Containing 15 scenes from DTU and 18 scenes from BlendedMVS, all with monocular normals estimated using Omnidata and foreground masks.