ファイサル ズバイル クェレシ
Construction of Facial Tissue using Cyberware and
Computer Tomography Data
Abstract:Physically-based facial modeling has proven promising for facial animation. These models, consisting of mathematical abstractions for skin tissue, muscles and skeletal sub-surface, allow a wide range of expressions at high refresh rates. Geometery of the face is captured using stereo photogrammetery or optical surface scanners and used to setup the physically-based model. Facial expressions are dependant on the mechanical properties of the skin. These properties, in turn, depend upon elastin, collagen and ground susbtance [LS86]. The mechanics of the skin also depend upon the tissue thickness at various parts of the face. At forehead, the skin tissue is only few millimeters thick where as around cheeks, it is much more dense. Similarly, facial tissue has a variety of sub-cutaneous attachments. To correctly model the physical properties of the skin tissue, we need to take into consideration all the above things. In a realistic physically-based facial model we need to include both skin tissue and the underlying skeletal bone sub-surface.
Waters ([WT90]) proposed a way to estimate skeletal sub-surface geometery from the epidermis data. Although this simplification is good enough for animation purposes, it under-estimates the influence of the varying skin thickness. Also, lack of an accurate skull model makes jaw motion both difficult and unrealistic. Data from optical scanners and/or stereo photogrammetery contains information about the color and geometery of facial skin exo-surface and computer tomography (CT) data can be used to get skin thickness and underlying skeletal information. Using these data sets, we can compute skin thickness at various points on the face and build a more accurate skull-bone model.
In this report, we propose a way to combine information from the optical surface scanner (in our case cyberware range and reflectance data) and CT data to construct a physically-based facial model. This model takes into account the variations in skin thickness and creates more accurate facial expressions.