June 2011
Spotlight Summary by Laurie Wilcox
Depth rendition of three-dimensional displays
Vision science has investigated and documented the extremely fine resolution of the stereoscopic system. The assessment of depth magnitude, as documented here, has also received considerable attention in the literature. Virtually all experiments of this type have used simple lines or dots as stimuli and relatively small displays (computer monitors, oscilloscopes). The fact that observers in Westheimer’s study required more disparity than predicted from the geometric analysis is consistent with other work using disparity-defined cylinders. Westheimer raises the possibility that the apparent foreshortening is due to a mis-estimation of the distance to the display (another topic discussed in other work). Another potentially important factor in interpreting experiments of this type is that the human visual system is rarely presented with objects whose depth is defined by disparity alone. Arguably, the lack of supporting depth information from other sources (such as perspective, size, and occlusion) may make the depth from disparity more susceptible to errors in distance estimation.
Experiments like the ones outlined in this study are an important start toward correctly predicting the amount of depth that will be perceived in 3D displays. As discussed by Westheimer, the use of complex naturalistic scenes that contain multiple depth cues, including movement, may also prove critical in future experiments. The inclusion of multiple depth cues could influence the depth perceived from stereopsis in many ways. For instance, the presence of additional depth cues such as perspective and occlusion may help counter distortions in perceived depth from stereopsis.
To fully understand and anticipate the amount of depth obtained via stereopsis in 3D content, it will also be necessary to consider the effects of camera optics, and separation, on the resulting footage. For example, as noted by Westheimer, the separation of the eyes in the head (see “ac”in Fig. 1) determines the amount of disparity in a given stimulus configuration. The average human interocular separation is 65 mm, so one might expect that to obtain veridical depth percepts from 3D imagery one should replicate this separation in the camera configuration. However, in filmmaking it is not unusual to use camera separations as small as 10 mm, particularly when filming indoor scenes. The effects of differences like these between the twin optics of the human visual system and the 3D cameras are important but not necessarily obvious.
The challenges for scientists attempting to bridge the gap between laboratory experimentation and depth percepts from stereopsis in 3D displays are significant. Depth distortions may arise at all stages, including capture, processing, and display. Thus, a complete understanding of the geometry of stereopsis is essential, but insufficient. As the author has demonstrated, in the absence of additional cues to depth the amount of depth perceived from simple line stimuli is not fully consistent with geometric predictions. The use of more complex scenes containing multiple depth cues will undoubtedly influence the depth perceived via stereopsis. Advances in 3D display technology are making such experiments viable and opening exciting avenues of collaboration between science and industry.
Add Comment
You must log in to add comments.
Article Information
Depth rendition of three-dimensional displays
Gerald Westheimer
J. Opt. Soc. Am. A 28(6) 1185-1190 (2011) View: Abstract | HTML | PDF