Efficient deformable geometry image-maps

  • Authors:
  • Bin Sheng;Hanqiu Sun

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

  • Venue:
  • Proceedings of the 17th ACM Symposium on Virtual Reality Software and Technology
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Multiresolution rendering of deformable models, performing fast rendering of global deformations while preserving local surface details, is usually a computation-costly and time-consuming process, because two non-trivial operators are involved. We propose a novel GPU-based adaptive rendering of deforming mesh sequence on sole-cube maps (SCM), which is a variant of geometry images built upon spherical parameterizations. We also introduce the differential coordinates to bound the local resampling error for supporting details preservation and view-dependent visualization. By precomputing the adaptive SCM texture atlas of deforming mesh sequences as well as their differential coordinates, we can map both the deformation and the level-of-details (LOD) operator to the GPU. The proposed algorithm enables us to reconstruct the deformed positions and sufficiently fine-scale approximations of deforming mesh sequences for efficient GPU processing. The GPU-friendly data structure and process allow us to render dynamically deforming 3D models with GPU parallelization, also our system improves the efficiency of the manipulations of node selection and boundary stitching, significantly alleviating the computing load on CPU.