M. Bailey, T. Rebotier, an D. Kirsh (USA)
Perception, three dimensions, shadows, stereopsis, virtualreality, augmented reality.
The goal of three-dimensional visualization is to present
information in such a way that the viewer suspends disbelief and
uses the screen imagery the same way as he or she would use an
identical, real 3D scene. To do this effectively, programmers
employ a variety of 3D depth cues. Our own anecdotal
experience says that shadows and stereopsis are two of the best
for visualization. The nice thing is that both of these are
possible to do in interactive programs. They sacrifice a certain
amount of interactive speed, but they are possible.
But, there is very little information detailing exactly what these
cues add to the perception process. The purpose of this project
was to quantify how worthwhile using these two depth cues are,
that is, is it worth losing interactivity to get them? Using a large
number of student subjects, we performed a series of depth-test
trials and analyzed the results. Finally, as an upper-bound
control on these experiments, we also ran subject trials on
physically fabricated 3D objects, viewing them through a
pinhole in a controlled-lighting situation to factor out both
shadows and stereopsis, leaving only focal accommodation.
This paper shows the design of the experiments and the results
expressed in reaction times and error rates. The results have a
significant bearing on the design of 3D interactive visualization
systems, particularly those that use virtual or augmented reality.