【作者】 徐颂华；

【导师】 潘云鹤；

【作者基本信息】 浙江大学 ， 计算机科学与技术， 2007， 博士

【摘要】 从电子计算机诞生伊始起,无数富有才干的艺术家们就孜孜不倦地尝试把计算机用作一种艺术创作的工具。但是早期的计算机非常昂贵,计算能力也十分有限。因此当时将计算机用于艺术创作的尝试困难重重。近年来,随着计算机硬件技术发展的突飞猛进,即便是普通个人电脑也拥有了比早期大型计算机更为强大的运算能力。这为艺术家们达成利用计算机进行艺术创作的宿愿提供了丰富的技术可能性。本学位论文对计算机艺术的研究就是在上述计算机硬件技术发展带来的技术可能性与诸多艺术工作者将计算机引入艺术创作领域的强烈希冀下萌生的。本学位论文的研究围绕着两种经典的中国传统艺术形式所展开——中国书法艺术和绘画艺术(以下简称中国书画艺术)。这一研究对象的选择是基于对目前计算机艺术研究现状的认识:尽管悠久的中国历史孕育了灿烂的中华文化与艺术,并且经过无数先人毕生的努力,中国书画艺术的独特魅力已为世界所广泛承认,但是在国际计算机科学界,针对中国书画艺术的计算机艺术的研究仍非常有限。此外,目前针对西方艺术形式所展开的计算机艺术的研究尚无法解决中国传统书画艺术创作电子化、信息化过程中所特有的许多富有挑战性的问题并为之提供有效的技术支持。鉴于上述原因,本学位论文选择了将中国书画艺术电子化创作的论题作为研究中心。本学位论文对中国书画艺术电子化创作中涉及到的若干智能设计与美学创作的重要问题以计算机科学研究的方式展开了一系列算法设计工作与软件工程的实践。我们对中国书画艺术电子化创作进行的算法性探索主要包括三个方面的研究工作:1)计算机智能书法生成的研究,包括计算机智能书法生成算法的研究及与之相应的面向计算机自动书法生成的智能计算机辅助设计系统的研制;2)国画分解及国画风格动画智能生成技术的研究,具体包括面向中国国画作品图像的笔划分解技术及与之配套的基于笔划的国画风格计算机动画智能生成制作技术的研究和相应软件原型系统的研制;3)交互式中国书画创作的研究,包括面向交互式电子书画创作的算法研究及与之相应的软件平台工具的研制,具体体现在对基于物理模拟的面向计算机交互式电子书画创作的虚拟画笔模型的算法设计和相应软件原型系统的研制以及对富有艺术表现力的高精度虚拟颜料行为模拟仿真模型的设计及配套算法技术的研究。下面将对上述三方面研究的主要工作作一个简要的介绍。在计算机智能书法生成的子课题研究中,我们探究了如何研制具有机器创造力的可以从事自动设计的智能计算机辅助设计系统的论题。这一研究的主要难点在于如何对人脑的创造力进行数学建模并基于该模型通过计算模拟的方式实现机器创造力。在该研究中,我们采用了综合推理的创作思维模型来模拟人脑在从事设计活动时展开创造性思维的过程。之所以选用综合推理作为我们计算模拟的算法设计基础,是因为该模型在前人的研究中已被证明是一种尤其适合于模拟人脑在形象思维领域从事创造性思维活动的心象计算模型。在本研究中,我们根据计算机书法生成的特定领域问题背景,基于综合推理模型的一般设计方法,提出了基于综合推理的计算机智能书法生成算法。在研究如何模拟书法家在进行书法创作时所开展的创造性思维的问题同时,我们还研究了智能书法生成这一问题所涉及到的知识表示与获取的问题。这是因为知识表示与获取的方法是否完备且利于知识重用会对我们的推理模型解决实际问题的能力起着至关重要的影响。本研究对汉字书法作品引入了可以实现组件重用的多层次参数化知识表示方法。这一知识表示方法是基于对汉字构成图像中存在的大量冗余、可重用图像特征及结构特征的观察认识所设计提出的。因此该知识表示方法可以非常有效地支持对关于汉字书法的图像信息的知识表达和重用。综合上述两部分对基于综合推理的计算机智能书法生成算法的研究以及对汉字书法信息知识表达与获取方法的研究,我们初步现实了在计算机书法这一特定领域问题内的形象设计思维中创造力的计算模拟,获得了计算机在书法创作活动中丰富的机器想象力与创新能力——计算机在学习了有限的输入字帖后,能够在一个巨大的创作可能性空间里,实时全自动地创作出风格迥异的计算机书法作品。由于知识的组合爆炸,这个计算机书法创作的可能性空间十分巨大,荧屏上常出现出乎意料之外的崭新的书写风格和别有韵味的书法作品。通过我们的方法生成的计算机书法作品书写效果变化多样、风格迥异,并且常常别有韵味,可以激发起书法爱好者对美学创作活动强烈的兴趣和无限的创作灵感。对于专业书法创作人士来说,通过观察计算机智能生成的书法作品,可以获得由于思维定势约束或缺乏创作灵感而无法发现的艺术创作动机与可能性,从而实现计算机智能辅助设计与创作。在基础研究的层面上,我们对计算机智能书法生成的研究以中国传统书法作品的智能生成作为实验背景,探索了计算机美术中创作思维模拟的问题。值得一指的是,我们对计算机智能书法生成的研究本身是以一个案例分析的形式所展开的,旨在探索在形象思维领域对人脑从事创造性思维活动进行机器模拟的可行性。这一研究的最终目的在于研制针对形象思维的先进智能计算机辅助设计系统和工具,用于辅助设计人员在从事富有挑战性的设计作业时拥有更为开阔的想像力和开展更为活跃的创造性思维活动。由于人脑本身受到的诸如思维定势等约束所导致的功能上的局限性,即便是最富有才华的设计师和最富有创新能力的思考者也会受益于上述的先进智能计算机辅助设计系统。在我们计算机智能书法生成的个案研究中,我们通过展示计算机自动书法生成系统的研制过程,详细示例了如何根据综合推理的一般方法,研制能够解决具体领域问题的智能计算机辅助设计系统。这一智能系统设计研发过程的成功个案示范,表明了研制上述先进智能计算机辅助设计系统的可能性,也预示着针对其他特定领域问题研制类似高端智能计算机辅助设计系统的技术可行性。这些系统将智能地辅助设计人员更富有创造力地完成设计工作,实现设计创新。就此,我们最后提出了开发基于综合推理的面向特定应用领域的智能计算机辅助设计系统的一般方法,用以作为研发类似系统的方法论指导。在国画分解及国画风格动画智能生成的子课题研究中,我们提出了一种新的基于静态国画作品生成国画风格动画的智能技术。为了生成一幅国画风格的动画,首先需要提取构成该静态画面的笔划。笔划提取过程的关键是利用一个样本笔划库进行笔划分解。一旦构成原画的笔划被计算提取出来之后,该绘画就可以在笔划级完成动画的编辑和生成工作。根据上述算法,我们成功地开发了一个基于笔划的国画风格动画编辑、创作环境。我们的实验结果包括对几幅国画作品笔划分解及动画生成的实例。这些实验结果均获得了满意的视觉效果,从而实证了本研究所提出的“基于笔划的国画分解与动画技术”的可行性与有效性。在交互式中国书画创作的子课题研究中,我们探索并设计了一个与用户友好交互、实时响应,并且具有丰富艺术表现力的全电子化的书画创作软件环境。对这一问题的研究是计算机图形学和计算机人机交互领域长期以来存在的热点与难点。这项研究不仅富有理论研究价值,而且有着广阔的市场应用前景,也对信息时代中国传统书画艺术的创新与电子化变革有着重要的意义。在技术层面,本研究提出了面向书画创作、基于实体造型技术的虚拟毛笔模型以及利用它进行交互式电子书画创作的计算模拟框架。我们还对影响该计算模拟框架综合性能的若干关键模块做了算法优化,旨在提升性能。根据上述模拟框架,我们研发了一个基于虚拟毛笔模型的支持电子书画交互式创作的软件原型系统。实验结果表明,用户可以利用虚拟毛笔创作出艺术表现力非常丰富的电子书画作品。为了进一步提高上述系统的艺术表现力,我们还提出了一种新颖的面向数字绘画——特别是中国水墨画和水彩画创作的虚拟颜料行为模拟仿真模型。该模型不但计算模拟了颜料水溶液的独立行为;还考虑了在颗粒级上颜料和画笔、画纸之间发生相互作用的物理化学过程,从而实现了对颜料颗粒在画笔笔毛上的扩散过程以及在画纸表面的吸附和沉降过程的精细模拟仿真。在算法的具体设计过程中,我们大量参考了表面化学和纺织工业中对吸附和扩散过程研究所得出的定性规律和定量方程。这样高精度的模拟有效地保证了我们为艺术家提供的电子书画创作工具的艺术表现力。例如我们新的算法可以同时模拟包括非常湿润和极其干枯的笔划在内的一系列绘画笔触效果,从而为数字绘画提供了广泛的可能性。这是现有的颜料模型所没有能够很好解决的问题。为了验证上述模型的有效性,我们将该模型的算法实现集成到我们在第一阶段研究中已经开发的基于虚拟画笔的电子书画创作系统中。实验结果表明,我们提出的这一新颖的虚拟颜料仿真模型可以有效地支持高质量东方绘画和水彩画的创作过程,从而肯定了该模型设计的成功性及其实用价值。总的来说,本学位论文的研究综合运用了智能计算机辅助设计技术、人工智能、计算机图形学与图形交互技术以及计算机人机交互技术等多种计算机科学与技术的手段,对中国书画艺术电子化创作的论题从多个视角展开了深入而细致的算法性探索。该研究成功地解决了中国书画艺术电子化创作中所特有的多个富有挑战性的学术难题,取得了许多算法研究的成果,并积累了丰富的研制相关算法性软件原型系统的工程实践经验。该学位沦文汇报了上述对中国书画艺术电子化创作开展的算法性探索工作与相应的软件工程实践所取得的主要研究成果。更多还原

【Abstract】 There is a long trace of efforts by talented artists to try unremittingly to transform the computer into a novel art creation tool, even at times when computers were outrageously expensive, clumsy and had very limited power. But until recently the limited amount of computing power available to any ordinary person had meant more hindrances than opportunities to the innovative artists who fancied using the computer to do art. This thesis study is formulated in the light of recent technical advances that are turning what has been largely a dream into the reality of the computer having a serious role in art. Our study concentrates on two traditional Chinese art forms—calligraphy and painting. This is because despite the fact that the very long Chinese history has nurtured an extremely rich culture and system of Chinese arts of calligraphy and painting, which are increasingly becoming more popular around the globe, very limited serious research efforts have been dedicated to the challenges of Chinese painting and calligraphy in computer science circles. Many of the research problems are quite unique and not present in Western arts. This motivates us to explore a computational approach to Chinese calligraphy and painting.In our research, we try to bridge the beauty and intelligence involved in the two traditional forms of Chinese visual arts with the algorithmic possibilities brought about by advanced modern computing techniques. There are three main components in our study: 1) automatic generation of Chinese calligraphy, 2) a stroke-based intelligent Chinese painting decomposition and animation method, and 3) an interactive approach for electronically creating Chinese calligraphy and painting using a physically-based virtual hairy paint brush model and an expressive digital pigment behavior simulation model. In the following, we give the highlights of each.In the first part of our research, intelligent Chinese calligraphy generation, we address the demanding task of developing intelligent systems with creativity abilities that can perform design tasks automatically. The main challenge is how to model human beings’ creativity mathematically and mimic such creativity computationally. At the heart of this research, we adopt the synthesis reasoning model as the underlying reasoning mechanism to simulate human beings’ creative thinking when they handle design tasks. We adopt this model because it is a mental simulation device grounded in cognitive sciences, and there is proof by prior studies that the model is suited for reproducing human’s innovative thinking in the imagery thinking domain. We carefully tailor the standard form of the general synthesis reasoning model according to our application context and propose a concrete analogous reasoning algorithm based on the general synthesis reasoning model for automatic generation of Chinese calligraphy. We also look into knowledge representation and acquisition issues in our automatic calligraphy generation problem as without the support of a comprehensive and efficient knowledge representation and reference mechanism, no reasoning model will be capable enough to handle real-world problems. We propose a hierarchical and parametric schema to represent Chinese calligraphic writings in the form of static images. Our design facilitates efficient knowledge representation and reference by observing repetitive image patterns and structures exhibited in the calligraphic writings. Based on the synthesis reasoning model and our calligraphy knowledge representation and acquisition schema, once a computer has learned a few calligraphic samples through acquiring the parametric and hierarchical representation of these sample calligraphic writings, it can intelligently and automatically generate Chinese calligraphic writings in novel writing styles. When knowledge is combined, the space in which computer can create visually interesting Chinese calligraphic artworks is immense. Given such a large space of potential solutions, the computer can generate Chinese calligraphic writings with writing styles that we have never seen before. In addition to generating novel artistic Chinese calligraphic writings for calligraphy amateurs, for those advanced calligraphic fans and practicing calligraphists, the computer generated artworks, writing styles, and visual effects can be a useful inspiration. Our system therefore functions as an intelligent computer-aided design system that can stimulate calligraphy art masters to explore new calligraphy art creation possibilities and opportunities which might otherwise be inaccessible due to limitations on the creativity of human brains such as habitual thinking and the lack of art creation motives or inspirations.Our study on automatic generation of Chinese calligraphic artwork spans a broad research context in which our work can be considered a case in point aiming at revealing the possibilities of simulating human being’s imagery thinking in creative problem solving. Our purpose is to emulate the human thinking process for the design and development of advanced intelligent computer aided design systems or tools. With these systems or tools, people can conduct challenging design tasks that call for a high degree of creativity and imagination which could be challenging even for the most gifted designers and original thinkers. We present the detailed procedure of designing and developing an intelligent computer aided design system based on the synthesis reasoning model. It demonstrates that the notion of using the computer to assist human designers on their non-trivial design tasks is technically feasible. This also implies the feasibility of developing similar intelligent computer aided design tools in other domain specific applications. In the light of such a vision, we propose a generic methodology for constructing intelligent systems using the synthesis reasoning model based on the design and implementation experiences of our intelligent calligraphy generation system as well as several other systems for solving real-world problems. We describe an extended version of the synthesis reasoning model and its associated software development methodology to serve as a reference for the design and development of synthesis reasoning based intelligent CAD systems for arbitrary domain specific applications.In the second part of our research, we propose a technique to animate a Chinese style painting given its image. What is particularly interesting is that the animation is done at the brush-stroke level. To do that, we first extract descriptions of the brush strokes that hypothetically produced the painting. The key to the extraction process is the use of a brush stroke library which aids region segmentation. The brush stroke library is obtained by digitizing single brush strokes drawn by an experienced artist. The steps in our extraction technique are first to segment the input image, then to find the best set of brush strokes that fit the regions, and finally to refine these strokes to account for the local appearance. We model a single brush stroke using its skeleton and contour, and we characterize texture variation within each stroke by sampling perpendicularly along its skeleton. Our brush stroke model plays a critical role in allowing the painting’s appearance to be captured and subsequently rendered with good fidelity. We also introduced a brush stroke overlap separation algorithm which allows full appearance of strokes to be extracted despite the presence of overlaps. Once these brush descriptions have been obtained, the painting can be animated at the brush stroke level. A key contribution of our work is the automatic recovery of separate, vectorized brush strokes. This is a tremendous time saver compared to manual segmentation especially when the painting has hundreds of brush strokes. In addition, proper automatic color separation in the overlap regions is not trivial and is not a feature in common image editing tools such as Photoshop. The animation is significantly easier once the segmentation is done. We present several animations of real paintings using our technique. In our study, we focus on Chinese paintings with relatively sparse strokes. The animation is produced using a graphical application we developed. Experimental results show that our method of decomposition is capable of producing high-quality reconstructions of paintings. The quality of the sample animations also serves to illustrate the effectiveness of our decomposition approach for producing stroke-based animation of Chinese paintings.The third part of our research is on the design and development of a novel interactive software system for Chinese calligraphy and painting. We aim at user friendliness and user expressiveness in the design. Many researchers in computer graphics and human-computer interaction circles for a long time have attempted to design such a system. Providing a digital environment for paper-less artwork creation is not only challenging in terms of algorithm design, but requires also significant system research skills, and much effort in constructing a functioning system satisfying user needs. This pursuit has promising market values and could lead to new software tools that may reshape the current digital art software industry. This thesis describes the following two main pieces of our work: 1) a novel algorithmic framework for interactive digital painting and calligraphy based on a new virtual hairy brush model, including an advanced design to optimize the overall performance, and 2) a novel digital pigment behavior simulation model to serve as a foundation for the creation of expressive digital painting and calligraphy.For our algorithmic framework supporting interactive digital painting and calligraphy, we propose a novel virtual hairy brush model for which solid modeling and parametric design techniques are employed. Our model meets all the criteria for a good digital paintbrush. We use only four attributes to capture the essential features of the brush, and a suitably powerful modeling metaphor for its behavior. The virtual hairy brush model’s geometry, dynamic motions, and pigment changes are all dealt with in a single model. A single model simplifies the synchronization between the various system modules, thus giving rise to a more stable system, and lower costs. By a careful tradeoff between the complexity of the model and computation efficiency, elaborate simulation of the virtual hairy brush model’s deformation and its recovery for interactive painterly rendering is made possible. We also propose a novel paper-ink model to complement the brush’s model, and a machine intelligence module to empower the user to easily create beautiful calligraphy and painting. We have successfully developed a prototype system with which end users can interactively produce high-quality digital paintings and calligraphic artwork. Despite the complexity of the modeling behind the scene, the high-level user interface has a simple and friendly design. The final results created by our virtual hairy brush model can achieve visually very compelling results.We also propose a novel generic pigment model suitable for digital painting in a wide range of genres including traditional Chinese painting and water-based painting. The model embodies a simulation of the pigment-water solution and its interaction with the brush and the paper at the level of pigment particles; such a level of details is needed for achieving highly aesthetic effects by the artist. The simulation covers pigment diffusion in the brush, sorption processes at the paper surface, and aspects of pigment particle deposition on the paper. We follow rules and formulations from quantitative studies of adsorption and diffusion processes in surface chemistry and the textile industry. The result is a pigment model that spans a continuum from the very wet to the very dry brush effects, and hence is capable of simulating many intricate effects in digital painting. To experiment with the proposed model, we have embedded the model in our virtual hairy brush based digital painting and calligraphy prototype system. The combined system exhibits interactive speeds on a modest PC platform with improved system performance in terms of the system’s artistic expressiveness, as reflected in the extended range of visual effects users can create using our system.In summary, this thesis studies an interesting problem in computer science—the application of computing to Chinese calligraphy and painting. Our approach represents a multi-disciplinary treatment involving intelligent computer aided design, artificial intelligence, computer graphics and interactive techniques, and human-computer interaction. We address many unique difficulties and challenges of using the computer to produce Chinese painting and calligraphy. This thesis presents both our research results and the lessons and engineering experiences obtained during our research process. As a whole, this thesis offers a fresh view on Chinese calligraphy and painting from a computational perspective.更多还原