It is common in scientific visualization for one 2D scalar field variable to be colormapped and draped onto a shaded surface representing a second variable displayed as a height field. The experiments reported here investigate the properties that make a colormap suitable for this application; specifically, how much colormap accuracy is lost because of draping and shading as well as the degree to which the colormap impacts the perception of surface shape. A new task is used to evaluate surface shape perception; it involves study participants clicking on the peaks of hills and the bottoms of valleys in the display. The results provided surprisingly little support for the hypothesis that colormaps varying greatly in luminance impacted surface shape perception more than those that are close to isoluminant. Stepped colormaps proved to be more accurately read. However, stepped colormaps also interfered much more with the perception of surface shape. The implication of these results are that when selecting a colormap to be used for draping, careful consideration should be given to which of the two variables is more important, the one represented by the shaded surface, or the one which is colormapped and draped.

Abstract Visualizing 3D vector fields is challenging because of occlusion problems and the difficulty of providing depth cues that adequately support the perception of direction of flow lines in 3D space. One of the depth cues that has proven most valuable for the perception of other kinds of 3D data, notably 3D networks and 3D point clouds, is structure‐from‐motion (also called the Kinetic Depth Effect); another powerful depth cue is stereoscopic viewing. We carried out an experiment of the perception of direction for short streamlines passing through a cutting plane. The conditions included viewing with and without structure‐from‐motion and with and without stereoscopic depth. Conditions also include comparing streamtubes to lines. The results show that for this particular task, stereo provided an effective depth cue, but structure‐from‐motion did not. Ringed streamtubes and streamcones provided good 3D direction information, even without stereoscopic viewing. We conclude with guidelines for viewing slices through vector fields.

One advantage of touch interaction is the sense of direct manipulation; there is perhaps no more-intuitive interface than just reaching out and touching virtual entities. However, direct manipulation is generally limited to objects located on the 2D display surface. For 3D spaces extending behind or in front of a touchscreen, the direct manipulation metaphor quickly falls apart. In these cases, gestures are needed to convert 2D finger positions into 3D cursor positions. This paper presents the pantograph technique, a simple two-finger interaction method for positioning a 3D cursor within mono and stereoscopic applications. The pantograph's pseudomechanical linkage between fingers and cursor provides helpful depth cues and maintains the sense of direct manipulation. Extensions to the technique, which integrate selection and other advanced actions, are explored within the context of real-world visual analysis applications. A series of human factors experiments showed that, while the pantograph technique outperformed other similar multitouch 3D positioning techniques, multi-touch was still inferior to other traditional, non-touch-based interfaces for sustained 3D positioning tasks.

Pseudocoloring is one of the most common techniques used in scientific visualization. To apply pseudocoloring to a scalar field, the field value at each point is represented using one of a sequence of colors (called a colormap). One of the principles applied in generating colormaps is uniformity and previously the main method for determining uniformity has been the application of uniform color spaces. In this paper we present a new method for evaluating the feature detection threshold function across a colormap. The method is used in crowdsourced studies for the direct evaluation of nine colormaps for three feature sizes. The results are used to test the hypothesis that a uniform color space (CIELAB) will accurately model colormapped feature detection thresholds compared to a model where the chromaticity components have reduced weights. The hypothesis that feature detection can be predicted solely on the basis of luminance is also tested. The results reject both hypotheses and we demonstrate how reduced weights on the green-red and blue-yellow terms of the CIELAB color space creates a more accurate model when the task is the detection of smaller features in colormapped data. Both the method itself and modified CIELAB can be used in colormap design and evaluation.

Particle-based simulations are used across many science domains, and it is well known that stereoscopic viewing and kinetic depth enhance our ability to perceive the 3D structure of such data. But the relative advantages of stereo and kinetic depth have not been studied for point cloud data, although they have been studied for 3D networks. This article reports two experiments assessing human ability to perceive 3D structures in point clouds as a function of different viewing parameters. In the first study, the number of discrete views was varied to determine the extent to which smooth motion is needed. Also, half the trials had stereoscopic viewing and half had no stereo. The results showed kinetic depth to be more beneficial than stereo viewing in terms of accuracy and so long as the motion was smooth. The second experiment varied the amplitude of oscillatory motion from 0 to 16 degrees. The results showed an increase in detection rate with amplitude, with the best amplitudes being 4 degrees and greater. Overall, motion was shown to yield greater accuracy, but at the expense of longer response times in comparison with stereoscopic viewing.

A myriad of design rules for what constitutes a "good" colormap can be found in the literature. Some common rules include order, uniformity, and high discriminative power. However, the meaning of many of these terms is often ambiguous or open to interpretation. At times, different authors may use the same term to describe different concepts or the same rule is described by varying nomenclature. These ambiguities stand in the way of collaborative work, the design of experiments to assess the characteristics of colormaps, and automated colormap generation. In this paper, we review current and historical guidelines for colormap design. We propose a specified taxonomy and provide unambiguous mathematical definitions for the most common design rules.

Three-dimensional vector fields are common datasets throughout the sciences. Visualizing these fields is inherently difficult due to issues such as visual clutter and self-occlusion. Cutting planes are often used to overcome these issues by presenting more manageable slices of data. The existing literature provides many techniques for visualizing the flow through these cutting planes; however, there is a lack of empirical studies focused on the underlying perceptual cues that make popular techniques successful. This paper presents a quantitative human factors study that evaluates static monoscopic depth and orientation cues in the context of cutting plane glyph designs for exploring and analyzing 3D flow fields. The goal of the study was to ascertain the relative effectiveness of various techniques for portraying the direction of flow through a cutting plane at a given point, and to identify the visual cues and combinations of cues involved, and how they contribute to accurate performance. It was found that increasing the dimensionality of line-based glyphs into tubular structures enhances their ability to convey orientation through shading, and that increasing their diameter intensifies this effect. These tube-based glyphs were also less sensitive to visual clutter issues at higher densities. Adding shadows to lines was also found to increase perception of flow direction. Implications of the experimental results are discussed and extrapolated into a number of guidelines for designing more perceptually effective glyphs for 3D vector field visualizations.

Two experiments were conducted to test the hypothesis that animated representations of vector fields are more effective than common static representations even for steady flow. We compared four flow visualization methods: animated streamlets, animated orthogonal line segments (where short lines were elongated orthogonal to the flow direction but animated in the direction of flow), static equally spaced streamlines, and static arrow grids. The first experiment involved a pattern detection task in which the participant searched for an anomalous flow pattern in a field of similar patterns. The results showed that both the animation methods produced more accurate and faster responses. The second experiment involved mentally tracing an advection path from a central dot in the flow field and marking where the path would cross the boundary of a surrounding circle. For this task the animated streamlets resulted in better performance than the other methods, but the animated orthogonal particles resulted in the worst performance. We conclude with recommendations for the representation of steady flow patterns.

Abstract In many planning applications, a computational model is used to make predictions about the effects of management or engineering decisions. To understand the implications of alternative scenarios, a user typically adjusts one or more of the input parameters, runs the model, and examines the outcomes using simple charts. For example, a time series showing changes in productivity or revenue might be generated. While this approach can be effective in showing the projected effects of changes to the model's input parameters, it fails to show the mechanisms that cause those changes. In order to promote understanding of model mechanics using a simple graphical device, we propose dynamic change arcs. Dynamic change arcs graphically reveal the internal model structure as cause and effect linkages. They are signed to show both positive and negative effects. We implemented this concept using a species interaction model developed for fisheries management based on a system of Lotka‐Volterra equations. The model has 10 economically important fish species and incorporates both predation and competition between species. The model predicts that changing the catch of one species can sometimes result in changes in biomass of another species through multi‐step causal chains. The dynamic change arcs make it possible to interpret the resulting complex causal chains and interaction effects. We carried out an experiment to evaluate three alternative forms of arcs for portraying causal connections in the model. The results show that all linkage representations enabled participants to reason better about complex chains of causality than not showing linkages. However, none of them were significantly better than the others.

It is common in infographics for quantities to be represented by stacks of discrete blocks. For example, a magazine illustration showing automobile production in different countries might use stacks of blocks with each block representing a thousand cars. This is unlike what is done to represent quantity in the charts used by statisticians, or for quantitative glyphs used in maps. In these cases, solid bars or solid area glyphs such as circles are commonly used to represent quantity. This raises the question of whether breaking bars, area, or volume glyphs into discrete blocks can improve the rapid estimation of quantity. We report on a study where participants compared quantities represented using bar, area, and volume glyphs in both solid and discrete variants. The discrete variants used up to 4, 4 × 4, and 4 × 4 × 4 blocks or 10, 10 × 10, and 10 × 10 × 10 blocks for bar, area, and volume, respectively. The results show that people are significantly more accurate in estimating quantities using the discrete versions, but they take somewhat longer. For both areas and volumes, the accuracy gains were considerable.

Most professional wind visualizations show wind speed and direction using a glyph called a wind barb in a grid pattern. Research into flow visualization has suggested that streamlines better represent flow patterns but these methods lack a key property-unlike the wind barb, they do not accurately convey the wind speed. With the goal of improving the perception of wind patterns, and at least equaling the quantitative quality of wind barbs, we designed two variations on the wind barb and designed a new quantitative glyph. All of our new designs space glyph elements along equally spaced streamlines. To evaluate these designs, we used a North American mesoscale forecast model. We tested the ability of subjects to determine direction and speed using two different densities each of three new designs as well as the classic wind barb. A second experiment evaluated how effectively each of the designs represented wind patterns. The results showed that the new design is superior to the classic, but they also showed that the classic barb can be redesigned and substantially improved. We suggest that flow patterns with integrated glyphs may have widespread application in flow visualization.

We present a method for automatically evaluating and optimizing visualizations using a computational model of human vision. The method relies on a neural network simulation of early perceptual processing in the retina and primary visual cortex. The neural activity resulting from viewing flow visualizations is simulated and evaluated to produce a metric of visualization effectiveness. Visualization optimization is achieved by applying this effectiveness metric as the utility function in a hill-climbing algorithm. We apply this method to the evaluation and optimization of 2D flow visualizations, using two visualization parameterizations: streaklet-based and pixel-based. An emergent property of the streaklet-based optimization is head-to-tail streaklet alignment. It had been previously hypothesized the effectiveness of head-to-tail alignment results from the perceptual processing of the visual system, but this theory had not been computationally modeled. A second optimization using a pixel-based parameterization resulted in a LIC-like result. The implications in terms of the selection of primitives is discussed. We argue that computational models can be used for optimizing complex visualizations. In addition, we argue that they can provide a means of computationally evaluating perceptual theories of visualization, and as a method for quality control of display methods.

It has been previously proposed that understanding the mechanisms of contour perception can provide a theory for why some flow rendering methods allow for better judgments of advection pathways than others. In this article, we develop this theory through a numerical model of the primary visual cortex of the brain (Visual Area 1) where contour enhancement is understood to occur according to most neurological theories. We apply a two-stage model of contour perception to various visual representations of flow fields evaluated using the advection task of Laidlaw et al. In the first stage, contour enhancement is modeled based on Li's cortical model. In the second stage, a model of streamline tracing is proposed, designed to support the advection task. We examine the predictive power of the model by comparing its performance to that of human subjects on the advection task with four different visualizations. The results show the same overall pattern for humans and the model. In both cases, the best performance was obtained with an aligned streamline based method, which tied with a LIC-based method. Using a regular or jittered grid of arrows produced worse results. The model yields insights into the relative strengths of different flow visualization methods for the task of visualizing advection pathways.

Optimizing complex displays is difficult because there are many alternative ways of mapping the data to its graphical representation. In this paper we report on a study employing a method we call ¿interactive design space hill climbing¿. This involves first parameterizing the mapping from data to display. Next, using an interactive interface, designers and domain experts attempt to construct good designs by interactively changing parameters settings based on a random starting point. In our study we applied this method to the problem of 2D flow visualization: users adjusted 22 different sliders under each of 11 mappings to try to create an optimal display of a flow field from an ocean flow model. The results suggest that some variables should have settings in a narrow range. We conclude that employing designers with a human in the loop hill climbing interface can be a good overall solution for complex visual display designs in cases where a relatively simple parameterization is possible.

Representing bivariate scalar maps is a common but difficult visualization problem. One solution has been to use two dimensional color schemes, but the results are often hard to interpret and inaccurately read. An alternative is to use a color sequence for one variable and a texture sequence for another. This has been used, for example, in geology, but much less studied than the two dimensional color scheme, although theory suggests that it should lead to easier perceptual separation of information relating to the two variables. To make a texture sequence more clearly readable the concept of the quantitative texton sequence (QTonS) is introduced. A QTonS is defined a sequence of small graphical elements, called textons, where each texton represents a different numerical value and sets of textons can be densely displayed to produce visually differentiable textures. An experiment was carried out to compare two bivariate color coding schemes with two schemes using QTonS for one bivariate map component and a color sequence for the other. Two different key designs were investigated (a key being a sequence of colors or textures used in obtaining quantitative values from a map). The first design used two separate keys, one for each dimension, in order to measure how accurately subjects could independently estimate the underlying scalar variables. The second key design was two dimensional and intended to measure the overall integral accuracy that could be obtained. The results show that the accuracy is substantially higher for the QTonS/color sequence schemes. A hypothesis that texture/color sequence combinations are better for independent judgments of mapped quantities was supported. A second experiment probed the limits of spatial resolution for QTonSs.

It has been known for some time that larger graphs can be interpreted if laid out in 3D and displayed with stereo and/or motion depth cues to support spatial perception. However, prior studies were carried out using displays that provided a level of detail far short of what the human visual system is capable of resolving. Therefore, we undertook a graph comprehension study using a very high resolution stereoscopic display. In our first experiment, we examined the effect of stereoscopic display, kinetic depth, and using 3D tubes versus lines to display the links. The results showed a much greater benefit for 3D viewing than previous studies. For example, with both motion and stereoscopic depth cues, unskilled observers could see paths between nodes in 333 node graphs with less than a 10% error rate. Skilled observers could see up to a 1000-node graph with less than a 10% error rate. This represented an order of magnitude increase over 2D display. In our second experiment, we varied both nodes and links to understand the constraints on the number of links and the size of graph that can be reliably traced. We found the difference between number of links and number of nodes to best account for error rates and suggest that this is evidence for a “perceptual phase transition.” These findings are discussed in terms of their implications for information display.

Summary form only given. Current theories of perception and language can help answer the question, "What can we expect of a visual language?" There are fundamental differences between visual and linguistic forms of expression. Linguistic forms of expression are characterized by the use of a rich set of socially invented arbitrary symbols as well as a form of logic exemplified by the 'ifs', 'ands' and 'buts' of natural language. The grammar of visual representation is quite different having to do with pattern relationships such as 'connected', 'inside', 'outside', or 'part of. These derive from our visual processing mechanism developed through interaction with the visual environment. This means that certain kinds of logic are best left to verbal or written language whereas structural relationships are best conveyed visually. Narrative, however, can be carried either through language or purely visual techniques or both. This is because narrative is about leading the attention of an audience and this can be achieved through either modality. Nevertheless, what can be expressed within a narrative is very different depending on the modality. Language can convey complex logical relationships between abstract ideas and support conditional actions whereas visual media can support the perception of almost instantaneous scene gist, rapid explorations of spatial structure and relationships between objects as well as emotions and motivations. Both can maintain and hold the thread of audience attention which is the essence of narrative.

In order to investigate large information spaces effectively, it is often necessary to employ navigation mechanisms that allow users to view information at different scales. Some tasks require frequent movements and scale changes to search for details and compare them. We present a model that makes predictions about user performance on such comparison tasks with different interface options. A critical factor embodied in this model is the limited capacity of visual working memory, allowing for the cost of visits via fixating eye movements to be compared to the cost of visits that require user interaction with the mouse. This model is tested with an experiment that compares a zooming user interface with a multi-window interface for a multiscale pattern matching task. The results closely matched predictions in task performance times; however error rates were much higher with zooming than with multiple windows. We hypothesized that subjects made more visits in the multi-window condition, and ran a second experiment using an eye tracker to record the pattern of fixations. This revealed that subjects made far more visits back and forth between pattern locations when able to use eye movements than they made with the zooming interface. The results suggest that only a single graphical object was held in visual working memory for comparisons mediated by eye movements, reducing errors by reducing the load on visual working memory. Finally we propose a design heuristic: extra windows are needed when visual comparisons must be made involving patterns of a greater complexity than can be held in visual working memory.

This paper is a contribution to the literature on perceptually optimal visualizations of layered three-dimensional surfaces. Specifically, we develop guidelines for generating texture patterns, which, when tiled on two overlapped surfaces, minimize confusion in depth-discrimination and maximize the ability to localize distinct features. We design a parameterized texture space and explore this texture space using a "human in the loop" experimental approach. Subjects are asked to rate their ability to identify Gaussian bumps on both upper and lower surfaces of noisy terrain fields. Their ratings direct a genetic algorithm, which selectively searches the texture parameter space to find fruitful areas. Data collected from these experiments are analyzed to determine what combinations of parameters work well and to develop texture generation guidelines. Data analysis methods include ANOVA, linear discriminant analysis, decision trees, and parallel coordinates. To confirm the guidelines, we conduct a post-analysis experiment, where subjects rate textures following our guidelines against textures violating the guidelines. Across all subjects, textures following the guidelines consistently produce high rated textures on an absolute scale, and are rated higher than those that did not follow the guidelines.

This paper proposes a new experimental framework within which evidence regarding the perceptual characteristics of a visualization method can be collected, and describes how this evidence can be explored to discover principles and insights to guide the design of perceptually near-optimal visualizations. We make the case that each of the current approaches for evaluating visualizations is limited in what it can tell us about optimal tuning and visual design. We go on to argue that our new approach is better suited to optimizing the kinds of complex visual displays that are commonly created in visualization. Our method uses human-in-the-loop experiments to selectively search through the parameter space of a visualization method, generating large databases of rated visualization solutions. Data mining is then used to extract results from the database, ranging from highly specific exemplar visualizations for a particular data set, to more broadly applicable guidelines for visualization design. We illustrate our approach using a recent study of optimal texturing for layered surfaces viewed in stereo and in motion. We show that a genetic algorithm is a valuable way of guiding the human-in-the-loop search through visualization parameter space. We also demonstrate several useful data mining methods including clustering, principal component analysis, neural networks, and statistical comparisons of functions of parameters.

This paper reports the development of a real-time human-computer interface (HCI) system that enables a human operator to more effectively utilize the large volume of quantitative data (navigation, scientific, and vehicle status) generated in real-time by the sensor suites of underwater ro botic vehicles. The system provides interactive 3-D graphical interfaces that display, under user control, the quantitative spatial and temporal sensor data presently available to pilots and users only as two-dimensional plots and numerical dis plays. The system can presently display real-time bathymetric renderings of the sea-floor based upon vehicle navigation and sonar sensor data; vehicle trajectory data; a variety of scalar valued sensor data; and geo-referenced targets and waypoints. We report the accuracy of the real-time navigation and bathymetric sonar data processing by comparing the real-time sonar bathymetry of our test tank floor to a high-resolution laser scan of the same tank floor. The real-time sonar bathymetry is shown to compare favorably to the laser scan data

The problem of perceptually optimizing complex visualizations is a difficult one, involving perceptual as well as aesthetic issues. In our experience, controlled experiments are quite limited in their ability to uncover interrelationships among visualization parameters, and thus may not be the most useful way to develop rules-of-thumb or theory to guide the production of high-quality visualizations. In this paper, we propose a new experimental approach to optimizing visualization quality that integrates some of the strong points of controlled experiments with methods more suited to investigating complex highly-coupled phenomena. We use human-in-the-loop experiments to search through visualization parameter space, generating large databases of rated visualization solutions. This is followed by data mining to extract results such as exemplar visualizations, guidelines for producing visualizations, and hypotheses about strategies leading to strong visualizations. The approach can easily address both perceptual and aesthetic concerns, and can handle complex parameter interactions. We suggest a genetic algorithm as a valuable way of guiding the human-in-the-loop search through visualization parameter space. We describe our methods for using clustering, histogramming, principal component analysis, and neural networks for data mining. The experimental approach is illustrated with a study of the problem of optimal texturing for viewing layered surfaces so that both surfaces are maximally observable.

Language theorists argue that the reason why spoken language is acquired so rapidly is that we have an innate predisposition for understanding linguistic structures. Theories of perception also hold that there may be deeply seated mechanisms for decomposing visual objects and analyzing them into both component parts and the structural interrelationships of those parts. We propose the theory that diagrams that activate the mechanisms for structural object perception should be similarly easy to learn. This builds on previous work in which we have developed diagramming principles based on the theory of structural object perception. We call these geon diagrams. We have previously shown that such diagrams are easy to remember and to analyze. To evaluate our hypothesis that geon diagrams should also be easy to understand we carried out an empirical study to evaluate the learnability of geon diagram semantics in comparison with the well-established UML convention. The results support our theory of learnability. Both "novices" and "experts" found the geon diagram syntax easier to apply in a diagram-to-textual description matching task than the equivalent UML syntax.

Many different problems can be represented as graphs displayed in the form of node--link diagrams. However, when a graph is large it becomes visually uninterpretable because of the tangle of links. We describe a set of techniques that use motion in an interactive interface to provide effective access to larger graphs. Touching a node with the mouse cursor causes that node and the subgraph of closely connected nodes to oscillate. We argue from perceptual principles that this should be a more effective way of interactively highlighting a subgraph than more conventional static methods. The MEGraph system was developed to gain experience with different forms of motion highlighting. Based on positive feedback, three experiments were carried out to evaluate the effectiveness of motion highlighting for specific tasks. All three showed motion to be more effective than static highlighting, both in increasing the speed of response for a variety of visual queries, and in reducing errors. We argue that motion highlighting can be a valuable technique in applications that require users to understand large graphs.

It is common for 3D visualization systems to provide multiple points of view to a user, but there have been many solutions to the problem of linking these views so that users can understand the spatial relationships between them. Toward developing guidelines for view-linking devices, we have carried out two experiments that compare the utility of three different classes of linking devices: a directional proxy, tethers from one view to another, and a track-up map coupling. The task we apply them to is what we call the multi-perspective identification task: subjects are asked to identify an item seen in a local, forward-looking view in the context of a global, overhead view. Our results indicate that the directional proxy is the most beneficial device, and that the track-up map coupling is also beneficial. The results suggest that tethers provide little benefit. The results also suggest that when multiple local views are present, it may be beneficial to emphasize one window as being of primary interest.

The class of diagrams known collectively as node-link diagrams are used extensively for many applications, including planning, communications networks, and computer software. The defining features of these diagrams are nodes, represented by a circle or rectangle connected by links usually represented by some form of line or arrow. We investigate the proposition that drawing three-dimensional shaded elements instead of using simple lines and outlines will result in diagrams that are easier to interpret. A set of guidelines for such diagrams is derived from perception theory and these collectively define the concept of the geon diagram. We also introduce a new substructure identification task for evaluating diagrams and use it to test the effectiveness of geon diagrams. The results from five experiments are reported. In the first three experiments geon diagrams are compared to Unified Modeling Language (UML) diagrams. The results show that substructures can be identified in geon diagrams with approximately half the errors and significantly faster. The results also show that geon diagrams can be recalled much more reliably than structurally equivalent UML diagrams. In the final two experiments geon diagrams are compared with diagrams having the same outline but not constructed with shaded solids. This is designed to specifically test the importance of using 3D shaded primitives. The results also show that substructures can be identified much more accurately with shaded components than with 2D outline equivalents and remembered more reliably. Implications for the design of diagrams are discussed.

A common problem in visualization applications is the display of one surface overlying another. Unfortunately, it is extremely difficult to do this clearly and effectively. Stereoscopic viewing can help, but in order for us to be able to see both surfaces simultaneously, they must be textured, and the top surface must be made partially transparent. There is also abundant evidence that all textures are not equal in helping to reveal surface shape, but there are no general guidelines describing the best set of textures to be used in this way. What makes the problem difficult to perceptually optimize is that there are a great many variables involved. Both foreground and background textures must be specified in terms of their component colors, texture element shapes, distributions, and sizes. Also to be specified is the degree of transparency for the foreground texture components. Here we report on a novel approach to creating perceptually optimal solutions to complex visualization problems and we apply it to the overlapping surface problem as a test case. Our approach is a three-stage process. In the first stage we create a parameterized method for specifying a foreground and background pair of textures. In the second stage a genetic algorithm is applied to a population of texture pairs using subject judgments as a selection criterion. Over many trials effective texture pairs evolve. The third stage involves characterizing and generalizing the examples of effective textures. We detail this process and present some early results.

Zooming and multiple windows are two techniques designed to address the focus-in-context problem. We present a theoretical model of performance that models the relative benefits of these techniques when used by humans for completing a task involving comparisons between widely separated groups of objects. The crux of the model is its cognitive component: the strength of multiple windows comes in the way they aid visual working memory. The task to which we apply our model is multiscale comparison, in which a user begins with a known visual pattern and searches for an identical or similar pattern among distracters. The model predicts that zooming should be better for navigating between a few distant locations when demands on visual memory are low, but that multiple windows are more efficient when demands on visual memory are higher, or there are several distant locations that must be investigated. To evaluate our model we conducted an experiment in which users performed a multiscale comparison task using both zooming and multiple-window interfaces. The results confirm the general predictions of our model.

Perceptual factors that affect monocular, transparent (a.k.a "see-thru") head-mounted displays include binocular rivalry, visual interference, and depth of focus. We report the results of an experiment designed to evaluate the effects of these factors on user performance in a table look-up task. Two backgrounds were used. A dynamic moving background was provided by a large screen TV and an untidy bookshelf was used to provide a complex static background. With the TV background large effects were found attributable to both rivalry and visual interference. These two effects were roughly additive. Smaller effects were found with the bookshelf. In conclusion we suggest that monocular transparent HMDs may be unsuitable for use in visually dynamic environments. However when backgrounds are relatively static, having a transparent display may be preferable to having an opaque display.

Most diagrams, particularly those used in software engineering, are line drawings consisting of nodes drawn as rectangles or circles, and edges drawn as lines linking them. In the present paper we review some of the literature on human perception to develop guidelines for effective diagram drawing. Particular attention is paid to structural object recognition theory. According to this theory as objects are perceived they are decomposed into 3D set of primitives called geons, together with the skeleton structure connecting them. We present a set of guidelines for drawing variations on node-link diagrams using geon-like primitives, and provide some examples. Results from three experiments are reported that evaluate 3D geon diagrams in comparison with 2D UML (Unified Modeling Language) diagrams. The first experiment measures the time and accuracy for a subject to recognize a sub-structure of a diagram represented either using geon primitives or UML primitives. The second and ...

The term Eye-hand co-ordination refers to hand movements controlled with visual feedback and reinforced by hand contact with objects. A correct perspective view of a virtual environment enables normal eye-hand co-ordination skills to be applied. But is it necessary for rapid interaction with 3D objects? A study of rapid hand movements is reported using an apparatus designed so that the user can touch a virtual object in the same place where he or she sees it. A Fitts tapping task is used to assess the effect of both contact with virtual objects and real-time update of the centre of perspective based on the user's actual eye position. A Polhemus tracker is used to measure the user's head position and from this estimate their eye position. In half of the conditions, head tracked perspective is employed so that visual feedback is accurate while in the other half a fixed eye-position is assumed. A Phantom force feedback device is used to make it possible to touch the targets in selected conditions. Subjects were required to change their viewing position periodically to assess the importance of correct perspective and of touching the targets in maintaining eye-hand co-ordination, The results show that accurate perspective improves performance by an average of 9% and contact improves it a further 12%. A more detailed analysis shows the advantages of head tracking to be greater for whole arm movements in comparison with movements from the elbow.

Times for virtual object rotations reported in the literature are of the order of 10 seconds or more and this is far longer than it takes to manually orient a “real” object, such as a cup. This is a report of a series of experiments designed to investigate the reasons for this difference and to help design interfaces for object manipulation. The results suggest that two major factors are important. Having the hand physically in the same location as the virtual object being manipulated is one. The other is based on whether the object is being rotatted to a new, randomly determined orientation, or is always rotated to the same position. Making the object held in the hand have the same physical shape as the object being visually manipulated was not found to be a significant factor. The results are discussed in the context of interactive virtual environments.

This paper reports on an experimental approach to adjusting stereo parameters automatically and thereby providing a low eye strain, easily accommodated stereo view for computer graphics applications. To this end, the concept of virtual eye separation is defined. Experiment 1 shows that dynamic changes in virtual eye separation are not noticed if they occur over a period of a few seconds. Experiment 2 shows that when subjects are given control over their virtual eye separation, they change it depending on the amount of depth in the scene. Based partly on these results, an algorithm is presented for enhancing stereo depth cues for moving computer generated 3D images. It has the effect of doubling the stereo depth in flat scenes and limiting the stereo depth for deep scenes. It also reduces the occurrence of double images and the discrepancy between focus and vergence. The algorithm is applied dynamically in real time with an optional damping factor applied so the disparities never change too abruptly. Finally, Experiment 3 provides a qualitative assessment of the algorithm with a dynamic "flight" over a digital elevation map.

In fish tank VR environments, the screen is used as a window into a virtual environment. This effectively creates a useful 3D workspace in the vicinity of the monitor screen, near to the user. However, many geographical applications require the user to cover large virtual distances and to support this a flying interface is often provided. The authors describe a method for combining the flying and fish tank metaphors to create a practical working environment. The central insight developed in the paper is that a geometric transformation that they call the "cyclopean scale" enables the simple combination of flying and fish tank VR interaction metaphors. Cyclopean scale continuously scales the working environment to lie just behind the screen in terms of stereoscopic depth. Cyclopean scale allows for fish tank VR viewing and also places objects at a convenient distance for manipulation, optimizes stereo display parameters and reduces stereo display problems (vergence focus conflict). They have implemented this technique in a system called Fledermaus VR with a cable route editing task. This is an application for planning the layout of submarine cables.

The requirement to change scale frequently is common to many 2D and 3D applications.Users must "zoom in" to examine details and "zoom out" to appreciate the context.This presents a problem in the context of "fly by" interfaces that use a flying metaphor to enable the user to change the point of view and explore a data space.The problem is that radical changes in velocity sensitivity may needed when working at different scrdes.A method is described that uses continous depth sampling to modulate the flying speed.The distribution of depths in the current frame of animation is used to set the Device to Control ratio so that it is always comfortable when operating over a range of scales.This is called Depth Modulated Flying (DMF).A family of related methods are evaluated in a task that requires subjects to search for small targets in a scene.The results show that scaling the velocity control by the near point in the scene and by the average point in the scene are equally effective.

This study investigates the use of a 2D cursor presented to one eye for target selection in Fish Tank VR and other stereo environments. It is argued that 2D selection of 3D objects should be less difficult than 3D selection. Vision research concerning binocular rivalry and the tendency we have to project images onto surfaces suggests that this mode of viewing will not seem particularly unnatural. A Fitt's Law experiment was done to directly compare target acquisition with a one-eyed 2D cursor and target acquisition using a 3D cursor. In both cases we used the same input device (Polhemus Fastrak) so that the device lag and gain parameters were exactly matched. The results show a large improvement in target acquisition time using the 2D cursor. The practical implications of this is that the 2D selection method using a one-eyed cursor in preferable to the 3D selection method. Theoretical implications relate to methods for extending Fitts' Law from the one-dimensional task for which it was designed to 2D and 3D tasks. We conclude that the existing approaches to this problem are not adequate.

This article concerns the benefits of presenting abstract data in 3D. Two experiments show that motion cues combined with stereo viewing can substantially increase the size of the graph that can be preceived. The first experiment was designed to provide quantitiative measurements of how much more (or less) can be understood in 3D than in 2D. The 3D display used was configured so that the image on the monitor was coupled to the user's actual eye positons (and it was updated in real-time as the user moved) as well as bring in stereo. Thus the effect was like a local “virtual reality” display located in the vicinity of the computer monitor. The results from this study show that head-coupled stereo viewing can increase the size of an abstract graph that can be understood by a factor of three; using stereo alone provided an increase by a factor of 1.6 and head coupling along produced an increase by a factor of 2.2. The second experiment examined a variety of motion cues provided by head-coupled perspective (as in virtual reality displays), head-guided motion and automatic rotation, respectively, both with and without stereo in each case. The results show that structured 3D motion and stereo viewing both help in understanding, but that the kind of motion is not particularly important; all improve performance, and all are more significant than stereo cues. These results provide strong reasons for using advanced 3D graphics for interacting with a large variety of information structures.

Based on a review of the facts about human stereo vision, a case is made that the stereo processing mechanism is highly flexible. Stereopsis seems to provide only local additional depth information, rather than defining the overall 3D geometry of a perceived scene. New phenomenological and experimental evidence is presented to support this view. The first demonstration shows that kinetic depth information dominates stereopsis in a depth cue conflict. Experiment 1 shows that dynamic changes in effective eye separation are not noticed if they occur over a period of a few seconds. Experiment 2 shows that subjects who are given control over their effective eye separation, can comfortably work with larger than normal eye separations when viewing a low relief scene. Finally, an algorithm is presented for the generation of dynamic stereo images designed to reduce the normal eye strain that occurs due to the mis-coupling of focus and vergence cues.

Results from vision research are applied to the synthesis of visual texture for the purposes of information display. The literature surveyed suggests that the human visual system processes spatial information by means of parallel arrays of neurons that can be modeled by Gabor functions. Based on the Gabor model, it is argued that the fundamental dimensions of texture for human perception are orientation, size (1/frequency), and contrast. It is shown that there are a number of trade-offs in the density with which information can be displayed using texture. Two of these are (1) a trade-off between the size of the texture elements and the precision with which the location can be specified, and (2) the precision with which texture orientation can be specified and the precision with which texture size can be specified. Two algorithms for generating texture are included.

This article reports the results from three experimental studies of reaching behavior in a head-coupled stereo display system with a hand-tracking subsystem for object selection. It is found that lag in the head-tracking system is relatively unimportant in predicting performance, whereas lag in the hand-tracking system is critical. The effect of hand lag can be modeled by means of a variation on Fitts' Law with the measured system lag introduced as a multiplicative variable to the Fitts' Law index of difficulty. This means that relatively small lags can cause considerable degradation in performance if the targets are small. Another finding is that errors are higher for movement in and out of the screen, as compared to movements in the plane of the screen, and there is a small (10%) time penalty for movement in the Z direction in all three experiments. Low frame rates cause a degradation in performance; however, this can be attributed to the lag which is caused by low frame rates, particularly if double buffering is used combined with early sampling of the hand-tracking device.

Fish tank virtualreality" refers to the use of a standard graphics workstation to achieve real-time display of 3D scenes using stereopsis and dynamic head-coupled perspective.Fish tank VR has a number of advantages over head-mounted immersion VR which makes it more practical for many applications.After discussing the characteristics of fish tank VR, we describe a set of three experiments conducted to study the benefits of fish tank VR over a traditional workstation graphics display.These experiments tested user performance under two conditions: (a) whether or not stereoscopic display was used and (b) whether or not the perspective display was coupled dynamically to the positions of a user's eyes.Subjects using a comparison protocol consistently preferred head coupling without stereo over stereo without head coupling.Error rates in a tree-tracing task similar to one used by Sollenberger and Milgram showed an order of magnitude improvement for head-coupled stereo over a static (nonhead-coupled) display, and the benefits gained by head coupling were more significant than those gained from stereo alone.The final experiment examined two factors that are often associated with human performance in virtual worlds: the lag (or latency) in receiving and processing tracker data arid the rate at which frames are updated.For the tree-tracing task, lag had a larger impact on performance than did frame update rate, with lag having a multiplicative effect on response time.We discuss the relevance of these results for the display of complex 3D data and highlight areas requiring further study, Categories and Subject Descriptors: 1.

Vision research relating to the human perception of texture is briefly reviewed with a view to arriving at the principal dimensions of visual texture useful for data display. The conclusion is that orientation, size (1/spatial frequency), and contrast (amplitude) are the primary orderable dimensions of texture. Data displayed using these texture parameters will be subject to similar distortions to those found when color is used. Textures synthesized using Gabor function primitives can be modulated along the three primary dimensions. Some preliminary results from a study using Gabor functions to modulate luminance are presented which suggest that: perceived texture size difference are approximately logarithmic, a 5% change in texton size is detectable 50% of the time, and large perceived size differences do not predict small (just noticeable) size differences.

This paper introduces a predictor based visual feedback aid for navigating through virtual environments using velocity control. The predictor indicates to the user where and how fast he or she is travelling and has a direct manipulation feel to it. Experiences using the predictor to navigate over digital terrain maps are discussed, which show it to be an aide in learning to use velocity control and in creating smooth flight paths over thinned wire frame representation of a scene for subsequent single frame animation. Measurements of performance in using the predictor to fly through a tube scene show a benefit for the less experienced users.

This report describes an experiment designed to test the use of simple linear motion as an attention‐getting device for computer displays. The experiment involved two tasks: a primary task that required the typed transcription of a document onto a computer screen, and a secondary task that involved detecting and responding to a moving icon signal. The icon consisted of a rectangular bar that grew and shrank in an oscillatory fashion, as the top edge ascended and descended. The amplitude and the velocity of motion were varied systematically and the effect on response time was measured. The results from this secondary task show that there is an inverse relation between the velocity of the moving icon and the response time. No effect was found for amplitude. The speed of the responses suggests that simple motion is an effective attention‐getting device for events in the periphery of the visual field.

This paper evaluates three distinct metaphors for exploration and virtual camera control in virtual environments using a six degree of freedom input device. The metaphors are "eyeball in hand", "scene in hand", and "flying vehicle control". These metaphors have been implemented and evaluated using an IRIS workstation and a Polhemus 3Space. The system has the capability to record the motion path followed during an exploration session and this can be recorded and played back to create a movie. Evaluation is through intensive structured interview sessions wherein subjects are required to complete a number of tasks involving three different "toy" environments. None of the metaphors is judged the best in all situations, rather the different metaphors each have advantages and disadvantages depending on the particular task. For example, "scene in hand" is judged to be good for manipulating closed objects, but is not good for moving through an interior; whereas "flying vehicle control" is judged the best for navigating through the interior, but is poor for moving around a closed object.

Background: The gamut of a color CRT is defined by its three primary colors, each produced by a phosphor/electron gun combination. Light from the primaries combines additively, so the color gamut is a subset of a three dimensional vector space [1]. With the primaries as basis vectors normalized to 1.0, the color gamut is a unit cube, known as the RGB color geometry, since the three primaries are usually red, green, and blue. User interaction via RGB is generally thought to be counterintuitive, and transformations of RGB, such as Smith's HSV geometry [10] which is derived from centuries old artists' models [2], are more popular. More recent color theories, based on psychophysical and physiological models of early visual processing, suggest that more intuitive geometries may be possible. The RGYB geometry is based on two recent discoveries about the human visual system. First, the three color signals from the cone receptors are organized into three opponent channels [1, 7]. A single achromatic channel indicates lightness or brightness. Two chromatic channels, red/green and yellow/blue, signal the chromatic quantities. Second, signals on the achromatic channel are easily distinguishable from signals on the chromatic ones [6]. Consequently, it is usual to represent colors as a set of surfaces of colors that vary in chromaticity, each at a different level of brightness. Examples are as diverse as CIE chromaticity coordinates, the CIELUV uniform color space, the Munsell color system, and computer graphics color spaces such as HSV and HLS [10, 12].

A geometry is described for converting hand position and orientation into six useful variables for computer input. The application is that of controlling form and color in an experimental computer “paint” program. We find that the most easily controlled parameters of hand placement are x,y and z cartesian coordinates and a twist parameter which approximates the wrist action that occurs when a dial is turned.

Since humans direct their visual attention by means of eye movements, a device which monitors eye movements should be a natural “pick” device for selecting objects visually present on a monitor. The results from an experimental investigation of an eye tracker as a computer input device are presented. Three different methods were used to select the object looked at; these were a button press, prolonged fixation or “dwell” and an on screen select button. The results show that an eye tracker can be used as a fast selection device providing that the target size is not too small. If the targets are small speed declines and errors increase rapidly.