外文参考文献译文及原文一种设计三维形状的工具.doc

目 录1 引言12 目前做法23 相关工作34 游艇的绘制55 特点和数据结构86 建立曲线96.1 自由曲线96.2 约束曲线96.3 反射曲线10结论11Contents1 Introduction122 Current practice133 Related work144 Sketch of a 12-meter yacht165 Features and data structures206 Curve creation226.1 Free-form curves226.2 Constrained curves226.3 Reflected curves23Conclusions2424一种设计三维形状的工具1 引言形状设计通过产品差异化，视觉吸引力，符合人体工程学的表现有助于产品的商业竞争力。遗憾的是，目前的CAD工具是无法达到外形设计很好，因为它们不能掌握快速素描的概念。因此，设计人员更喜欢使用纸和铅笔进行初步构思。我们的目标是建立一个从根本上形成的CAD设计系统的新类型。我们希望它足够直观，易于使用，功能强大以致工业设计人员可以将概念设计权的初始阶段通过适当的模式制造它。我们根据我们的方法来塑造一个新的范例，我们可以称之为“在三维中直接设计。”我们尤其要提供一个对用户友好的CAD界面，允许用户通过手持的六位自由传感器直接输入三维空间信息来设计复杂自由曲面的形状。这样的CAD系统将大大提高速度和形状设计质量。2 目前做法在产品设计的最初阶段，工业设计师往往造出粗糙的2D草图，只有最终产品的近似实际尺寸。一旦设计师一直致力于某一具体方案，草图将转化为更精确的三维模型。这个过程可能涉及到将粘贴版面编排或输入数据进计算机数据库。在这两种情况下，从二维草图到三维模型的转变是主观的和临时性的。该任务需要提取的素描（如剪影线，突出线条，线条和字符）及标注他们从而提供一个三维模型。我们可以描绘出他们使用这种“二维到三维”模式的复杂性和目前的CAD系统限制的事实，其中二维形状的输入建立了三维。用户必须通过使用二维输入设备如鼠标，操纵杆，或牌进行三维信息沟通。除了运用二维到三维范式，今天的CAD系统的大部分人也需要一个非常详细的基本形状的代表性。因此，用户必须有计划地和有条理地使用今天的三维CAD系统去设计模型，涉及指定层，局部坐标系统，除了工作架的坐标必须与样条拟合。有些系统允许设计师绘制平面（二维），然后将与样条曲线拟合。要获得三维曲线，然而，设计者必须通过指定移动的方向和幅度个别控制点外的平面位移。这些系统都需要用户在控制工作点的水平工作，而不是本身的形状。因此，用户被迫花费更多的时间和地点如何规定的比例，以完成一项任务，而不是任务本身。这总是一个“非直线前进”的系统和复杂的用户界面。3 相关工作几个研究工作重点是开发设计CAD系统的初始阶段。彭特兰开发这个叫Supersketch的CAD工具，在此前提下，人们自然地把看到的物体看为“块状粘土。”Supersketch的用户发展的原始形状和变形加入他们形成对象模型。在另一项研究，彭特兰还制定了素描工具，提取二维透视图的三维特征。编辑可以初步绘制曲线之后。计算机接口最近的发展之一是三维输入设备。几个早期的设计纳入附加的机械测量装置之间的联系。在另一种方法，费尔德曼提供力反馈以及位置和方向感应。他的Joystring包括一个手持3英寸的T型杆连接到伺服电机的电线杆和一台超级计算机控制。这种机械检测工具提供的费用和罚款使得它是尴尬的和限制自由运动的。所谓的3D输入设备传感器框架是在一个CRT的附近感觉手指的位置和方向。使用光学传感器阵列检测手部动作，这个装置可以用来“塑造”在屏幕上的部分。最近的另一个发展是一个6位的自由度球，目前可从多个供应商获得利用。这个工具，它类似于换档手柄上的杠杆，使用户能够在虚拟世界操纵的对象。推，拉，扭力和旋转，转换成翻译。具有3D输入设备的研究工作有重点关注波默斯三维六位的自由度设备。这种输入设备采用磁场，消除之间的手持传感器和计算机机械联系的需要。在一个项目中，沃尔和杰瑟姆创造了6个自由度的自由度鼠标命名为“蝙蝠”来探讨三维场景操控。他们发现，最有效的办法来克服深度感觉模糊之处是一个关于切换垂直轴90度翻转，同时禁用到屏幕的动作。他们还发现，与使用传统的鼠标接口相比，近似安置对象一个简单的任务，他们把这个结果记为手与物体运动之间的动觉对应关系。该波默斯设备也被用于在图形工作站探索的空间之间的输入和立体显示的对应关系。施曼特实施单传感器波默斯6D“魔术棒”。他的目标是发现一个清晰投影到空间，让用户在投影图像达到立体影像与棒的立体显示的互动功能。他首先制定了一个方案，创建三维涂料涂在棒附近地区。项目到工作区的图像，施曼特用半镀银镜和传统视频监控安装以上，以45度角在用户面前。用户戴眼镜，看从镜上CRT反映的时间复用立体影像。施曼特的初步结果显示，棒子和投影图像的良好的自然对应产生于棒尖所画的线。然而，渲染和图形之间的手部动作的检测和时间间隔所造成的磁场干扰有损于该工具的互动感觉。该波默斯设备也被用于Eyephone和数据手套产品，让用户在虚拟世界内运作的VPL之研究。该Eyephone是一个头戴式显示屏，能够提供广泛的周边立体虚拟世界的视觉。通过采用一个波默斯传感器跟踪头部运动，Eyephone使计算机匹配观看者在虚拟世界中的位置和用户的头部方向。总之，这两个装置允许用户通过移动一个虚拟世界，通过互手套的手部动作与对象交互。最后，VPL手套已被用在探讨立体显示三维曲面造型互动。德罗斯的3D设计空间使用户能够使用手套和移动三维点，形成了复杂曲面多项式的“控制网”。4 游艇的绘制在试图掌握3-Draw特性的过程中，我们开始了一个12米长的游艇样品模型的构建。下面的文本描述的步骤，设计人员使用3-Draw目前实现的功能完成了该模型。设计师的过程伴随着数字记录。首先，设计人员首次提请两个游艇模型曲线确定右舷甲板和龙骨轮廓（见图4.1）。他画了一个作为一个独立的空间三维曲线的单手运动的甲板曲线。接着，他使用Pick-Draw画出龙骨轮廓。若要使用Pick-Draw，设计师在右舷甲板预选两点（例如，一点在船头，另一点在船尾）作为新曲线的端点。预先选择结束点使得设计师可以集中捕捉曲线的形状，而不用在绘制过程中担心其大小，位置和方向。之后，并自动绘制曲线折点，以适应结束点，设计师通过扭曲手持笔指定新的曲线方向。然后，按下触控笔的按钮，设计师在所需的方向释放曲线。在创建游艇下一步，设计师确定一个反射面来建立了船中心线。在3-Draw的使用下，默认反射面可以“拿起”手写笔，然后以手写笔的自然动进行导向和定位。在释放时（松开手写笔按钮），相对于其他对象组成的曲线，反射面仍处于相对固定的位置和方向。然后，用户通过使用手写笔创建曲线的镜像版本。图4.1反映了甲板和龙骨曲线建立在船的左舷。图4.1 展示甲板和龙骨曲线建立船体形状，设计师下一步使用Pick-Draw，通过利用两个预选点绘制右舷舱壁（见图4.2）。在捕捉每个舱壁的形状之后，设计人员通过扭曲手写笔设置加入舱壁的轴的端点的扭量。然后，他按下手写按钮以画下曲线。（图4.2还说明了一个前舱壁被扭曲到正确位置。）此外，设计人员可以通过使用3-Draw的轴扭功能来改变任何曲线的轴旋转量。在完成舱壁反应整个中心平面轮廓之后，设计师像画龙骨和甲板那样，使用两点绘制出一个舵。图4.2 展示舱壁，注意使用笔完成舱壁曲线的扭转图4.3 展示舵，龙骨和甲板的梁接下来，设计师开始勾画翼龙骨。不是预选两个端点，他绘制一个独立的曲线作为右舷剖面，使得他稍后可以使用转换和旋转功能来移动它的位置。设计者在空间勾画出一条自由曲线作为翼龙骨。他用镊子和位移枪的功能来微调曲线的形状。镊子让设计者能够使用虚拟笔抓住一点，并能在手写笔的运动方向拉伸曲线。当位移枪使用时，3-Draw从该虚拟手写笔尖端呈现半透明圆锥辐射。当设计师横扫这个圆锥曲线部分，偏转的曲线距离正如从空气枪偏转一个字符串的浮动空间。类似镊子，位移枪个别点的坐标偏移，以使曲线能理想变形。一旦变形，曲线保持其新的形状。在加上翼龙骨的三维轮廓之前，设计人员使用擦除功能削减龙骨基底的一条曲线。然后，他使用保险丝功能挑选三条曲线联系在一起从而在船体上附上新的曲线。一旦曲线融合成单一的曲线，它们可以轻松地移动，并反映在一个步骤。设计者使用两点绘制法增加了一些弯曲过梁来获得图4.4的模型。图4.4 展示翼龙骨的三位轮廓该游艇模型再加上桅杆，索具，帆，和国旗就能完成。设计者使用单一的预选点，通过Pick-Draw绘制桅杆和旗杆。这使得它在形状确定后在规模，方向，形状和位置确定前很容易锚一曲线的结束。随着桅杆，这个绘图功能，让设计人员很容易保持和旋转游艇，在锁定曲线到位前同时移动桅杆以获得正确的向前或向后倾斜。锁定桅杆到位后，设计师使用功能Ball和Socket抓住和精细定位桅杆，使得它好像是通过一个球窝关节连接在其基底。5 特点和数据结构3-Draw保持一个三维虚拟世界类似于设计师的工作领域。当设计师在源附近的移动波默斯传感器，这个动作就被映射到虚拟物体在虚拟世界中的动作。这致使手部动作和虚拟物体移动产生一一对应。从波默斯设备发送得来的点坐标立即抵消一个固定转译。这抵消将绘制波默斯传感器的位置和方向，成为一个描述世界坐标的系统。传入的信息流也被缩放，使得一个完整的三维手部运动的范围反映在屏幕上的虚拟物体运动。该系统目前可显示一枝笔，类似于波默斯手写笔，以及产品组合的形状相对应的对象传感器。在一种情况中，对象传感器贴在一个类似剪贴板的调色板的顶部。由于易于操作，它为绘制一个新模型的前几条三维曲线作为了一个粗略的指导。该虚拟类似物对于调色板来说是一个网格线，大小类似调色板。通常，用户为了实现自己手中持有的曲线组成对象的感觉，会在绘制足够数量的曲线后关闭虚拟片。参考图5，曲线是存储在3-Draw中关于它们自己的局部坐标系CSline。此外，构成模型的每一组曲线的集合都有一个局部的坐标系统CScurve-list，正如以及协调系统的虚拟调色板CSobject和虚拟笔CSpen. 巢层次的坐标系统。因此，在一个曲线显示之前，它的点坐标转换首先就说明该模型，然后就说明了虚拟调色板，最后一个关于虚拟世界CSworld。3-Draw作为第一度连续基数样条呈现曲线。曲线插值点从波默斯设备获得。图5.1 坐标嵌套系统的树结构6 建立曲线3-Draw允许设计人员绘制三种曲线：自由曲线，约束曲线和反射曲线。6.1 自由曲线3-Draw在目前的实现的主要任务是绘制三维曲线和在虚拟工作区中相对于其他曲线建立自己的位置和方向。在这种运作绘图模式，用户通过按住手写笔的按钮按他所需的尺寸移动他的手来画一条曲线。松开按钮曲线就画出来了。在曲线的绘制过程中，3-Draw 转换直角点的坐标并对从笔传感器来的欧拉角到虚拟对象进行描述。然后计算和存储新坐标点的位置。起初，3-Draw 为虚拟对象和新的曲线指定坐标系统是为了与虚拟片制度同步和协调。然而，后来由于通过刚体变换曲线和对象被移动，该虚拟片的坐标系统与移动坐标系统进行对齐。在一个自由形式曲线图，3-Draw不断更新虚拟笔和片剂的位置。因此，设计者可能在获得多角度视觉的同时移动了牌。此外，一条曲线的增长完全取决于触笔的移动。因此，移动牌本身不会造成绘图。添加最后一个增量点在曲线上确定曲线上增设一个点的位置。这增量的变化是最后一次波默斯读取到现在手写笔的位置变化，是关于对虚拟对象的描述。6.2 约束曲线当创建曲线，设计者可以选择充分利用绘画的六位自由手写笔完全不受约束的规定，制定自由形式曲线。对于它们的位置，方向，在虚拟工作区的大小，这种自由形式的曲线，确实具有7个自由度：曲线上一点的坐标X、Y、Z的位置，曲线上这点的三维定向，和曲线的缩放。（当然，曲线的形状是手持式手写笔直接作用的结果。因此，在形状域所有绘制曲线具有许多额外的自由度。）另外，在绘制曲线之前，设计者可以用一种或多种方法限制曲线的7个自由度。 在3-Draw 的目前实现中，用户可以在Pick Draw模式下预先制定一个或曲线的两个端点（如游艇的例子所说那样）。预先指定限制的好处是抓住了一个曲线的形状，使得曲线的定位，定向，以及其它方面的扩展可以与其它曲线脱钩。用户可以通过单独集中形状获得更精确的形状，没有了在满意位置被限制的负担。用户使用使用“行瞄准采摘技术选择曲线和点。用户将笔尖对齐屏幕上的对象并点击手写笔按钮。之后3-Draw 在采摘区搜索RGB颜色的像素字节，为了显示存在的曲线和点的代码。3-Draw可以从这些颜色字节得到一个点或曲线的数目，因为它呈现每个对象都有独特的颜色代码。线的视线采摘使用户能够确定点的三维位置和曲线，没有实际履行虚拟笔困难的三维路线。要选择模糊的对象，用户只需简单地移动虚拟片来揭示隐藏的对象。采摘可用在默认的绘图模式下，这样，用户可以通过按住笔按钮和移动手写笔来画曲线，或通过点击笔的按钮来选择一个曲线或点。6.3 反射曲线在用户需要模型对称性的情况下，只有一半的模型需要被创建，另一半通过反射面反射曲线生成（如游艇的例子所讲）。结论3-Draw已经证明，在计算机绘制的三维模型是自然和快速的。一个熟练的设计家呈现了一个12米长含有近100条曲线的全3D游艇线框模型约需1.5小时。在某种程度上，这种高水平是同时使用双手提供复杂的三维信息的结果。3-Draw 提供强大的动觉反馈给用户，使他们感到他们持有在屏幕上观看的对象，并与他们迅速自然地互动。我们也可以归功于使用3-Draw时的简单设计方法， 这方法是基于我们可以把一个复杂的自由曲面的形状创造分成四个步骤。这四个步骤使纸和铅笔的绘制扩展到三维图形，计算机可以生成无视觉限制的模型，能让设计师简单地执行图像投影。3-Draw通过使对捕捉形状、缩放和方向脱钩，进一步简化了绘图过程。用户目前可以预先指定的一个或两个曲线的端点位置，在画曲线后再指定剩余的七个自由度。新手用户需要时间去了解一个想法如何转换成三维手势，但所需三维图形的协调远远比绘制角度或成为善于雕塑所需的技巧简单得多。3-Draw体现了一种形状设计的新方法，该方法允许直接探索三维设计和解放设计者在使用二维工具时的限制性。A Tool for Designing 3D Shapes1 IntroductionShape design contributes to a products commercial competitiveness through product differentiation, visual appeal, and ergonomic performance. Unfortunately, current CAD tools do not serve shape design well because they cant handle quick sketching of concepts. As a result, designers prefer to use manual sketching tools such as paper and pencil to rough out their initial ideas. Our goal is to develop a fundamentally new type of CAD system for designing shape. We want it to be sufficiently intuitive, easy to use, and powerful that industrial designers can and will use it from the initial stages of conceptual design right through to models suitable for manufacturing. We based our approach to shape design on a new paradigm that we could call “design directly in 3D. ” We particularly want to provide a user-friendly CAD interface that allows users to design complex free-form shapes by entering information directly in three dimensions using a pair of hand-held, six-degree-of-freedom sensors. Such a CAD system would significantly increase the speed and quality of shape design. 2 Current practiceIn the first stages of product design, industrial designers often create rough 2D sketches that only approximate the actual dimensions of the final product. Once a designer has committed to a particular alternative, the sketches are translated into a more precise 3D model. This process may involve clay mockups or entering data into a computer database. In either case, the transformation from 2D sketch to 3D model is subjective and ad hoc. The task requires extracting salient geometric features from the sketches (such as silhouette lines, highlight lines, and character lines) and dimensioning them to provide a 3D model. We can trace the complexity and limitations of current CAD systems to the fact that they employ this “2D-to-3D” paradigm, in which 3D shape is built up from 2D inputs. The user must communicate 3D information using a 2D input device such as a mouse, joystick, or tablet. In addition to employing the 2D-to-3D paradigm, most of todays CAD systems also require a highly detailed underlying representation for shape. As a result, users must model a shape in a planned and methodical A typical session on todays 3D CAD systems involves specifying layers, local coordinate systems, and working planes in addition to point coordinates that must later be fitted with splines. Some systems allow the designer to draw planar (2D) curves that are then approximated with splines. To obtain 3D curves, however, the designer must move individual control points by specifying the direction and magnitude of out-of-plane displacement. These systems all require the user to work at the control-point level rather than with the shape itself. As a result, the user is forced to spend a greater percentage of time specifying how and where to accomplish a task rather than on the task itself. This invariably results in a “nonstraightforward” system and a complex user interface. 3 Related workSeveral research efforts have focused on developing CAD systems for the initial stages of design. Pentland developed such a CAD tool, called Supersketch, on the premise that people naturally see objects as “lumps of clay. ” Users of Supersketch develop models by deforming primitive shapes and joining them to form objects. In separate research, Pentland also developed a sketching tool that extracts 3D features from 2D perspective drawing. Editing is possible after drawing the initial curves. One of the most recent developments in computer interfaces is the 3D input device. Several early designs incorporated mechanical linkages attached to measuring devices. In another approach, Feldman provided force feedback as well as position and orientation sensing. His Joystring consisted of a hand-held three-inch T-bar connected by wires to servomotors and controlled by a supercomputer. Such mechanical sensing tools provided fine resolution at the expense of being awkward and limiting in the freedom of motion. A 3D input device called the Sensor Frame continuously senses finger position and orientation in the vicinity of a CRT. Using an array of optical sensors to detect hand movements, this device can be used to “sculpt” parts on a screen. Another recent development is a six-degree-of-freedom ball, currently available from several vendors. This tool, which resembles the handle on a gearshift lever, allows users to manipulate objects within a virtual world. Pushing, pulling, and twisting forces are converted into translations and rotations. Several research efforts with 3D input devices have focused attention on the Polhemus 3Space six-degree-of-freedom device. This input device employs magnetic fields, eliminating the need for mechanical linkages between the hand-held sensors and the computer. In one project, Ware and Jessome created a six-degree-of-freedom mouse named the “Bat” to explore manipulation of 3D scenes. ” They discovered that the most effective way to overcome depth perception ambiguities was to toggle a 90 degree flip about a vertical axis while disabling motion into the screen. They also found that approximate object placement became a trivial task with the Polhemus device compared with using conventional mouse interfaces. They credited this result to the kinesthetic correspondence between hand and object movement. The Polhemus device has also been used to explore the correspondence between spatial input and stereoscopic display in a graphics workstation. Schmandt implemented a single-sensor Polhemus as a 6D “magic wand. ” His goal was to discover the interactive capabilities of stereoscopic displays by projecting stereo images into an unobscured space and letting the user reach into the projected image with the wand. He first developed a 3D paint program that created painted regions in the vicinity of the wand. To project images into the workspace, Schmandt used a half-silvered mirror and a conventional video monitor mounted above and in front of the user at a 45 degree angle. The user wore electrically shuttered glasses and viewed time-multiplexed stereo images reflected from the CRT onto the mirror. Schmandts initial results indicated a good natural correspondence between the wand and the projected images of painted lines emanating from the tip. However, magnetic field disturbances from the monitor and the time lag between hand motion and graphic rendering detracted from the interactive feel of the tool. The Polhemus device has also been used by VPL Research in the Eyephone and DataGlove products to enable users to operate within virtual worlds. The Eyephone is a head-mounted display that provides wide peripheral stereoscopic views of virtual worlds. By incorporating a Polhemus sensor to track head motions, the Eyephone enables the computer to match the position and orientation of the viewer in the virtual world with the position and orientation of the users head. Together, these two devices allow a user to move through a virtual world and interact with objects via the hand motions of the glove. Finally, the VPL glove has been used in conjunction with a stereoscopic display to explore interactive modeling of 3D surfaces. DeRoses 3D Design Space lets users reach in using the glove and move 3D points that form the “control nets” of complex polynomial surfaces. 4 Sketch of a 12-meter yachtIn an attempt to capture the flavor of 3-Draw, we initiated construction of a sample model of a 12-meter yacht. The following text describes the steps the designer took to complete the model using 3-Draws currently implemented features. Accompanying figures trace the designers progress. To begin, the designer first drew two curves of the yacht model to establish the starboard deck and keel contours (see Figure 4.1). He drew the deck curve with a single hand motion as an unattached 3D curve in space. Next he drew the keel contour using Pick-Draw. To use Pick-Draw, the designer preselected two points on the starboard deck (for example, one at the bow, the other at the stern) as the new curves endpoints. Preselecting endpoints in advance allowed the designer to concentrate on capturing the shape of the curve without worrying about its scale, placement, and orientation during the drawing process. After the curve was drawn and automatically snapped to fit the endpoints, the designer specified the new curves orientation about the axis joining its endpoints by twisting the hand-held stylus. Then, by pressing the stylus button, the designer released the curve in the desired orientation. In the next step of creating the yacht, the designer positioned a reflection plane to establish the center line of the boat. In 3-Draw, a default reflection plane can be “picked up” with the stylus, then positioned and oriented with natural motions of the hand holding the stylus. When released (by releasing the stylus button), the reflection plane remains fixed in its position and orientation relative to the other curves comprising the object. The user then creates mirrored versions of curves by picking them with the stylus. Figure 4.1 illustrates the reflected deck and keel curves that establish the port side of the boat. Figure 4.1 3-Draw image showing newly reflected port deck and keel curvesTo establish shape in the hull, the designer next drew the starboard bulkheads using the Pick-Draw method with two preselected points (see Figure 4.2). After capturing the shape of each bulkhead, the designer set the amount of twist through the axis joining the bulkheads endpoints by twisting the stylus. He then pressed the stylus button to release the curve. (Figure 4.2 also illustrates one of the bulkheads prior to being twisted into the proper position.) Additionally, the designer can later alter the amount of axis twist of any curve by using the 3-Draw feature Axis-Twist. After reflecting the bulkheads across the center plane to complete the port contour, the designer added a rudder as well as a keel and deck crosspieces using two-point Pick-Draw (see Figure 4.3). Figure 4.2 3-Draw image showing bulkheads. Note the twisting of a bulkhead curve, accomplished using the pen Figure 4.3 3-Draw image showing rudder as well as keel and deck crosspiecesNext, the designer began to sketch the winged keel. Rather than preselect two endpoints, he drew the starboard profile as a freestanding curve so that he could later move it into position using the features Translate and Rotate. The designer sketched the winged keel as a free-form curve in space. He used the features Tweezers and Displacement Gun to fine-tune the curves shape. Tweezers allows the designer to grab a point with the virtual pen and stretch the curve in the styluss direction of motion. When the Displacement Gun is executed, 3-Draw renders a translucent cone radiating from the tip of the virtual stylus. As the designer sweeps this cone through portions of curves, the curves are deflected away as if from an air gun deflecting a string floating in space. Similar to Tweezers, the Displacement Gun offsets individual point coordinates in order to bring about the desired deformation of the curve. Once deformed, a curve retains its new shape. Before attaching the 3D profile of the winged keel, the designer cut one of the curves at the base of the keel using Erase Point. He deleted the internal curve segment using Erase Line. Then he attached the new curve to the hull by picking the three curves to be linked with the feature Fuse. Once curves are fused into single curves, they can be moved and reflected easily in one step. The designer added some curved crosspieces using twopoint Pick-Draw to obtain the model shown in Figure 4.4. Figure 4.4 3-Draw image showi