Broad-Based Computational Shape Design

点击翻译按钮获取中文摘要

Digital 3D models are ubiquitously used to communicate product designs, describe virtual assets in digital media, and specify target geometries for advanced manufacturing tools. Millions of designers and innovators who seek to utilize these technologies require the means to convert their imagined shapes into customized 3D digital models. Currently available 3D modeling tools are not suitable for their needs since they require time and specialized modeling expertise, and are often optimized for narrow sets of geometries. I will enable 3D geometric content creation by this emerging broad user base by developing modeling approaches that are easy to use yet support modeling of the broad range of shapes these users seek to create. I will achieve this goal by enabling users to describe their conceived shapes using the medium they are most comfortable with and then algorithmically translating the provided user input into the user-intended virtual 3D models. The algorithms I will develop will leverage a new line of modeling research that I helped pioneer that utilized insights about human perception and communication of shapes for algorithmic user-driven shape processing and modeling. My program will take advantage of emerging interface technologies for shape communication that are only now becoming broadly available, and will leverage user studies and collections of 3D data to both validate and fine-tune the proposed methods. 3DInterfaces: To support the modeling needs of a broad user base, I will develop a palette of modeling methods customized to different user-sets and modeling domains, while leveraging common observations about communication of user-intent as an underlying design principle. In the short term I will focus on modeling frameworks that employ three types of user-facing interfaces for shape communication: touch-and-pen displays, 3D Virtual and Augmented Reality manipulation devices, and photometric scanning of user-created real-life mockups. These technologies enable more natural 3D content communication compared to traditional mouse-and-keyboard modeling systems. However, leveraging the additional power they provide for user-guided modeling requires new algorithms and methodologies. 3DOptimization: Amateur users frequently lack the artistic skills to fully communicate the shapes they imagine. Thus inputs generated by less skilled users may lack the aesthetics or functionality they desire. My research will optimize the quality of the output models produced from such inputs while ensuring that the results remain consistent with user intent. Combined together these methods will drastically simplify the creation of virtual models, facilitating content creation by a broad user base and across a wide spectrum of domains. They will greatly expand the accessibility of advanced manufacturing techniques, a research priority identified by the federal government.