Traditionally, archaeologists have recorded sites and artefacts via a combination of ordinary still photographs, 2D line drawings and occasional cross-sections. Given these constraints, the attractions of 3D models have been obvious for some time, with digital photogrammetry and laser scanners offering two well-known methods for data capture at close range (e.g. Bates et al., 2010; Hess and Robson, 2010). The highest specification laser scanners still boast better positional accuracy and greater true colour fidelity than SfM–MVS methods (James and Robson, 2012), but the latter produce very good quality models nonetheless and have many unique selling points. Unlike traditional digital photogrammetry, little or no prior control of camera position is necessary, and unlike laser scanning, no major equipment costs or setup are involved. However, the key attraction of SfM–MVS is that the required input can be taken by anyone with a digital camera and modest prior training about the required number and overlap of photographs. A whole series of traditional bottlenecks are thereby removed from the recording process and large numbers of archaeological landscapes, sites or artefacts can now be captured rapidly, in the field, in the laboratory or in the museum. Fig. 2a–c shows examples of terracotta warrior models for which the level of surface detail is considerable.
