SfM-MVS (Structure from Motion and Multiple-View Stereophotogrammetry) is part of a series of technological progresses brought to the field of earth-sciences during the last decade or so, which has allowed geoscientists to collect unprecedented precise and extensive DSMs (Digital Surface Model) for virtually no cost, rivaling LiDAR (Light Detection and Ranging) technology. Previous work on SfM-MVS in geosciences has been solely exploring data acquired for the purpose of SfM-MVS, but no research has been done in the exploration of photographic archives for geomorphological purposes. Therefore, the present publications aims to present the usage of SEM-MVS applied to historical aerial photographs in Japan, in order to (1) demonstrate the potentials to extract topographical and vegetation data and (2) to present the potential for chronological analysis of landscape evolution. SfM-MVS was implemented on black-and-white and colour aerial photographs of 1966, 1976, 1996, 2006 and 2013, using the commercial software Photoscanpro (R). Firstly, the photographs were masked, tied to GPS points; secondly the positions of the cameras and the 3D pointcloud were calculated; and thirdly the 3D surface was created. Data were then exported in the GIS software ArcGIS for analysis. Results also proved satisfactory for the reconstruction of 3D past-geomorphological landscapes in coastal areas, riverine areas, and in hilly and volcanic areas. They also prove that the height of trees and large vegetation features can also be calculated from aerial photographs alone. Diachronic analysis of the evolution in 3D landforms presented more difficulties, because the resolution of the early photographs was lower than the recent ones. Volume and surface calculations should therefore be conducted carefully. Although the method holds merit and great promise in the exploration of active landscapes that have widely changed during the 20th century; the authors have also reflected on the issues linked to large datasets, mostly because the processing of these large datasets is still in need of improvement. Moreover there is no proof that an ever increasing resolution brings any major advance to the geomorphological paradigms. (C) 2015 Elsevier B.V. All rights reserved.