We present a smooth, every-where $C^k$, analytic surface representation for closed surfaces of arbitrary topology. We demonstrate fitting this representation to meshes of varying resolutions and sampling quality. The fitting process is adaptive and provides controls for both the average and the maximum allowable error. The representation is suitable for applications which require consistent parameterizations across different surfaces.

We present a framework for interactive sketching that allows users to create three-dimensional (3D) architectural models quickly and easily from a source drawing. The sketching process has four steps. (1) The user calibrates a viewing camera by specifying the origin and vanishing points of the drawing. (2) The user outlines surface polygons in the drawing. (3) A 3D reconstruction algorithm uses perceptual constraints to determine the closest visual fit for the polygon. (4) The user can then adjust aesthetic controls to produce several stylistic effects in the scene: a smooth transition between day and night rendering, a horizon knockout effect and entourage figures. The major advantage of our approach lies in the combination of perception-based techniques, which allow us to minimize unnecessary interactions, and a hinging-angle scheme, which shows significant improvement in numerical stability over previous optimization-based 3D reconstruction algorithms. We also demonstrate how our reconstruction algorithm can be extended to work with perspective images, a feature unavailable in previous approaches.

A new experiment is presented which demonstrates the usefulness of an image space modulation technique called Subtle Gaze Direction (SGD) for guiding t he user in a simple searching task. SGD uses image space modulations in the luminance channel to guide a viewer's gaze about a scene without interruptin g their visual experience. The goal of SGD is to direct a viewer's gaze to certain regions of a scene without introducing noticeable changes in the ima ge. Using a simple searching task we compared performance using no modulation, using subtle modulation and using obvious modulation. Results from the e xperiments show improved performance when using subtle gaze direction, without affecting the user's perception of the image. Results establish the pote ntial of the method for a wide range of applications including gaming, perceptually based rendering, navigation in virtual environments and medical sear ch tasks.

Viewing data sampled on complicated geometry, such as a helix or a torus, is hard because a single camera view can only encompass a part of the object. Either multiple views or non-linear projection can be used to expose more of the object in a single view, however, specifying such views is challenging because of the large number of parameters involved. We show that a small set of versatile widgets can be used to quickly and simply specify a wide variety of such views. These widgets are built on top of a general framework that in turn encapsulates a variety of complicated camera placement issues into a more natural set of parameters, making the specification of new widgets, or combining multiple widgets, simpler. This framework is entirely view-based and leaves intact the underlying geometry of the dataset, making it applicable to a wide range of data types.

Cardiac fluorescent optical imaging provides the unique opportunity to investigate the dynamics of propagating electrical waves during ventricular arrhythmias and the termination of arrhythmias by strong electric shocks. Panoramic imaging systems using charge-coupled device been developed to overcome the inability to monitor electrical activity from the entire cardiac surface. Photodiode arrays known to have higher temporal resolution and signal quality, but lower spatial resolution compared to CCD cameras. We construct a panoramic imaging system with three PDAs and image Langendorff perfused rabbit hearts pacing, and arrhythmias. The recorded spatiotemporal dynamics of electrical activity is texture mapped onto a reconstructed 3-D geometrical heart model specific to each heart studied. The PDA-based system provides sufficient spatial resolution The reconstructed 3-D electrical activity provides us with a powerful tool to investigate the fundamental mechanisms of arrhythmia maintenance and termination. features is a nontrivial task.

This paper presents a novel rendering technique inspired by artistic approaches. Instead of trying to recreate a traditional medium, such as charcoal or watercolor, this approach is a mixture of both photo-realism and abstraction. Artists use a process of abstraction to provide structural information about subjects that do not have clearly defined shapes, such as groups of leaves in a tree. For example, an artist will use a color wash to first approximate a group of leaves. They then add detail on top of parts of this wash to indicate the presence of individual leaves. Similarly, we use an abstract shape that approximates the image of leaves clustered at the end of a branch. To prevent oversimplification, we add photo-realistic detail using a blending process. Interframe coherence is achieved both by smoothly interpolating the abstract shapes, and the continuity inherent in the photo-realistically rendered detail.

We present a system for painting how the appearance of an object changes under different lighting and viewing conditions. The user paints what the object should look like under different lighting conditions (dark, partially dark, fully lit, etc.), or different viewing angles, or both. The system renders the object under new lighting conditions and a new viewing angle by combining these paintings. For surfaces without a pre-defined texture map the system can construct texture maps directly from the user’s paintings.

Alexa’s method for linearly interpolating matrices is well-suited for application to camera matrices. This paper discusses implementation issues which arise when applying this method to camera interpolation, and shows how to sidestep several major pitfalls. We present implementation details for three interpolation methods: linear, spline-based smooth approximation, and smooth interpolation using subdivision. We also discuss general issues involved in implementing Alexa’s method. We note changes to his provided pseudocode, and we dicuss suitable values for e to maximize efficiency. Finally, we show examples of the technique and compare it to traditional interpolation methods.

Orthopedists invest significant amounts of effort and time trying to understand the biomechanics of arthrodial (gliding) joints. While new image acquisition and processing methods currently generate richer-than-ever geometry and kinematic datasets that are individual-specific, the computational and visualization tools needed to enable the comparative analysis and exploration of these datasets lag behind. In this paper, we present a framework that enables the crossdataset visual exploration and analysis of arthrodial joint biomechanics. Central to our approach is a computer-vision inspired markerless method for establishing pairwise correspondences between individual-specific geometry. Manifold models are subsequently defined and deformed from one individual-specific geometry to another such that the markerless correspondences are preserved while minimizing model distortion. The resulted mutuallyconsistent parameterization and visualization allow the users to explore the similarities and differences between two datasets, and to define meaningful quantitative measures. We present two applications of this framework to human wrist data: articular cartilage transfer from cadaver data to in vivo data, and cross-dataset kinematics analysis. The method allows our users to combine complementary geometry acquired through different modalities, and thus overcome current imaging limitations. Results demonstrate the technique useful in the study of normal and injured anatomy and kinematics of arthrodial joints. In principle, the pairwise cross-parameterization method applies to all spherical topology data from the same class, and should be particularly beneficial in instances where identifying salient object

Existing environment mapping techniques include spherical mapping and cube mapping. These techniques have inherent flaws that cause sampling issues and aliasing. Continuous cube mapping is offered as an alternative environment mapping approach that effectively folds the cube onto the sphere, providing a better parameterization of cube mapping. We provide a hardware implementation.

We introduce a novel concept called perceptually meaningful image editing and present techniques for manipulating the apparent depth of objects and creating the illusion of motion in 2D images. Our techniques combine principles of human visual perception with approaches developed by traditional artists. For our depth manipulation technique, the user loads an image, selects an object and specifies whether the object should appear closer or further away. The system automatically determines luminance or color temperature target values for the object and/or background that achieve the desired depth change. Our approach for creating the illusion of motion exploits the differences between our peripheral vision and our foveal vision by introducing spatial imprecision to the image.

The work presented in this paper is an interactive interface for automatic synthesis of textures from a given sketch. Reaction-Diffusion (RD) is used as the basis for texture synthesis. The major difficulty of determining appropriate initial values of the RD system is resolved by employing machine learning techniques. The paper describes the implementation of this system which allows a user to sketch a pattern, automatically generates a pattern with the same attributes and allows the user to interactively create more complex textures by adding other patterns as well as manipulate the color of the resulting texture.

We apply simplified image-based lighting methods to reduce the equipment, cost, time, and specialized skills required for high-quality photographic lighting of desktop-sized static objects such as museum artifacts. We place the object and a computer-steered moving-head spotlight inside a simple foam-core enclosure and use a camera to record photos as the light scans the box interior. Optimization, guided by interactive user sketching, selects a small set of these photos whose weighted sum best matches the user-defined target sketch. Unlike previous image-based relighting efforts, our method requires only a single area light source, yet it can achieve high-resolution light positioning to avoid multiple sharp shadows. A reduced version uses only a handheld light and may be suitable for battery-powered field photography equipment that fits into a backpack.

The goal of this research is to develop new and more intuitive ways for editing a mesh from a static camera angle. I present two ways to edit a mesh via a simple sketching system. The first method is a gray-scale editor which allows the user to specify a fall off function for the region being deformed. The second method is a profile editor in which the user can re-sketch a mesh’s profile. Lastly, the types of edits possible will be discussed and our results will be presented.

Camera manipulation is a hard problem since a graphics camera is defined by specifying 11 independent parameters. Manipulating such a high-dimensional space to accomplish specific tasks is difficult and requires a certain amount of expertise. We present an intuitive interface that allows novice users to perform camera operations in terms of the change they want see in the image. In addition to developing a natural means for camera interaction, our system also includes a novel interface for viewing and organizing previously saved views. When exploring complex 3D data-sets a single view is not sufficient. Instead, a composite view built from multiple views may be more useful. While changing a single camera is hard enough, manipulating several cameras in a single scene is still harder. In this thesis, we also present a framework for creating composite views and an interface that allows users to manipulate such views in real-time.

A method is described for texturing surfaces using decals, images placed on the surface using local parameterizations. Decal parameterizations are generated with a novel O(N logN) discrete approximation to the exponential map which requires only a single additional step in Dijkstra's graph-distance algorithm. Decals are dynamically composited in an interface that addresses many limitations of previous work. Tools for image processing, deformation/feature-matching, and vector graphics are implemented using direct surface interaction. Exponential map decals can contain holes and can also be combined with conformal parameterization to reduce distortion. The exponential map approximation can be computed on any point set, including meshes and sampled implicit surfaces, and is relatively stable under resampling. The decals stick to the surface as it is interactively deformed, allowing the texture to be preserved even if the surface changes topology. These properties make exponential map decals a suitable approach for texturing animated implicit surfaces.

Automatic detection of features in three-dimensional objects is a critical part of shape matching tasks such as object registration and recognition. Previous approaches often required some type of user interaction to select features. Manual selection of corresponding features and subjective determination of the difference between objects are time consuming processes requiring a high level of expertise. The Curvature Map represents shape information for a point and its surrounding region and is robust with respect to grid resolution and mesh regularity. It can be used as a measure of local surface similarity.We use these curvature map properties to extract feature regions of an object. To make the selection of the feature region less subjective, we employ a min-cut/max-flow graph cut algorithm with vertex weights derived from the curvature map property. A multi-scale approach is used to minimize the dependence on user defined parameters. We show that by combining curvature maps and graph cuts in a multi-scale framework, we can extract meaningful features in a robust way.

Colors that appear closer to the red end of the visible spectrum are said to be warm while the colors that appear closer to the blue end are said to be cool. The phenomenon of warmer colors appearing nearer in depth to viewers than cooler colors has been studied extensively by psychologists and other vision researchers (see [Sundet 1978] for a summary). The vast majority of these studies have asked human observers to view physically equidistant, colored stimuli and compare them for relative depth. However, in most cases, the stimuli presented were rather simple: straight colored lines, uniform color patches, point light sources, or symmetrical objects with uniform shading. Additionally, the colors used were typically highly saturated. Although such stimuli are useful in isolating and studying depth cues in certain contexts, they leave open the question of whether the human visual system operates similarly for realistic objects. This paper presents the results of an experiment designed to explore the color-depth relationship for realistic, colored objects with varying shading and contours.

This paper takes a systematic look at methods for estimating the curvature of surfaces represented by triangular meshes. We have developed a suite of test cases for assessing both the detailed behavior of these methods, and the error statistics that occur for samples from a general mesh. Detailed behavior is represented by the sensitivity of curvature calculation methods to noise, mesh resolution, and mesh regularity factors. Statistical analysis breaks out the effects of valence, triangle shape, and curvature sign. These tests are applied to existing discrete curvature approximation techniques and common surface fitting methods.We provide a summary of existing curvature estimation methods, and also look at alternatives to the standard parameterization techniques. The results illustrate the impact of noise and mesh related issues on the accuracy of these methods and provide guidance in choosing an appropriate method for applications requiring curvature estimates.

Cardiac fluorescent optical imaging provides the unique opportunity to investigate the dynamics of propagating electrical waves during ventricular arrhythmias and the termination of arrhythmias by strong electric shocks. Panoramic imaging systems using charge-coupled device CCD cameras as the photodetector have been developed to overcome the inability to monitor electrical activity from the entire cardiac surface. Photodiode arrays PDAs are known to have higher temporal resolution and signal quality, but lower spatial resolution compared to CCD cameras. We construct a panoramic imaging system with three PDAs and image Langendorff perfused rabbit hearts n=18 during normal sinus rhythm, epicardial pacing, and arrhythmias. The recorded spatiotemporal dynamics of electrical activity is texture mapped onto a reconstructed 3-D geometrical heart model specific to each heart studied. The PDA-based system provides sufficient spatial resolution 1.72 mm without interpolation for the study of wavefront propagation in the rabbit heart. The reconstructed 3-D electrical activity provides us with a powerful tool to investigate the fundamental mechanisms of arrhythmia maintenance and termination.

We present a surface reconstruction technique that constructs a smooth, Ck, analytic surface from scattered data. The technique is robust to noise and both poorly and non-uniformly sampled data, making it well-suited for use in medical applications. In addition, the surface can be parameterized in multiple ways, making it possible to represent additional data, such as electromagnetic potential, in a different (but related) coordinate system to the geometric one. The parameterization technique also supports consistent parameterizations of multiple data sets.

This paper demonstrates the use of image-space constraints for key frame interpolation. Interpolating in image-space results in sequences with predictable and controlable image trajectories and projected size for selected objects, particularly in cases where the desired center of rotation is not fixed or when the key frames contain perspective distortion changes. Additionally, we provide the user with direct image-space control over how the key frames are interpolated by allowing them to directly edit the object’s projected size and trajectory. Image-space key frame interpolation requires solving the inverse camera problem over a sequence of point constraints. This is a variation of the standard camera pose problem, with the additional constraint that the sequence be visually smooth. We use image-space camera interpolation to globally control the projection, and traditional camera interpolation locally to avoid smoothness problems. We compare and contrast three different constraint-solving systems in terms of accuracy, speed, and stability. The first approach was originally developed to solve this problem [Gleicher and Witken 1992]; we extend it to include internal camera parameter changes. The second approach uses a standard single-frame solver. The third approach is based on a novel camera formulation and we show that it is particularly suited to solving this problem.

Automatic detection of features in three-dimensional objects is a critical part of shape matching tasks such as object registration and recognition. Previous approaches to local surface matching have either focused on man-made objects, where features are generally well-defined, or required some type of user interaction to select features. Manual selection of corresponding features and subjective determination of the difference between objects are time consuming processes requiring a high level of expertise. Curvature is a useful property of a surface, but curvature calculation on a discrete mesh is often noisy and not always accurate. However, the Curvature Map, which represents shape information for a point and its surrounding region, is robust with respect to grid resolution and mesh regularity. It can be used as a measure of local surface similarity. We use these curvature map properties to extract features and segment the surface accordingly. Although thresholding techniques can be used to generate reasonable features, the choice of a threshold is very subjective and the results may be very sensitive to this choice. To avoid the threshold dilemma and to make the selection of the feature region less subjective, we employ a min-cut/max-flow graph cut algorithm, with vertex weights derived from the curvature map property. A multi-scale approach is used to minimize the dependence on user defined parameters. We show that by combining curvature maps and graph cuts in a multi-scale framework, we can extract meaningful features in a robust way.

The phenomenon of warmer colors appearing nearer in depth to viewers than cooler colors has been studied extensively by psychologists and other vision researchers. The vast majority of these studies have asked human observers to view physically equidistant, colored stimuli and compare them for relative depth. However, in most cases, the stimuli presented were rather simple: straight colored lines, uniform color patches, point light sources, or symmetrical objects with uniform shading. Additionally, the colors used were typically highly saturated. Although such stimuli are useful in isolating and studying depth cues in certain contexts, they leave open the question of whether the human visual system operates similarly for realistic objects. This paper presents the results of an experiment designed to explore the color-depth relationship for realistic, colored objects with varying shading and contours.

We introduce the concept of perceptually meaningful image editing and present two techniques for manipulating the apparent depth of objects in an image. The user loads an image, selects an object and specifies whether the object should appear closer or further away. The system automatically determines target values for the object and/or background that achieve the desired depth change. These depth editing operations, based on techniques used by traditional artists, manipulate either the luminance or color temperature of different regions of the image. By performing blending in the gradient domain and reconstruction with a Poisson solver, the appearance of false edges is minimized. The results of a preliminary user study, designed to evaluate the effectiveness of these techniques, are also presented.

We present the implementation of a new widget, the IBar, for controlling all aspects of a perspective camera. This widget provides an intuitive interface for controlling the perspective distortion in the scene by providing single handles that manipulate one or more projection parameters simultaneously (eg distance-to-object and lens aperture) in order to create a single perceived projection change (increasing the perspective distortion without changing the scene size).

Linear perspective is a good approximation to the format in which the human visual system conveys 3D scene information to the brain. Artists expressing 3D scenes, however, create nonlinear projections that balance their linear perspective view of a scene with elements of aesthetic style, layout and relative importance of scene objects. Manipulating the many parameters of a linear perspective camera to achieve a desired view is not easy. Controlling and combining multiple such cameras to specify a nonlinear projection is an even more cumbersome task. This paper presents a direct interface, where an artist manipulates in 2D the desired projection of a few features of the 3D scene. The features represent a rich set of constraints which define the overall projection of the 3D scene. Desirable properties of local linear perspective and global scene coherence drive a heuristic algorithm that attempts to interactively satisfy the given constraints as a weight-averaged projection of a minimal set of linear perspective cameras. This paper shows that 2D feature constraints are a direct and effective approach to control both the 2D layout of scene objects and the conceptually complex, high dimensional parameter space of nonlinear scene projection.

We present a surface modeling technique that supports adaptive resolution and hierarchical editing for surfaces of spherical topology. The resulting surface is analytic, $C^k$, and has a continuous local parameterization defined at every point. To manipulate these surfaces we describe a user-interface based on multiple, overlapping subdivision-style meshes.

The concept of "curved perspective\'\' has been used by artists such as M.C. Escher in order effectively convey a sense of three dimensional space while being restricted to a two dimensional canvas. We present an interactive system to create and manipulate projections with a curvilinear perspective. Our system presents the user with a set of intuitive screen-space perspective primitives that control the vanishing points of the scene. This allows the user to generate diverse projections having curved perspective.

The ability to identify similarities between shapes is important for applications such as medical diagnosis, object registration and alignment, and shape retrieval. In this paper we present a method, the Curvature Map, that uses surface curvature properties in a region around a point to create a unique signature for that point. These signatures can then be compared to determine the similarity of one point to another. To gather curvature information around a point we explore two techniques, rings (which use the local topology of the mesh) and Geodesic Fans (which trace geodesics along the mesh from the point). We explore a variety of comparison functions and provide experimental evidence for which ones provide the best discriminatory power. We show that Curvature Maps are both more robust and provide better discrimination than simply comparing the curvature at individual points.

We apply simplified image-based lighting methods to reduce the equipment, cost, time, and specialized skills required for high-quality photographic lighting of desktop-sized static objects such as museum artifacts. We place the object and a computer-steered moving-head spotlight inside a simple foam-core enclosure, and use a camera to quickly record low-resolution photos as the light scans the box interior. Optimization guided by interactive user sketching selects a small set of frames whose weighted sum best matches the target image. The system then repeats the lighting used in each of these frames, and constructs a high resolution result from re-photographed basis images. Unlike previous image-based re-lighting efforts, our method requires only one light source, yet can achieve high resolution light positioning to avoid multiple sharp shadows. A reduced version uses only a hand-held light, and may be suitable for battery-powered, field photography equipment that fits into a backpack.

This work presents a novel rendering technique inspired by artistic approaches. Instead of trying to recreate the appearance of a traditional medium, such as charcoal or watercolor, this approach is a mixture of both photo-realism and abstraction. Artists use a process of abstraction to provide structural information about subjects that do not have clearly defined shapes, such as groups of leaves in a tree. For example, an artist will first use a color wash to approximate a group of leaves, then add detail on top of parts of this wash to indicate individual leaves. Similarly, we use an abstract shape that approximates the image of leaves clustered at the end of a branch. To prevent oversimplification, we add photo-realistic detail using a blending process. Inter-frame coherence is achieved by smoothly interpolating the abstract shapes as well as by the continuity inherent in the photo-realistically rendered detail.

The Edinburgh Logical Framework (LF) has been proposed as a system for expressing inductively defined sets. I will present an inductive definition of the set of manifold meshes in LF. This definition takes into account the topological characterization of meshes, namely their Euler Characteristic. I will then present a set of dependent data types based on this inductive definition. These data types are defined in a programming language based on LF. The language’s type checking guarantees that any typeable expression represents a correct manifold mesh. Furthermore, any mesh can be represented using these data types. Hence, the encoding is sound and complete.

Linear perspective is a good approximation to the format in which the human visual system conveys 3D scene information to the brain. Artists expressing 3D scenes, however, create nonlinear projections that balance their linear perspective view of a scene with elements of aesthetic style, layout and relative importance of scene objects. Manipulating the many parameters of a linear perspective camera to achieve a desired view is not easy. Controlling and combining multiple such cameras to specify a nonlinear projection is an even more cumbersome task. This paper presents a direct interface, where an artist manipulates in 2D the desired projection of a few features of the 3D scene. The features represent a rich set of constraints which define the overall projection of the 3D scene. Desirable properties of local linear perspective and global scene coherence drive a heuristic algorithm that attempts to interactively satisfy the sketched constraints as a weight-averaged projection of a minimal set of linear perspective cameras. This paper shows that 2D feature constraints are a direct and effective approach to control both the 2D layout of scene objects and the conceptually complex, high dimensional parameter space of nonlinear scene projection. The simplicity of our interface also makes it an appealing alternative to standard through-the-lens and widget based techniques to control a single linear perspective camera.

We present a new screen space widget, the IBar, for effective camera control in 3D graphics environments. The IBar provides a compelling interface for controlling scene perspective based on the artistic concept of vanishing points. Various handles on the widget manipulate multiple camera parameters simultaneously to create a single perceived projection change. For example, changing just the perspective distortion is accomplished by simultaneously decreasing the camera\'s distance to the scene while increasing focal length. We demonstrate that the IBar is easier to learn for novice users and improves their understanding of camera perspective.

There are a variety of surface types (such as meshes and implicit surfaces) that lack a natural parameterization. We believe that manifolds are a natural method for representing parameterizations because of their ability to handle arbitrary topology and represent smooth surfaces. Manifolds provide a formal mechanism for creating local, overlapping parameterization and defining the functions that map between them. In this paper we present specific manifolds for several genus types (sphere, plane, n-holed tori, and cylinder) and an algorithm for establishing a bijective function between an input mesh and the manifold of the appropriate genus. This bijective function is used to define a smooth embedding of the manifold that approximates or interpolates the mesh. The smooth embedding is used to calculate analytical quantities, such as curvature and area, which can then be mapped back to the mesh. Possible applications include texture mapping, surface fitting, arbitrary topology surface modeling, feature detection, and surface comparison.

We present a novel method for modeling contact areas and ligament lengths in articulations. Our approach uses volume images generated by computed tomography and allows the in vivo and non-invasive study of articulations. In our method bones are modeled both implicitly (scalar distance fields) and parametrically (manifold surfaces). Using this double representation we compute inter-bone distances and joint contact areas. Using the same types of representation we model ligament paths; in our model the ligaments are approximated by shortest paths in a 3D space with bone obstacles. We demonstrate the method by applying our contact area and ligament model to the distal radioulnar joints of a volunteer diagnosed with malunited distal radius fracture in one forearm. Our approach highlights focal changes in the articulation at the distal radioulnar joint (location and area of bone contact) and potential soft-tissue constraints (increased `length\' of the distal ligaments and ligament-bone impingement in the injured forearms). Results suggest that the method could be useful in the study of normal and injured anatomy and kinematics of complex joints.

We have developed an autonomous robot system that takes well-composed photographs of people at social events, such as weddings and conference receptions. The robot, Lewis, navigates through the environment, opportunistically taking photographs of people. In this paper, we outline the overall architecture of the system and describe how the various components inter-relate. We also describe our experiences of deploying the robot photographer at a number of real-world events.

This paper takes a systematic look at calculating the curvature of surfaces represented by triangular meshes. We have developed a suite of test cases for assessing the sensitivity of curvature calculations, to noise, mesh resolution, and mesh regularity. These tests are applied to existing discrete curvature approximation techniques and three common surface fitting methods (polynomials, radial basis functions and conics). We also introduce a modification to the standard parameterization technique. Finally, we examine the behaviour of the curvature calculation techniques in the context of segmentation.

We present an image-space camera manipulation widget that supports visualization of the relationship of the camera with respect to the scene. The form of the widget presents the user with natural affordances for camera manipulation. Visual aids such as ghosting of the scene and preview animations are used to acquaint novice users with the functions of different parts of the widget. Mouse gestures are used to transition between different perspective views of the scene in an intuitive way. Finally, we provide a novel method for visualizing camera bookmarks.

The primary goal of this research is to develop a perception based image editing system. The input to this system will be either a rendered image, a photograph, or a high dynamic range image. We are currently developing techniques that allow the user to edit these images in a perceptually intuitive manner. Specifically we are considering the following image editing features:

We present a new widget, the IBar, for controlling all aspects of a perspective camera. This widget provides an intuitive interface for controlling the perspective distortion in the scene by providing single handles that manipulate one or more projection parameters simultaneously (e.g., distanceto- object and lens aperture) in order to create a single perceived projection change (increasing the perspective distortion without changing the scene size). We demonstrate that novice users more easily learn how to manipulate the camera using the IBar.

This paper presents several methods for shading meshes from scanned paint samples that represent dark to light transitions. Our techniques emphasize artistic control of brush stroke texture and color. We first demonstrate how the texture of the paint sample can be separated from its color gradient. We demonstrate three methods, two real-time and one off-line for producing rendered, shaded images from the texture samples. All three techniques use texture synthesis to generate additional paint samples. Finally, we develop metrics for evaluating how well each method achieves our goal in terms of texture similarity, shading correctness and temporal coherence.

We describe a complete, end-to-end system for taking well-composed photographs using a mobile robot. The general scenario is a reception, or other event, where people are roaming around talking to each other. The robot serves as an "event photographer\'\', roaming around the same space as the participants, periodically taking photographs. These images are then sent to a workstation where participants can print the photographs out, or email them.

We define a parameterization for an n-holed tori based on the hyperbolic polygon. We model the domain using a manifold with 2n+2 charts, and linear fractional transformations for transition functions. We embed the manifold using standard spline techniques to produce a surface.

The goal of this research is to explore techniques for shading 3D computer generated models using scanned images of actual paint samples. The techniques presented emphasize artistic control of brush stroke texture and color. We first demonstrate how the texture of a paint sample can be separated from its color transition. Four methods, three real-time and one off-line, for producing rendered images from the paint samples are then presented. Finally, we develop metrics for evaluating how well each method achieves our goal in terms of texture similarity, shading correctness, and temporal coherence.

We have developed an autonomous robot system that takes well-composed photographs of people at social events, such as weddings and conference receptions. The robot, Lewis, navigates through the environment, opportunistically taking photographs of people. In this paper, we outline the overall architecture of the system and describe how the various components inter-relate. We also describe our experiences of deploying the robot photographer at a number of real-world events.

We present a 3D input device consisting of a stiff piece of paper which is tracked by a digital video camera. The user can also draw on the paper using a pen-like device. The user moves the paper to specify the location of a virtual plane. By drawing on the paper, the user can specify points in 3D space. The primary technical contribution of this paper is a new pose estimation algorithm suitable for a hand-held, moving pattern. To demonstrate the usefulness of the device we developed a sketching application for simple characters. The characters are constructed by sketching and joining together 3D ellipses, much as traditional cartoon characters are created in 2D using 2D ellipses.

We describe a new pose estimation approach for a 3D object capture system. This 3D pose estimation approach offers several advantages: increased visibility, robustness to lighting conditions, and improved reliability with evenly distributed errors. The calibration pattern is built using 3D conic features. We use simplex search to find the camera position and orientations that minimizes the error between the projected 3D cone features and the corresponding 2D image features. We demonstrate that our approach is accurate, efficient and robust.

The goal of this project is to develop a 3D input device using a stiff piece of paper and a camera. The camera tracks the piece of paper in 3D space. The user orients the paper in 3D space and then draws on the paper using a pen-like device. The camera tracks the movement of the pen on the piece of paper. The location of the pen in 3D space can then be calculated from the orientation of the paper. A drawing application that uses this 3D input device was also developed. The application allows a user to make characters by sketching ellipses. The drawing application creates a virtual rendering of the paper and displays this to the user. As the user positions the real paper, the virtual one mirrors its movements. The user can draw shapes on the paper. These shapes then get rendered in the virtual scene.

We have deployed a robot "photographer\'\' at several events. The robot, Lewis, navigates through the space, opportunistically taking photographs of people. We summarize the different types of human-robot interactions we have observed at these events, and put forth some possible explanations for the different behaviors. We also discuss potential models for human-robot interactions in this constrained setting.

This paper presents several methods for shading meshes from scanned paint samples that represent dark to light transitions. Our techniques emphasize artistic control of brush stroke texture and color. We first demonstrate how the texture of the paint sample can be separated from its color gradient. We demonstrate three methods, two real-time and one off-line for producing rendered, shaded images from the texture samples. All three techniques use texture synthesis to generate additional paint samples. Finally, we develop metrics for evaluating how well each method achieves our goal in terms of texture similarity, shading correctness and temporal coherence.

We present a system for painting how the appearance of an object changes under different lighting and viewing conditions. The user paints what the object should look like under different lighting conditions (dark, partially dark, fully lit, etc.) and (optionally) different viewing angles. The system renders the object under new lighting conditions and a new viewing angle by combining these paintings. We also provide a technique for constructing texture maps directly from the user’s paintings.

We explain how to use simple composition rules to drive an automated, mobile photography system. The composition rules are used to determine both the location for a good photograph, and how to frame that photograph. We describe the composition component in the context of a larger application, a robotic photographer. The robot moves around an area with people in it, opportunistically looking for faces and taking photographs. We describe both how to find faces in the world and how to create “good” photographs of those faces.

Altered kinematics of the distal radioulnar joint (DRUJ) and/or bone impingement are considered causes of long-term complications associated with malunited distal radius fractures. However, a recent CT image-based in vivo study of patients with malunited distal radius fractures found that malunion did not alter forearm kinematics and that limitations of pronosupination were not caused by bony impingement (Moore et al., in press). In this study, data from the previous study was reanalyzed to explore focal changes in the articulation at the DRUJ (location and area of bone contact) and potential soft tissue constraints (\'length\' of the dorsal and palmar radioulnar ligaments).

We present a surface modeling technique using manifolds. Our approach uses a single, simple parameterization for all surfaces of a given genus. This differs from previous approaches which build a parameterization based on the elements of a mesh. The simple parameterization is more appropriate for applications that do complex operations in parameter space or on the mesh surface. We define a manifold and a corresponding embedding function for three genera (plane, sphere, and torus). The manifold can be used simply as a parameterization tool or as a smooth surface approximating the original mesh. We demonstrate how to build a correspondence between the mesh and the manifold, then how to build an embedding that approximates the mesh.

We present a technique for fitting a smooth, locally parameterized surface model (called the manifold surface model) to unevenly scattered data describing an anatomical structure. This data is acquired from medical imaging modalities such as CT scans or MRI. The manifold surface is useful for problems which require analyzable or parametric surfaces fitted to data acquired from surfaces of arbitrary topology (e.g., entire bones). This surface modeling work is part of a larger project to model and analyze skeletal joints, in particular the complex of small bones within the wrist and hand. To demonstrate the suitability of this model we fit to several different bones in the hand, and to the same bone from multiple people.

We present a technique for blending multiple images of an object into a single, view-dependent texture map for that object. This technique can be used for image-based rendering, when the object is known, or for "painting" a view-dependent texture map of an object. The technique provides a structured mechanism for combining images at different resolutions, producing a mip-map like structure with the different levels constructed from different images. The user controls the camera angles for which a given image is valid. The problem of gaps caused by self-occlusion and non-overlapping images is also dealt with. This technique is also suitable for use on an object that will be animated.

We present a technique for adding view-dependent information to geometry, allowing us to write down, for each point on the surface, what the surface looks like from every direction. This data structure allows the incorporation of "image-based" objects into ordinary rendering, allows broader viewing conditions for image-based objects (one can look at a torus from "inside the hole," for example), and provides opportunities for compression of image-based objects and for separate interpolation of intensity and color data for image-based objects. We based our work on a surface representation that supports models of arbitrary topology, and discuss various methods for capturing, representing, and compressing the view-dependent information.

We demonstrate a method for visualizing inter-bone distances in articular joints. Visualization of inter-bone distances has the potential to characterize 3D structures and spatial relations non-invasively in complex joints.

In traditional art a painter displays a 3D scene on a 2D image plane in a manner that is aesthetically pleasing. The arrangement of objects and colors is called composition and is the subject of many art books and classes. While a painter may use perspective to create depth in a scene they may also alter the perspective and color, either subtly or dramatically, to influence the focus of viewer and the effect of the image. To date, traditional 3D graphics packages have largely concentrated on modeling, textures, and lighting to create images and provide few tools for altering the composition post-rendering. In this paper we present several simple techniques for creating images with non-standard perspective and color using standard 3D rendering packages. The scene is modeled in 3D but each object has its own camera, color balance, and image size, allowing the user to alter the composition after the 3D rendering step. The purpose of this paper is not to present a complete composition system but rather to illustrate the potential of composition-based tools.

Editing curves and surfaces is difficult in part because their mathematical representations rarely correspond to most people\'s idea of a curve or surface. The implementation (and hence, behavior) of most manipulation tools is intertwined with a particular curve or surface representation; this can make reimplementing the tool with a different representation problematic. A system using a single representation must therefore either limit the types of tools available or convert existing tools to work on the system\'s representation. In this paper we present a framework for editing curves or surfaces which supports multiple representations and ensures that they stay synchronized. As a proof of concept, we have created a curve editor which contains several tools each of which manipulate one of three different curve representations: polylines, NURBs, and multi-resolution B-splines.

We have created a system for capturing both the three-dimensional geometry and color and shading information for human facial expressions. We use this data to reconstruct photorealistic, 3D animations of the captured expressions. The system uses a large set of sampling points on the face to accurately track the three dimensional deformations of the face. Simultaneously with the tracking of the geometric data, we capture multiple high resolution, registered video images of the face. These images are used to create a texture map sequence for a three dimensional polygonal face model which can then be rendered on standard 3D graphics hardware. The resulting facial animation is surprisingly life-like and looks very much like the original live performance. Separating the capture of the geometry from the texture images eliminates much of the variance in the image data due to motion, which increases compression ratios. Although the primary emphasis of our work is not compression we have investigated the use of a novel method to compress the geometric data based on principal components analysis. The texture sequence is compressed using an MPEG4 video codec. Animations reconstructed from 512x512 pixel textures look good at data rates as low as 240 Kbits per second.

We introduce an implicit generalized cylinder which is constructed from an axis and one or more profile curves. This surface is related to the sweep which is traditionally formed by an axis curve and one or more cross sections. Instead of cross sections, this definition uses profile curves to define how far the surface is from the axis. This facilitates the construction of surfaces which have continually varying cross sections. We extend this definition to a generalized cylinder with two axis curves. This model is useful for surfaces where one side of the surface curves sharply away from the axis while the axis itself is curving. We also present a user interface for editing these sweeps.

This paper explores the use of visual operators for solids modeling. We focus on designing interfaces for free-form operators such as blends, sweeps, and deformations, because these operators have a large number of interacting parameters whose effects are often determined by an underlying parameterization. In this type of interactive modeling good solutions to the design problem have aesthetic as well as engineering components. Traditionally, interaction with the parameters of these operators has been through text editors, curve editors, or trial-and-error with a slider bar. Parametric values have been estimated from data, but not interactively. These parameters are usually one- or two-dimensional, but the operators themselves are intrinsically three-dimensional in that they are used to model surfaces visualized in 3D. The traditional textual style of interaction is tedious and interposes a level of abstraction between the parameters and the resulting surface. A 3D visual interface has the potential to reduce or eliminate these problems by combining parameters and representing them with a higher-level visual tool. The visual tools we present not only speed up the process of determining good parameter values but also provide visual interactions that are either independent of the particular parameterizations or make explicit the effect of the parameterizations. Additionally, these tools can be manipulated in the same 3D space as the surfaces produced by the operators, supporting quick, interactive exploration of the large design space of these free-form operators. This paper discusses the difficulties in creating a coherent user interface for interactive modeling. To this end we present four principles for designing visual operators, using several free-form visual operators as concrete examples.

We describe an extension of B-splines to surfaces of arbitrary topology, including arbitrary boundaries. The technique inherits many of the properties of B-splines: local control, a compact representation, and guaranteed continuity of arbitrary degree. The surface is specified using a polyhedral control mesh instead of a rectangular one; the resulting surface approximates the polyhedral mesh much as a B-spline approximates its rectangular control mesh. Like a B-spline, the surface is a single, continuous object. This is achieved by modeling the domain of the surface with a manifold whose topology matches that of the polyhedral mesh, then embedding this domain into 3-space using a basis-function/control-point formulation. We provide a constructive approach to building a manifold.

We present a method for approximating an isosurface with a smooth parametric representation. From the isosurface we first produce a patch mesh, a description of how many surface patches there are and how they are connected. We then create a smooth surface from the patch mesh.