基于图像的光影绘制生成是计算机图形学、计算机视觉等领域的研究热点与难点。本项目面向肖像图像视频的高逼真光影绘制需求，立足相关方向的国际技术前沿，针对“单相机条件”、“毛孔级人脸材质”、“皮肤褶皱级几何”以及“基于图像的光影绘制”等技术难点，重点开展多维度高分辨率人脸材质与几何数据集构建、皮肤属性驱动的高分辨率人脸材质几何生成、基于单个图像视频的肖像光影逼真绘制生成等方面的关键技术方法研究。力争在皮肤属性驱动的人脸皮肤材质参数化模型、基于图像的人脸材质几何生成网络、肖像视频光影绘制等方面的理论方法创新。并构建覆盖度较广的高分辨率人脸材质与几何数据集，研究开发相应的软件工具。本项目的研究工作在影视制作、互动娱乐、安防监控等行业领域具有较广阔的应用前景，例如：数字化演员的自动构建、肖像图像光影编辑、针对人脸识别的多光照训练集构建等。

Image based portrait lighting is a hot research topic in computer graphics and computer vision. The research purpose of this project is the high-fidelity portrait image/video relighting and the challenges of this projects are: single view, facial reflectance model with pore level detail and facial wrinkles modelling. Therefore, the primary works of this project are: Data set of high-resolution facial geometry and material, Facial reflectance and micro geometry generation network based on skin attribute and Single portrait image/video relighting. The technical contributions of this project will be: Facial reflectance procedure model based on facial skin property, and Portrait video relighting. Furthermore, this project will build a HD facial reflectance and micro geometry data set and develop the image-based portrait relighting software. The research contributions of this project will find wide application in the field of film production, interactive entertainment, public security and so on. The typical applications are automatic generation of digital actor, portrait image illumination editing and multi-light face data set for face recognition.