Sequential Point Clusters for Efficient Point-Based Rendering
|
|
Advancement in modeling technologies has enabled us to obtain 3D models composed of an extremely large number of polygons. Unfortunately, however, the conventional polygon-based rendering method is not rapid enough for rendering such models. We propose a system for effectively rendering such models based on using the Point-Based Rendering technique that considers multi-resolution representations. The object model is first converted into a novel data structure, Sequential Point Clusters by spatial clustering, in order that we can display the model in an interactive manner with the aid of graphics hardware. Additionally, we propose a hybrid rendering method that combines point- and polygon-based rendering to improve the quality in rendering results.
Publications
“Efficient Point-based Rendering Method for Huge 3D Models using Sequential Point Clusters”,
Proc. Meeting on Image Recognition and Understanding, vol.1, pp.207-212,
July 2004
(Japanese)
Supplementary Material
Figures in Paper
Sequentialization of a hierarchy structure: The hierarchical sets of points are sorted into a 1D list.
Prunning of positional clustering: The row of bars means the list of points, and yellow bars is rest points after prunning.
Clustering: The tree structure of hierarchical point sets are converted into point clusters.
Efficiency and overhead: This plot shows the relation between efficiency and overhead by clustering.
Results
Rendering result of a happy-buddha model (543652 vertices, 1087716 triangles).
Rendering results of Bayon-tower model (8681691 vertices, 13939070 triangles).
Miscellaneous
See also Yasuhide Okamoto's page