You are here: Home Research Activities Perceptually-based Real-time Selective Rendering Algorithms

Research on Perceptually-based Real-time Selective Rendering Algorithms

Introduction

In a Virtual Environment (VE), efficient techniques are often needed to economize on rendering computation without compromising the information transmitted. This research activity aims to devise a functional fidelity metric by exploiting research on memory ‘schemata’ related to the context of a scene (an office, a hospital, etc). According to the proposed measure, similar information would be transmitted between a synthetic and a real world scene, both depicting a specific schema. This would ultimately indicate which areas in a Virtual Environment could be rendered in lower quality without affecting information uptake. We examine whether computationally more expensive scenes with greater visual fidelity affect memory performance after exposure to immersive VEs, or are they merely more aesthetically pleasing than their diminished visual quality counterparts. Initial results indicate that memory schemata function in VEs similar to real world environments. “High-level” visual cognition is unaffected by ubiquitous graphics manipulations such as polygon count and depth of shadow rendering; “normal” cognition operates as long as the scenes look acceptably realistic. However, when the overall realism of the scene is greatly reduced, such as in wireframe, then visual cognition becomes abnormal. Effects that distinguish “congruent” from “incongruent” objects change because the whole scene now looks incongruent; we have shown that this effect is not due to a failure of basic recognition.

Research Topics

Perception-based Computer Graphics

N/A

Advanced Display Technologies

N/A

Architectural Visualization

N/A

Ambient Intelligence and Ubiquitous Computing

N/A

R&D Projects

Outcomes

Related Persons

Nikos Zervoudakis Mr. zervoudakis at ced.tuc.gr