Title

Visibility-Based Forest Walk-Through Using Inertial Level Of Detail Models

Keywords

graphics processing unit (GPU); level-of-detail; Simulation/training walk-through; synthetic natural environments; visibility; visualization

Abstract

Modern simulation and training applications often require visualization of large scenes with many complex objects. However, this places prohibitive requirements on graphics hardware. For example, while methods for rendering near photo-realistic vegetation scenes have been described in the literature, they require minutes of computation. In this paper we present scene and level of detail management (LOD) techniques for achieving interactive frame rates for a forest walk-through. The framework selects the LOD based on object visibility, in addition to projected size. Moreover, visibility is calculated at runtime so we can support dynamic scene modification, such as interactively altered trees due to wind, explosions or fire. We also introduce an inertial level of detail model to minimize popping artifacts – rather than instantly switching the LODs, discrete LODs are smoothly blended over a number of frames. The described techniques can be readily adapted for scenes other than forests, providing support for simulations involving both static and dynamic synthetic environments, whether natural (e.g., dense vegetation) or man-made (e.g., urban landscapes). © 2007, SAGE Publications. All rights reserved.

Publication Date

1-1-2007

Publication Title

The Journal of Defense Modeling and Simulation: Applications, Methodology, Technology

Volume

4

Issue

2

Number of Pages

80-96

Document Type

Article

Personal Identifier

scopus

DOI Link

https://doi.org/10.1177/154851290700400202

Socpus ID

84993660610 (Scopus)

Source API URL

https://api.elsevier.com/content/abstract/scopus_id/84993660610

This document is currently not available here.

Share

COinS