遗传算法及其在3D动漫造型设计中的应用研究
发布时间：2018-08-16 08:00
【摘要】：动漫产业有着广阔的发展前景,被称为21世纪创意经济中最有希望的朝阳产业和21世纪的“无烟工业”。动漫造型作为动漫中的灵魂,成为欧美日韩等动漫强国推销其动漫与衍生产品的重要武器。动漫造型一般包括动漫角色造型和动漫群体造型。从深受观众喜爱的唐老鸭、米老鼠到现今的机器人瓦力、功夫熊猫,每一部动画片都能诞生一个甚至多个明星,成功地设计一个动漫角色造型,不仅能提升一部动漫影片的票房号召力,更能成为推广整个产业的中坚力量;此外,不少动漫影片中还会涉及到大量的群体造型,例如,大型战争场面中大量的战士、角斗场中大量的观众、城市中大量的群众等。有效地设计、产生动漫中的角色造型及群体造型,对于提升动漫产品的质量至关重要,一直是动漫领域的研究热点之一。 3D动漫是当今动漫产业的发展方向,3D动漫造型作为3D动漫的核心元素,在3D动漫制作中具有非常重要的地位,其好坏将直接影响3D动漫的质量以及后期的制作效果。如何高效地设计一个能够吸引大众眼球的3D动漫角色造型、如何快速高效地生成动漫影片中数以万计的3D动漫群体造型,一直是3D动漫设计领域重要的研究问题。3D动漫造型的传统设计方法是利用三维建模软件Maya,3D Max等直接绘制,这种方法获得的3D动漫造型美观、精确。但是,对于设计一个3D动漫角色造型而言,设计师往往会受到现有造型和设计观念的影响,导致很难设计出具有足够创意的、能够吸引大众眼球的作品;而对于设计数以万计的3D动漫群体造型而言,由于其造型数量太多、设计的工作量巨大,使用传统的设计方法成本太高、速度太慢、且无法自动生成大量的造型。因此,迫切需要寻找合适的方法来解决这些问题。 本文通过研究并利用进化计算技术、协同技术和可视化技术,以设计支持3D动漫造型协同创新设计的方法、提高造型的新颖程度和生成效率为目标开展研究,本文的主要工作与创新点概括如下: 1.为了将单目标和多目标遗传算法较好地应用到3D动漫造型协同创新设计中,首先将自适应遗传算法(AGA)和带精英策略的非劣排序遗传算法(NSGA2)进行了改进,有效地解决了AGA算法收敛速度慢和进化初期容易停滞的问题,并使改进的AGA和NSGA2算法可以很好地应用到创新设计中。 2.为了能够同时利用多位专家已有的知识对3D动漫造型进行设计,使用HOOPS/NET技术搭建了一个支持3D动漫造型创新设计的协同设计环境,并在此基础上,提出一种基于HSF(HOOPS Stream File)的多专家协同评价方法,有效地提高了设计效率。 3.基于以上提出的两种改进的遗传算法和多专家协同评价方法,提出了两种支持3D动漫造型协同创新设计的方法,有效地解决了传统的3D动漫造型设计方法难以设计出具有足够创意的3D动漫角色造型和数以万计的3D动漫群体造型的问题。 4.基于以上研究成果,设计并开发了一个3D动漫造型协同创新设计原型系统。通过在原型系统中的实验结果表明,本文提出的方法既可以高效地设计出创意新颖的3D动漫角色造型,又可以快速高效地生成大批的3D动漫群体造型。
[Abstract]:The animation industry has broad prospects for development, known as the most promising sunrise industry in the 21st century creative economy and the 21st century "smoke-free industry". As the soul of animation, animation modeling has become an important weapon for animation powers such as Europe, America, Japan and Korea to promote their animation and derivatives. From the popular Donald Duck, Mickey Mouse to today's robot Wally, Kung Fu Panda, each animation can produce one or more stars, successfully designed an animation character modeling, not only can enhance the appeal of an animation film box office, but also become the backbone of the promotion of the entire industry; Many animation films will also involve a large number of group modeling, for example, a large number of warriors in large-scale war scenes, a large number of spectators in the arena, a large number of people in the city and so on. Effective design, animation in the role of modeling and group modeling, to improve the quality of animation products is crucial, has been the animation field of research hot. One of the points.
3D animation is the development direction of today's animation industry, 3D animation modeling as the core element of 3D animation, has a very important position in the production of 3D animation, its quality will directly affect the quality of 3D animation and later production effect. Generating tens of thousands of 3D animation group modeling in animation movies has always been an important research issue in the field of 3D animation design. The traditional design method of 3D animation modeling is to use 3D modeling software Maya, 3D Max and other direct rendering, this method to obtain 3D animation modeling beautiful, accurate. However, for the design of a 3D animation character modeling. In other words, designers are often influenced by the existing modeling and design concepts, which makes it difficult to design works with enough creativity to attract the public's attention; and for the design of tens of thousands of 3D animation group modeling, because of its large number of modeling, design workload is huge, the cost of using traditional design methods is too high, speed. It is too slow to generate a large number of shapes automatically. Therefore, it is urgent to find a suitable way to solve these problems.
In this paper, through the research and use of evolutionary computing technology, collaborative technology and visualization technology, to design support 3D animation modeling collaborative innovative design methods, improve the novelty and productivity of modeling research objectives, this paper's main work and innovation are summarized as follows:
1. In order to apply single-objective and multi-objective genetic algorithm to 3D animation modeling collaborative innovation design, adaptive genetic algorithm (AGA) and non-inferior sorting genetic algorithm (NSGA2) with elite strategy were improved firstly, which effectively solved the problems of slow convergence speed and easy stagnation in the initial stage of evolution of AGA algorithm, and made the improved AGA. And NSGA2 algorithm can be well applied to innovative design.
2. In order to make use of the existing knowledge of many experts to design 3D animation modeling at the same time, a collaborative design environment supporting 3D animation modeling innovation design is built by using HOOPS/NET technology. Based on this, a multi-expert collaborative evaluation method based on HSF (HOOPS Stream File) is proposed, which effectively improves the design efficiency.
3. Based on the two improved genetic algorithms and multi-expert collaborative evaluation methods proposed above, two methods supporting collaborative and innovative design of 3D animation modeling are proposed, which effectively solve the problem that traditional 3D animation modeling design methods are difficult to design 3D animation character modeling with enough creativity and tens of thousands of 3D animation group modeling. Question.
4. Based on the above research results, a prototype system for collaborative and innovative 3D animation modeling design is designed and developed. The experimental results show that the proposed method can not only efficiently design novel 3D animation character modeling, but also quickly and efficiently generate a large number of 3D animation group modeling.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TP18;TP391.41
本文编号：2185388
[Abstract]:The animation industry has broad prospects for development, known as the most promising sunrise industry in the 21st century creative economy and the 21st century "smoke-free industry". As the soul of animation, animation modeling has become an important weapon for animation powers such as Europe, America, Japan and Korea to promote their animation and derivatives. From the popular Donald Duck, Mickey Mouse to today's robot Wally, Kung Fu Panda, each animation can produce one or more stars, successfully designed an animation character modeling, not only can enhance the appeal of an animation film box office, but also become the backbone of the promotion of the entire industry; Many animation films will also involve a large number of group modeling, for example, a large number of warriors in large-scale war scenes, a large number of spectators in the arena, a large number of people in the city and so on. Effective design, animation in the role of modeling and group modeling, to improve the quality of animation products is crucial, has been the animation field of research hot. One of the points.
3D animation is the development direction of today's animation industry, 3D animation modeling as the core element of 3D animation, has a very important position in the production of 3D animation, its quality will directly affect the quality of 3D animation and later production effect. Generating tens of thousands of 3D animation group modeling in animation movies has always been an important research issue in the field of 3D animation design. The traditional design method of 3D animation modeling is to use 3D modeling software Maya, 3D Max and other direct rendering, this method to obtain 3D animation modeling beautiful, accurate. However, for the design of a 3D animation character modeling. In other words, designers are often influenced by the existing modeling and design concepts, which makes it difficult to design works with enough creativity to attract the public's attention; and for the design of tens of thousands of 3D animation group modeling, because of its large number of modeling, design workload is huge, the cost of using traditional design methods is too high, speed. It is too slow to generate a large number of shapes automatically. Therefore, it is urgent to find a suitable way to solve these problems.
In this paper, through the research and use of evolutionary computing technology, collaborative technology and visualization technology, to design support 3D animation modeling collaborative innovative design methods, improve the novelty and productivity of modeling research objectives, this paper's main work and innovation are summarized as follows:
1. In order to apply single-objective and multi-objective genetic algorithm to 3D animation modeling collaborative innovation design, adaptive genetic algorithm (AGA) and non-inferior sorting genetic algorithm (NSGA2) with elite strategy were improved firstly, which effectively solved the problems of slow convergence speed and easy stagnation in the initial stage of evolution of AGA algorithm, and made the improved AGA. And NSGA2 algorithm can be well applied to innovative design.
2. In order to make use of the existing knowledge of many experts to design 3D animation modeling at the same time, a collaborative design environment supporting 3D animation modeling innovation design is built by using HOOPS/NET technology. Based on this, a multi-expert collaborative evaluation method based on HSF (HOOPS Stream File) is proposed, which effectively improves the design efficiency.
3. Based on the two improved genetic algorithms and multi-expert collaborative evaluation methods proposed above, two methods supporting collaborative and innovative design of 3D animation modeling are proposed, which effectively solve the problem that traditional 3D animation modeling design methods are difficult to design 3D animation character modeling with enough creativity and tens of thousands of 3D animation group modeling. Question.
4. Based on the above research results, a prototype system for collaborative and innovative 3D animation modeling design is designed and developed. The experimental results show that the proposed method can not only efficiently design novel 3D animation character modeling, but also quickly and efficiently generate a large number of 3D animation group modeling.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TP18;TP391.41
【引证文献】
相关硕士学位论文 前1条
1 孙玉灵;群体动画虚拟环境建模与路径规划方法研究[D];山东师范大学;2012年
，本文编号：2185388
本文链接：https://www.wllwen.com/wenyilunwen/dongmansheji/2185388.html
