Generation of Variable Human Faces from 3D Scan Dataset

摘要

Generating human faces is an important task in many research and application fields, including the gaming industry. When the scene contains many characters, it becomes impracticable to create all individual characters manually. On the other hand, the requirement for the different appearance of faces of individuals in a crowd is now more in demand. In this paper we propose our solution to the automatic generation of human faces. Our solution synthesizes facial parts coming from scans of real human faces; a process that is completely automatic. However, the user has the possibility to further adjust the composite by designing replacements, leading to a desired appearance. The final composite can be exported and attached to a given avatar. As the perception of the variability of resulting composites is in our focus as well, we also present results of the user study which was designed for this purpose. The study aims to reveal the minimal changes of the facial parts which are necessary to make in order to perceive a given facial model as different from the original one.