present work shows the results of a combined field and Structure from Motion (SfM) survey performed on the detachment surface of a shallow rockslide that occurred in the Rosandra Valley (Trieste, NE Italy), which was aimed at testing the use of 3D models obtained from Remote Sensing (RS) techniques to identify joint sets affecting unstable rock masses. According to discontinuity orientation data acquired from the field (N = 223), the investigated rock mass is affected by at least nine joint sets characterised by a notable variability. The extraction of joint sets from the 3D point cloud representing the surveyed rock outcrop was strongly sensitive to the point cloud density and the values of the controlling parameters of the density function embedded within the discontinuity extractor. This work demonstrates that, in order to properly identify rock joint sets, the exclusive application of a RS approach cannot fully substitute the traditional field survey, and the estimation of discontinuity sets should be integrated with joint orientation data acquired using a geological compass. To maximise its capabilities, the semi-automatic discontinuity set extraction from 3D point clouds should always be supported by a significant statistical sample of joint orientation measurements that are preliminarily collected from the field.