计算机辅助三维创意建模技术

发布时间：2018-01-23 20:20

本文关键词： 计算机辅助三维创意建模虚拟生物建模生物语法实时分割基于草图的建模三维制造蒙皮多姿态平衡分析　出处：《浙江大学》2016年博士论文　论文类型：学位论文

【摘要】：在三维影视特效及电子游戏中,三维角色模型是必不可少的部分。虽然现有大量的商业建模系统,可以满足各种建模需求,但这些系统非常复杂,难以掌握,更多地局限于专业人员使用。过去十几年里,图形学界在简单易用且功能强大的建模技术方面进行了大量研究,提出了草图式建模、组合式建模等方法,取得了重大进展。总体来看,现有的建模技术更多地将三维建模视为工艺过程,即在建模之前有明确的目标。然而,三维建模也是个开放式的艺术创作过程,其建模目标会经常发生不可预知的改变。显然,现有的建模技术不适用于开放式建模。因此,如何在开放式建模过程中提供创意支持以辅助用户更好地创作在三维建模中具有重要的理论研究意义与实际应用价值。其中一类计算机辅助三维创意建模技术向用户建议可添加到当前模型的部件。然而,这类技术都依赖预分割(或标记)的模型数据库。用户仅可使用预分割的部件组合模型。这极大地限制了建模空间与用户的创造力。另一类是模型建议技术。这类技术更适合处理具有相同结构的同类模型,产生的是与输入具有相同结构的同类模型。其最大缺点是不适合产生拓扑结构变异的模型。此外,在计算机动画中,建模仅是第一步,接下来还要编辑动画甚至三维打印制造实物。编辑动画之前需要蒙皮。三维打印之前需要经过很多处理,比如,平衡性分析,关节优化等。然而,现有的计算机辅助三维创意建模系统往往局限于建模。简单地将蒙皮与面向三维打印的模型处理做为建模的后续处理步骤无法达到实时交互。因为,"建模-蒙皮-面向三维打印的模型处理"是个迭代的步骤,简单组合现有技术会导致用户在不同工具间频繁变换,降低效率。本文针对现有计算机辅助三维创意建模技术不适合产生拓扑结构变异,依赖预分割的模型数据库限制建模空间,缺少蒙皮及面向三维打印的模型处理三方面的问题,进行了深入研究与探索,提出了一系列新技术新方法,在实际应用与理论研究上取得重大突破。本文主要研究内容如下:·提出一种辅助用户快速建模大量虚拟生物的技术。这些虚拟生物具有合理的奇异的拓扑结构。用户从数据库内选择若干个不同类的常规模型,我们的技术迭代地产生一代代结构奇异的虚拟生物模型做为建议提示给用户,以激发灵感。我们方法的核心是生物语法(Creature grammar)。生物语法是针对建模虚拟生物特化的形状语法(Shape grammar)。生物语法改变部件的排列方式与数目,同时保持生物的语义约束。·提出一种基于草图的按需部件建议方法—SketCh2Cut系统。用户输入草图表达设计概念,我们的方法依此生成三维代理部件并从数据库内搜索出一系列候选部件,提示给用户。我们的方法首先将三维模型局部匹配问题转化为多视角下的二维轮廓局部匹配。然后基于三维模型的超面片图表示,快速地从三维模型上提取与用户意图匹配的三维部件,提取过程由粗到细,边界逐步优化。与以前的工作不同,我们的方法不依赖预分割的三维模型数据库,并可以建议定制的任意部件,极大地扩展了用户设计空间。·提出一个统一的框架,将建模、蒙皮与面向三维打印的模型处理无缝地融合到一起。输入一组蒙皮的三维模型,我们的系统迭代地产生一代代新模型,做为建议提示给用户以激发灵感。用户选择感兴趣的模型用于动画编辑与三维打印分析。我们的方法首先融合部件表面及其骨架,然后快速更新融合模型的蒙皮权重,接着采用面向三维打印的骨架优化算法选择出最优关节位置与半径,最后优化模型内部质量分布,确保模型在不同姿势下保持平衡。通过将建模、蒙皮与面向三维打印的模型处理集成至统一的框架,我们的方法更好地辅助了用户的创新性建模过程。
[Abstract]:In 3D video effects and computer games, 3D role model is an essential part. Although a large number of existing commercial modeling system, can meet the modeling requirements, but these systems are very complex, difficult to grasp, more limited to professional use. Over the past ten years, made a lot of research in the field of graphics easy to use and powerful modeling techniques, presents a sketch of Shi Jianmo, combined modeling method, has made significant progress. Overall, the modeling technology of 3D modeling more as a process, which has clear objectives before modeling. However, 3D modeling is an open-ended process of artistic creation, its modeling the target is often unpredictable change. Obviously, the modeling technology is not suitable for open modeling. Therefore, how to provide creative support in open modeling process In order to assist users to better the creation of research has important theoretical significance and practical value in 3D modeling. One kind of computer aided 3D modeling technology to the user creative suggestions added to the current model of the part. However, this kind of technology is dependent on the pre segmentation (or mark) of the model database. Users can only model components the use of pre segmentation. This greatly limits the modeling space and user creativity. Another is the model suggested technology. This kind of technology is more suitable for the treatment of similar model with the same structure, the same model has the same structure and input. Its biggest disadvantage is not suitable for generating topological variation in addition model. In computer animation, modeling, is only the first step, then also edit animation and even 3D printing manufacturing objects. Editing animation is needed before skin. 3D printing is needed before the A lot of processing, for example, balance analysis, joint optimization. However, 3D creative modeling system is often limited to modeling the existing computer. Simply the skin and oriented 3D printing processing model for subsequent processing steps of modeling can not achieve real-time interaction. For modeling - skin - oriented 3D printing model processing is an iterative procedure, a simple combination of existing technology will lead the user to reduce efficiency between different tools. Frequent transform, aiming at the existing computer aided 3D modeling technology is not suitable to produce creative topology variation, dependent model database pre segmentation modeling space, lack of skin and oriented 3D printing model three the problem, in-depth research and exploration, put forward a series of new methods and new technologies, has made a major breakthrough in the practical application and theoretical research in this paper. The research contents are as follows: - provide a user assisted rapid modeling of a large number of virtual biological technology. The virtual life with strange topology structure reasonable. Users choose several different from the conventional model database, as suggestions to prompt the user to generate iterative virtual biology model generation structure of strange US. To inspire. The core of our approach is the biological Grammar (Creature grammar). The biological grammar is the shape grammar modeling virtual biological specialization (Shape grammar). The biological syntax changes the arrangement and the number of components, while maintaining the semantic constraints of organisms. And presents a sketch of the required components suggest method - SketCh2Cut system based on user input sketch expression design concept, and our method of generating 3D agent components and search out a series of candidate parts from the database, to prompt the use Our method first. The 3D model partial matching problem is transformed into the 2D contour under multi angle and local matching. Then hyperpatch diagram based on 3D model representation, quickly from the three-dimensional model of 3D intention extraction components matching with the user, the extraction process from coarse to fine, and gradually optimize the boundary. Unlike the previous work our method does not rely on the database, 3D model segmentation, and can be customized components suggest, greatly expand the user design space. We propose a unified framework for modeling, covering and oriented 3D printing model seamlessly together. A 3D model group covering input system iteration a new model of our real estate from generation to generation, as suggestions to prompt the user to inspire. Interest model is used to analyze the animation editing and 3D print users. First we method The fusion component surface and skeleton, and then quickly update the skin weights fusion model, then used for 3D printing frame optimization algorithm to select the optimal joint position and radius, finally optimize the quality distribution inside the model to ensure the model balance in different positions. By modeling, the skin and the oriented 3D printing processing integration to unified model the framework, our method better aided innovative user modeling process.

【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.41

【相似文献】