Single-Shot High-Quality Facial Geometry and Skin Appearance Capture

摘要

We propose a new light-weight face capture system capable of reconstructingboth high-quality geometry and detailed appearance maps from a singleexposure. Unlike currently employed appearance acquisition systems, the proposedtechnology does not require active illumination and hence can readilybe integrated with passive photogrammetry solutions. These solutions arein widespread use for 3D scanning humans as they can be assembled fromoff-the-shelf hardware components, but lack the capability of estimatingappearance. This paper proposes a solution to overcome this limitation, byadding appearance capture to photogrammetry systems. The only additionalhardware requirement to these solutions is that a subset of the cameras arecross-polarized with respect to the illumination, and the remaining camerasare parallel-polarized. The proposed algorithm leverages the images withthe two different polarization states to reconstruct the geometry and torecover appearance properties. We do so by means of an inverse renderingframework, which solves per texel diffuse albedo, specular intensity, andhigh-resolution normals, as well as global specular roughness considering thesubsurface scattering nature of skin.We show results for a variety of humansubjects of different ages and skin typology, illustrating how the capturedfine-detail skin surface and subsurface scattering effects lead to realisticrenderings of their digital doubles, also in different illumination conditions.