Spherical Q2-tree for sampling dynamic environment sequences

  • Authors:
  • Liang Wan;Tien-Tsin Wong;Chi-Sing Leung

  • Affiliations:
  • The Chinese University of Hong Kong;The Chinese University of Hong Kong;City University of Hong Kong

  • Venue:
  • EGSR'05 Proceedings of the Sixteenth Eurographics conference on Rendering Techniques
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Previous methods in environment map sampling seldom consider a sequence of dynamic environment maps. The generated sampling patterns of the sequence may not maintain the temporal illumination consistency and result in choppy animation. In this paper, we propose a novel approach, spherical Q2-tree, to address this consistency problem. The local adaptive nature of the proposed method suppresses the abrupt change in the generated sampling patterns over time, hence ensures a smooth and consistent illumination. By partitioning the spherical surface with simple curvilinear equations, we construct a quadrilateral-based quadtree over the sphere. This Q2-tree allows us to adaptively sample the environment based on an importance metric and generates low-discrepancy sampling patterns. No time-consuming relaxation is required. The sampling patterns of a dynamic sequence are rapidly generated by making use of the summed area table and exploiting the coherence of consecutive frames. From our experiments, the rendering quality of our sampling pattern for a static environment map is comparable to previous methods. However, our method produces smooth and consistent animation for a sequence of dynamic environment maps, even the number of samples is kept constant over time.