Affiliation:
1. Durham University, UK
2. University of Southampton
Abstract
Multiview auto-stereoscopic displays support both stereopsis and head motion parallax depth cues and could be superior for certain tasks. Previous work suggests that a high viewpoint density (100 views/10cm at the eye) is required to convincingly support motion parallax. However, it remains unclear how viewpoint density affects task performance, and this factor is critical in determining display and system design requirements. Therefore, we present a simulated multiview display apparatus to undertake experiments using a path-searching task in which we control two independent variables: the stereoscopic depth and the viewpoint density. In the first experiment, we varied both cues and found that even small amounts of stereo depth (2cm) reliably improved task accuracy and reduced latency, whereas there was no evidence of dependence on viewpoint density. In the second experiment, we switched off the stereoscopic cue and varied viewpoint density alone. We found that for these monoscopic images increasing viewpoint density resulted in some reduction in response latency (up to eight views/10cm) but had no effect on accuracy. We conclude for cases where occlusion is not an overriding factor that low viewpoint densities may be sufficient to enable effective path-searching task performance.
Publisher
Association for Computing Machinery (ACM)
Subject
Experimental and Cognitive Psychology,General Computer Science,Theoretical Computer Science
Cited by
5 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献