Mission Impossible Spaces
Using Challenge-Based Distractors to Reduce Noticeability of Self-Overlapping Virtual Architecture
5th-semester bachelor's project at Aalborg University Copenhagen. This is the first project where our new advisor at the time introduced us to VR locomotion and redirection research. We immediately resonated with his presentation that walking is the most natural and preferred way to traverse environments. However, one challenge is that virtual environments are much larger than the available physical space. Therefore, researchers are investigating techniques that enable users to walk in large virtual environments despite limited physical space.
We investigated overlapping architechture that leverages change blindness. While the user is transitioning from one room to another, the wall between the adjacent rooms can shift, creating larger rooms that use the same physical space. "Whodunnit" challenge is the perfect illustration of how people miss the changing environment.​
The main goal of this technique is to make impossible spaces seem possible by ensuring the user does not detect the manipulation. Researchers design studies to determine how much manipulation users can tolerate. Participants walk between the rooms and answer whether they perceived the space as impossible. Testing with a range of overlap levels lets us calculate detection thresholds. The threshold can later serve as a guideline for application design.
We identified a research gap, which motivated us to conduct our own study. The detection thresholds reported so far were measured in controlled scenarios in which users walked between two rooms for no other purpose. However, in VR applications the user might have multiple tasks and other type of distractors. Our hypothesis was that higher distraction and mental preoccupation would make it harder to detect the impossible space. To investigate the influence of distractors, we gamified the experiment as an escape room.
Participants had to visit two jail cells to find randommly scaterred numbers for a keycode. Using this code they could escape to the next floor. Each floor, in randomozed order, presented a level of overlap from 0 to 75 percent.
We theorized that one big clue for identifying impossible spaces is distance to walk between the two rooms. Therefore, we designed an environmental distractor during the walking transition. A security camera that rotates to observe the jail cells. The participants had to time their exits and entries to the next cell when the camera was briefly facing the other way.
The rooms could be overallped by up to 63% with the camera distractor, compared to only 36% without it. These findings reveal that overlapping architecture can be applied more liberally to VR applications with higher user engagement.