Abstract
The angle of declination from the horizon to a location on the ground plane is considered to be an effective absolute distance cue when scaled by eye height. We manipulated this visual cue in two different ways to determine if the two different manipulations had similar effects on distance judgments in a HMD-based virtual environment. We measured distance judgments using a direct blind walking task to targets on the floor at 3 to 6 m. In the first manipulation, the image displayed in the HMD was shrunk by 30%. This has the effect of reducing the visual angle between the implicit horizon defined by perspective cues in the image and locations of interest on the floor. In the second manipulation, the virtual environment was tilted up 5.7 degrees, reducing the angle between gravity-indicated level and locations of interest on the floor. The amount of tilt was chosen to approximately match the angular change due to making the image smaller. For correctly displayed images, distance judgments in HMD-based virtual environment are substantially compressed. Shrinking the image led to distance judgments that were 98% of the intended distance, in contrast to 80% in the control condition, (F(1,21)=5.94, p[[lt]]0.05). For the tilted condition, subjects' distance judgments were 84% of the actual distance and did not differ significantly from the control condition (F(1,21)=0.03, p=0.86). This work points to the importance of considering the frame of reference for the angle of declination distance cue. More generally, it suggests that some manipulations of perspective information for depth may have greater effects than others. The results also have implications for how displayed images might be modified to make spatial judgments in HMD-based virtual environments more accurate.
This work was supported by NSF grants 0121084 and 0745131.