【作者】 蔡飞龙；

【导师】 于金辉；

【作者基本信息】 浙江大学 ， 数字化艺术与设计， 2012， 博士

【摘要】 随着世界各国对传统非物质文化遗产保护的重视,非物质文化遗产的数字化建模越来越受到国内外学者的关注。被誉为国粹的京剧作为中国传统文化的重要代表,其独具特色的脸谱艺术更是我国传统文化宝库中的精华,具有高度审美价值、文化价值、研究价值和应用价值。本文以传统京剧脸谱为具体研究对象,通过对脸谱艺术中的纹样、程式、色彩及艺术特征进行分析和归纳,在计算机中对脸谱进行数字化建模和处理,并利用图形学技术合成脸谱图像和表情动画。通过对京剧脸谱的数字化建模和绘制技术的研究,可以让我们更有效的保护和继承这个民族瑰宝,同时也在一定程度上促进了计算机技术的发展。本文是针对京剧脸谱的数字化保护技术研究,不是简单对脸谱各种信息的进行数字化记录和呈现,更主要的是利用计算机技术对脸谱进行建模,并设计开发脸谱辅助设计和展示系统来促进脸谱的保护与研究。主要就如何建立脸谱纹样数据库、如何合成矢量化脸谱和表情脸谱、如何多角度的观察脸谱表情动画、如何真实感的绘制三维脸谱等问题展开研究。本文的主要贡献包括以下几个方面：1.建立了一个脸谱纹样库和脸谱辅助合成系统。首先分析京剧脸谱绘制过程以及基础纹样形状特征,然后基于分层原理构造出矢量化的纹样数据库。在合成脸谱阶段,用户只需按照脸谱绘制顺序逐层选取所需的纹样,根据自己的创作需要对各层的纹样进行组合得到最终的京剧脸谱图案。系统还提供一系列变形工具供用户对脸谱局部纹样进行编辑修改,以其生成更多的、富有变化的京剧脸谱图案。2.提出了一个基于矢量对象的分层驱动脸谱表情的生成方法。将脸谱纹样和表情动作分开处理,首先构造出矢量化纹样库,然后根据脸部运动编码系统标准将脸谱表情分解成40个动作单元。在合成脸谱过程中,用户只需按照脸谱绘制顺序和创作需要逐层选取所需的纹样即可得到一个脸谱图案,脸谱表情则通过自由变形(FFD)技术对脸谱纹样变形得到。3.提出了一个混合驱动的2.5维脸谱表情动画的方法。利用脸部运动编码系统,将脸谱表情动作分解成基本表情、眼睛、口型、头部转动四个部分。然后根据四个部分自身的运动原理采用不同动画技术来驱动。在生成脸谱动画过程中,用户只需选定所需脸谱图案和配音,设定好关键帧的表情,即可得到一段生动的脸谱动画。4.提出了两种适合京剧脸谱的真实感绘制方法：离线和实时绘制。研究比较了目前人脸皮肤真实感绘制的技术方法,找到了适合京剧脸谱的离线和实时真实感绘制的方法。在离线绘制方法中,先利用高精度法线图来增强皮肤细节,再把BSSRDF光照模型绘制的次表面散射和漫反射与高光反射相结合,离线绘制出真实感的脸谱；在实时绘制方法中,运用D’Eon的皮肤实时绘制技术,修改其中的模拟散射的高斯混合参数的比重以适合表现真实感的京剧脸谱化妆效果。更多还原

【Abstract】 Along with most countries attaching much value to the protection of traditional intangible cultural heritage, digital modeling of intangible cultural heritages raised attentions of researches at home and abroad. Peking Opera is an important representative of Chinese traditional cultures, and its unique facial makeup art is the essence of our cultural treasures. Facial makeups in Peking opera have very high values in terms of aesthetic, culture, research and application. This thesis focuses on traditional facial makeups in Peking Opera. Firstly we analyzed patterns, colors and artistic features of the traditional facial makeups, and used compute to model and process them. Then we used graphic technologies to synthesize digital images and animations of facial makeups and render photo-realistic 3D facial makeups. Digital modeling for facial makeup in Peking Opera can effectively protect and inherit this priceless national treasure; to a certain extent, it also promotes the development of digital modeling technologies.This thesis mainly studies digital modeling and rendering technologies of Facial makeup in Peking Opera. The related problems which were discussed in this thesis include:how to build a facial makeup’s synthesis system; how to control a facial makeup’s expression; how to observe facial expression animation in multi-view ways; how to render photo-realistic 3d facial makeups. The main contributions of this thesis are as follows:1. A computer-aided design system for facial makeup synthesis in Peking Opera is established. Firstly we analyzed the drawing process of facial makeups and characteristics of the patterns used in it, and then constructed a pattern bank based on layers corresponding to the drawing process. During the synthesis step, users picked up some patterns from the pattern bank and composed them to form new facial makeups. The system also provides a serial of tools to let users edit or modify local patterns.2. A vectorizated hierarchical driven model of facial makeup’s expression is proposed. Facial makeup’s patterns and expression actions are handled separately. Firstly we constructed a vectorised pattern database, and then facial expression was decomposed 40 local action units based on the standard of facial action coding system. In the synthesis process, the user only need to follow the facial makeup layer draw order and select the desired patterns, he can get a facial makeup. Facial makeup’s expression is achived through using the free-form deformation (FFD) technique to facial makeup patterns. Users can also control local expressions by some parameters.3. A hybrid-driven 2.5D model for expression animation of facial makeup in Peking Opera is proposed. Firstly we decomposed facial makeup’s motion into four parts:basic expressions, eye motions, mouth motions, and head rotations according of the standard of facial action coding system. Each part is controlled by different animation technolegies. In the synthesis process of facial makeup’s animation, users only need to select desired facial makeup and sound, and set expressions of keyframes, the system produces a vivid facial makeup’s animation.4. Two photo-realistic rendering methods for facial makeup is proposed. By analyzing the current photo-realistic rendering methods of human facial skin, we found off-line and real-time rendering methods suitable for facial makeup. For the off-line rendering method, Firstly we used a high resolution normal map to enhance the skin detail, and then combined the BSSRDF skin model and a physical highlight model to render realistic facial makeup; For the real-time rendering method, based D’Eon real-time skin model, we modified parameters of the Gaussian mixture to render realisticly subsurface scattering effect of real facial makeup.更多还原