Most information designers have few problems attaching text labels and other secondary information to graphic objects rendered in two-dimensional (2D) format. The real challenge appears when such information needs to be associated with objects in a three-dimensional (3D) virtual world, where the viewer can change an object’s degree of detail and angle at will.

The authors researched the challenge presented by an information design for an information-rich virtual environment (IRVE). An IRVE’s display spaces may consist of different screen sizes and resolutions, along with visual configurations that combine a virtual object with its related information, such as name, function, attributes, links, numbers, graph plots, and even video or audio annotations. The 3D information designer is presented with a complex problem--that of understanding the relationships existing between visibility, legibility, and data associations for the labels associated with their objects in a virtual space.

Polys, Kim, and Bowman, pursuing research involving interface designs for IRVEs, found little research about how different display presentations affect user performance. They wanted to understand how search and comparison tasks involving 3D objects and their label information were provided, and how users made choices under different display environments. Their research experiment was designed to evaluate the various tradeoffs needed to achieve the most accurate, user-satisfying visual configurations. The authors’ investigation involved how and where, within a virtual environment, information is displayed, and how specific label information is associated with its referenced object. The experiment was designed to answer these questions: Is a layout space with ensured visibility better than one where the object and its labeled annotations are tightly linked? Are the advantages of an explicitly sized and located layout space retained when screen size is increased? Do the advantages of the explicit layout space still hold when the software field of view (SFOV) is increased?

The researchers considered the roles played by a display’s resolution and physical size when deciding how much information can or should be shown. They examined the problems of layout variables (such as text size and readability) between those shown on a user’s desktop computer, compared to those displayed on a large, nine-screen display used for shared and collaborative work. Two layout spaces were used: the object space, in which labels were tightly coupled to a referent object, and the viewport space, where the labeling was rendered separately from the referent object. In all cases, a visual white line connected a label to its object. The researchers limited their study to the cues provided by these two layout spaces and their tradeoffs for performing basic task types across different monitor configurations and projection distortions: where the user needs to retrieve information about a given object in the virtual space, and where the user, by selecting specific abstract data, would display the corresponding desired objects in the virtual space.

After gaining information from a pilot study, the researchers designed their experiment. Sixteen participants, from the College of Engineering, were selected, who were very familiar with computers and had a high school level or higher course on cell biology. The study environment was based on a 3D model of a cell and its component parts. Each cell part was labeled with a callout that included its name, formula, and molecular weight. Within the object space, whose labels are located in the same world as their object, the labels magnify and distort along with the object being viewed. In the viewport space, where labels are viewed in a separate space, a BorderLayoutManager was used to define and constrain the textual information. These spaces are described in detail in the paper.

Users were asked to find a piece of abstract information: what molecule is just outside the nucleus? Then, they were asked to find a piece of perceptual/spatial information based on abstract label information: where in the cell is the molecule with a molecular weight of 85.15? Next, they were instructed to make a comparison using some spatial criteria, and to determine an abstract attribute. Finally, they were asked to make a comparison using some abstract label criteria, and to determine a spatial attribute.

When the results were collected, researchers found that at the start most users tended to search sequentially through the space. Some users tended to select, read, and deselect objects at first as they moved through the space, rather than keeping them visible at all times. As the users gained experience through repeated searches, they changed their strategy, leaving selected labels visible until they were a distraction. Overall, the viewport space layout was more satisfying to the users, who performed significantly better than those using the object space. The nine-screen display user process was hampered because of text size and distortion through scaling. However, this group found that tight coupling between object and label under the object space interface was more effective for comparison and accuracy. Overall, though, results showed that label information visibility was more important than a tied association with the object. Experiment evaluations indicated that, depending on the display condition, both object and viewport layouts could be used successfully. They also offered criteria to help guide the future of 3D design interfaces for the design of portable label information across different display sizes and distortions.