Yi Wang's Research -- Gaps Between Displays

The Concept of Gaps Between Displays (BIP)

The Tasks in CAVE

Experiment 1

Through the usage of a four-screen CAVE™ system and a tiled large display system with bezels, we noticed that one immersion component, the gap between displays, may affect user performance, but cannot be covered by the
concept of FOR defined in the book 3D User Interfaces: Theory and Practice. We define the gap between displays (GBD) as an area that does not show the virtual world, and lies between multiple displays that show the virtual world. As illustrated in the figure on the right, given a fixed FOR, we can still vary the size of the gap between the two displays or the real world stimuli in the gap. The open back wall on a four-screen CAVE is an example of GBD. We can vary the strength of the real world stimuli in the GBD without changing the FOR.

We lose several aspects of information when GBD exists:
1) visual cues about the size of this gap in virtual world, e.g. color, depth...
2) references for judging the virtual objects’ motion near this gap
3) visual cues that help the user to understand the virtual world on two displays as a whole.
Theoretically, lack of the above information may hinder task performance in some cases. Understanding GBD may help us effectively configure displays for VE and augmented reality systems to reduce this effect. This leads to several specific research questions:
1) Does the gap between displays affect task performance and usability of the system? What kinds of tasks are likely to be affected? How severe is this effect?
2) What is the best way to fill in the gap for different tasks? Should we reduce real world stimuli or not?
3) What is the tradeoff between GBD and FOR for different tasks? Is there a
threshold that indicates the minimum gap size that does not affect users’ performance?
We hypothesized that GBD may reduce users’ task performance in some cases, and choosing appropriate stimuli in GBD can alleviate its negative effects.

The following figure shows the distractions on the CAVE's missing wall.

Exp. 1: Fill GBD with what stimuli?

Using a four-screen CAVE™ as the experiment platform, we carefully designed three tasks that cause different levels of awareness of GBD. Using these tasks, we compared users’ performance and preference in three conditions. The low-stimulus GBD condition was created by hanging a black cloth across the missing back wall of a CAVE. The high-stimulus GBD condition was created by projected animations and real human motion outside the CAVE. The no-GBD condition was created by a normal usage of the front CAVE wall. The experiment results confirmed our hypotheses and showed that keeping some level of real world distraction may sometimes be better than no distraction at all.

Conclusion:
Our experiment showed that the level of distraction has a measurable effect on users’ performance for at least some tasks. The relationship between the level of distraction and users’ performance is more subtle than our initial hypothesis.

Publications:
Yi Wang, Kunmi Otitoju, Tong Liu, Sijung Kim, and Doug A. Bowman, “Evaluating the effects of real world distraction on user performance in virtual environments.” In Proceedings of the ACM Symposium on Virtual Reality Software and Technology. ACM Press, pages19-26, 2006.

The following two figurs explain the "gestination prediction" task.

Exp. 2: How Severe is GBD?

After the first experiment, we formalized the “real world
distraction” problem into the GBD problem. We noticed that
one essential question remained unclear: how severe is the
GBD effect? To answer this question we need to compare the
GBD condition with a baseline condition, where no GBD exists between the two side walls. We can create this condition by letting the user face the front wall, where the virtual world will be displayed; we call this the no-gap condition. This experiment also helps to test the validity of the two tasks. If the task is sensitive to different GBD conditions, there should be a significant difference in user performance between the black cloth and no-gap conditions. We used the gap estimation task and the destination prediction task because we saw performance differences in the previous experiment for these two tasks.

Conclusion:

We did not find that reducing the level of distraction with a
low-stimulus area consistently increases or decreases user
performance. Rather, the effect depends on the specific task
and the specific VE scenario. For better spatial perception, it may be better to fill in the gap with an appropriate amount of stimuli than to remove all stimuli.

Publications:

Yi Wang, Kunmi Otitoju, Tong Liu, Sijung Kim, and Doug A. Bowman, “The Effect of Gaps Between Displays on Spatial Perception and Cognition Tasks in Virtual Environments", The International Journal of Virtual Reality, vol. 6, no. 2, 2007

The following two figurs explain the "Crayoland Visit" task.