John Dill

Navigating hierarchically clustered networks through fisheye and full-zoom methods

Many information structures are represented as two-dimensional networks (connected graphs) of links and nodes. Because these network tend to be large and quite complex, people often perfer to view part or all of the network at varying levels of detail. Hierarchical clustering provides a framework for viewing the network at different levels of detail by superimposing a hierarchy on it. Nodes are grouped into clusters, and clusters are themselves place into other clusters. Users can then navigate these clusters until an appropiate level of detail is reached. This article describes an experiment comparing two methods for viewing hierarchically clustered networks. Traditional full-zoom techniques provide details of only the current level of the hierarchy. In contrast, fisheye views, generated by the “variable-zoom” algorithm described in this article, provide information about higher levels as well. Subjects using both viewing methods were given problem-solving tasks requiring them to navigate a network, in this case, a simulated telephone system, and to reroute links in it. Results suggest that the greater context provided by fisheye views significantly improved user performance. Users were quicker to complete their task and made fewer unnecessary navigational steps through the hierarchy. This validation of fisheye views in important for designers of interfaces to complicated monitoring systems, such as control rooms for supervisory control and data acquistion systems, where efficient human performance is often critical. However, control room operators remained concerned about the size and visibility tradeoffs between the fine room operators remained concerned about the size and visibility tradeoffs between the fine detail provided by full-zoom techniques and the global context supplied by fisheye views. Specific interface feaures are required to reconcile the differences.

The continuous zoom: a constrained fisheye technique for viewing and navigating large information spaces

Navigating and viewing large information spaces, such as hierarchically-organized networks from complex realtime systems, suffer the problems of viewing a large space on a small screen. Distorted-view approaches, such as fisheye techniques, have great potential to reduce these problems by representing detail within its larger context but introduce new issues of focus, transition between views and user disorientation from excessive distortion. We present a fisheyebased method which supports multiple focus points, enhances continuity through smooth transitions between views, and maintains location constraints to reduce the user’s sense of spatial disorientation. These are important requirements for the representation and navigation of networked systems in supervisory control applications. The method consists of two steps: a global allocation of space to rectangular sections of the display, based on scale factors, followed by degree-of-interest adjustments. Previous versions of the algorithm relied solely on relative scale factors to assign size; we present a new version which allocates space more efficiently using a dynamically calculated degree of interest. In addition to the automatic system sizing, manual user control over the amount of space assigned each area is supported. The amount of detail shown in various parts of the network is controlled by pruning the hierarchy and presenting those sections in summary form.

Capturing and supporting the analysis process

Visual analytics tools provide powerful visual representations in order to support the sense-making process. In this process, analysts typically iterate through sequences of steps many times, varying parameters each time. Few visual analytics tools support this process well, nor do they provide support for visualizing and understanding the analysis process itself. To help analysts understand, explore, reference, and reuse their analysis process, we present a visual analytics system named CzSaw (See-Saw) that provides an editable and re-playable history navigation channel in addition to multiple visual representations of document collections and the entities within them (in a manner inspired by Jigsaw [24]). Conventional history navigation tools range from basic undo and redo to branching timelines of user actions. In CzSaws approach to this, first, user interactions are translated into a script language that drives the underlying scripting-driven propagation system. The latter allows analysts to edit analysis steps, and ultimately to program them. Second, on this base, we build both a history view showing progress and alternative paths, and a dependency graph showing the underlying logic of the analysis and dependency relations among the results of each step. These tools result in a visual model of the sense-making process, providing a way for analysts to visualize their analysis process, to reinterpret the problem, explore alternative paths, extract analysis patterns from existing history, and reuse them with other related analyses.

A continuously variable zoom for navigating large hierarchical networks

We present the continuous zoom, a distorted view method for displaying hierarchically-organized, two-dimensional networks. The method is suitable for large networks, such as those found in interfaces to complex supervisory control systems. The continuous zoom shows detail in context, unlike simple pan and zoom techniques, and allows for more than one focus point, unlike fisheye viewing. The algorithm and variations are described.<<ETX>>

Haptic subdivision: an approach to defining level-of-detail in haptic rendering

Soft objects are often desired in applications such as virtual surgery training. Soft object simulations are computationally intensive because object deformation involves numerically solving a large number of differential equations. However, realistic force feedback requires deformation be computed fast and graphic feedback requires deformation be highly detailed. In this paper, we propose an approach that balances these requirements by subdividing the area of interest on a relatively coarse mesh model. Thus we keep the number of nodes of the model under control so that the simulation can be run at a sufficiently high rate for force feedback. The model we use is based on a mass-spring model. When a portion of the surface is subdivided, new values of mass and spring constants are determined such that computed force feedback offers the user the same reaction force as before subdivision.

System lag tests for augmented and virtual environments

We describe a simple technique for accurately calibrating the temporal lag in augmented and virtual environments within the Enhanced Virtual Hand Lab (EVHL), a collection of hardware and software to support research on goal-directed human hand motion. Lag is the sum of various delays in the data pipeline associated with sensing, processing, and displaying information from the physical world to produce an augmented or virtual world. Our main calibration technique uses a modified phonograph turntable to provide easily tracked periodic motion, reminiscent of the pendulum-based calibration technique of Liang, Shaw and Green. Measurements show a three-frame (50 ms) lag for the EVHL. A second technique, which uses a specialized analog sensor that is part of the EVHL, provides a “closed loop” calibration capable of sub-frame accuracy. Knowing the lag to sub-frame accuracy enables a predictive tracking scheme to compensate for the end-toend lag in the data pipeline. We describe both techniques and the EVHL environment in which they are used.

Expanding the Frontiers of Visual Analytics and Visualization

The field of computer graphics combines display hardware, software, and interactive techniques in order to display and interact with data generated by applications. Visualization is concerned with exploring data and information graphically in such a way as to gain information from the data and determine significance. Visual analytics is the science of analytical reasoning facilitated by interactive visual interfaces. Expanding the Frontiers of Visual Analytics and Visualization provides a review of the state of the art in computer graphics, visualization, and visual analytics by researchers and developers who are closely involved in pioneering the latest advances in the field. It is a unique presentation of multi-disciplinary aspects in visualization and visual analytics, architecture and displays, augmented reality, the use of color, user interfaces and cognitive aspects, and technology transfer. It provides readers with insights into the latest developments in areas such as new displays and new display processors, new collaboration technologies, the role of visual, multimedia, and multimodal user interfaces, visual analysis at extreme scale, and adaptive visualization.

On cutting and dissection of virtual deformable objects

Tissue dissection is an important procedure in surgical simulation systems. Dissection involves cutting through and separating the tissue after a cut. In this paper, we use a surface mass-spring model to simulate virtual dissection by progressive subdivision and re-meshing. We introduce novel algorithms to generate interior structures that show the cutting result generated by the interaction between instrument and model. In addition, a novel data structure for object representation after the cutting action is proposed which allows the original soft object to be divided and a portion manipulated away. The dissection environment can support a number of user interface devices which can manipulate different representation of virtual instruments. These techniques are being integrated into a training environment for both open and minimally invasive surgery.

Comparing CAVE, wall, and desktop displays for navigation and wayfinding in complex 3D models

Computer-aided design (CAD) and 3D visualization techniques are at the heart of many engineering processes such as aircraft, ship, and automobile design. These visualization tasks require users to navigate or wayfind through complex 3D geometric models consisting of millions of parts. Despite numerous studies, it remains unclear whether large-screen displays improve user performance for such activities. We present a user study comparing standard desktop, immersive room (i.e., CAVE), and wall displays with 3D stereo/head-tracking, and mono/no head-tracking. We observed individual differences between users and found that the presence of contextual structure greatly impacted performance, suggesting that providing structure and developing interaction techniques accommodating a wide range of users yields better performance than focusing on display characteristics alone

A study of level-of-detail in haptic rendering

This paper presents an initial study of an approach to reduce computational overhead in haptic rendering of physically based models. Haptic rendering refers to the notion of adding physical properties and behavior, specifically a sense of touch or force feedback, to models of objects. In this way, a user through a haptic feedback device can feel interaction forces while visually observing the objects. Physically based modeling is particularly important when representing deformable objects. In this paper, an approach based on a mass-spring damper system is used in modeling deformable objects. Deformation due to interaction forces is obtained by solving a set of differential equations, a process that is in general computationally demanding. To reduce this demand, the notion of level-of-detail in haptic rendering is introduced. Here the interplay between the graphical mesh and the haptic mesh as a function of various levels of subdivision is studied. The approach we describe is to adjust model parameters such that the user feels the same reaction force for a given deformation, regardless of the level of local subdivision.A preliminary user study with simple objects suggests there can be a local subdivision threshold such that the user cannot distinguish between global subdivision and the local subdivision introduced by the level-of-detail algorithm. This conclusion is beneficial for haptic rendering of deformable objects. Similar conclusions were obtained for haptic rendering of rigid objects. These results can be used as a guideline for other approaches to modeling deformable objects, such as finite element representations.