CAREER: 3D Sketchpad

这项研究的目标是一套工具（称为“3D Sketchpad”），用于创建交互式3D插图。例子应用包括老师解释物理，考古学家重现古代村庄模型，高中生制作引人注目的动画，以及医生向病人解释诊断。关键的创新是通过绘图来支持3D建模。生成的内容看起来像用户的绘图，但它是3D的，因为可以交互式地移动视点。这项研究的长期影响是四倍的。首先，让普通人创建3D内容的工具将极大地扩展3D图形的应用。第二，专家用户会设计他们用现有工具无法完成的内容，因为这太困难或太耗时。第三，这些工具支持3D内容，在美学上比目前可能的更接近传统插图。最后，该技术将与基于笔的计算的增长协同传播。用于创建3D绘图的工具最终将在平板电脑、掌上设备甚至手机上普及。一个关键的挑战是开发新的形状表示和相关的操作，以支持基于绘图的界面。研究人员正在开发一种新的表示，称为“分层网格”，由分层逐步建立的多分辨率网格组成。每个新层都动态地响应前一层的变化。分层网格支持在任何分辨率下添加新的结构（例如人体的四肢）和表面细节（凸起或凸起）。它们产生任意拓扑的分段光滑流形曲面。其他挑战包括开发新的非真实感渲染算法，基于手势输入的新交互技术，通过草图输入生成动画的新方法，以及可以从提供的示例生成新的形状特征，笔画模式和运动的模式合成工具。

This research targets a suite of tools (called "3D Sketchpad")for creating interactive 3D illustrations. Example applicationsinclude a teacher explaining physics, an archaeologist recreatinga model of an ancient village, a high school student producing aneye-catching animation, and a physician explaining a diagnosis toa patient. The key innovation is to support 3D modeling bydrawing. The resulting content looks like the user's drawing, butis 3D in the sense that the viewpoint can be moved interactively.The long-range impact of this research is fourfold. First, toolsthat let ordinary people create 3D content will dramaticallyexpand the applications of 3D graphics. Second, expert users willdesign content that they would not have undertaken with currenttools because it would be too difficult or time-consuming. Third,these tools support 3D content that is aesthetically closer totraditional illustration than is currently possible. Finally,this technology will propagate synergistically with the growth ofpen-based computing. Tools for creating 3D drawings willultimately be commonplace on tablet PCs, palm devices, and evencell phones.A key challenge is to develop new shape representations andassociated operations that support a drawing-based interface. Theresearchers are developing a new representation, called a"layered mesh", consisting of a multiresolution mesh that isbuilt up incrementally in layers. Each new layer respondsdynamically to changes in the previous layers. Layered meshessupport the addition of new structures (e.g. limbs of a humanfigure) and surface details (bumps or bulges) at anyresolution. They produce piecewise-smooth manifold surfaces ofarbitrary topology. Additional challenges include thedevelopment of new non-photorealistic rendering algorithms, newinteraction techniques based on gestural input, new methods forgenerating animation via sketched input, and pattern synthesistools whereby new shape features, stroke patterns, and motionscan all be generated from provided examples.