UNC Dynamic Scene Benchmarks
1. Introduction
We provide a few dynamic datasets, which have been used in the following papers:
Fast Continuous Collision Detection using Parallel Filter in Subspace,
Chen Tang, Sheng Li, and Guopin Wang
ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (i3D 2011), San Fransisco, CA, Feb. 18-20, 2011, 71-80.
Collision-Streams: Fast GPU-based Collision Detection for Deformable Models,
Min Tang, Dinesh Manocha, Jiang Lin, and Ruofeng Tong
ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (i3D 2011), San Fransisco, CA, Feb. 18-20, 2011, 63-70.
Fast and Scalable CPU/GPU Collision Detection for Rigid and Deformable Surfaces.
Simon Pabst, Artur Koch, Wolfgang Strasser
Comput. Graph. Forum 29(5): 1605-1612 (2010)
Fast Continuous Collision Detection using Deforming Non-Penetration Filters,
Min Tang, Dinesh Manocha, and Ruofeng Tong
ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games (i3D 2010)
MCCD: Multi-Core Collision Detection between Deformable Models using Front-based Decomposition,
Min Tang, Dinesh Manocha, and Ruofeng Tong
Graphical Models, Vol. 72, No. 2, pp.7-23, 2010
Min Tang, Sean Curtis, Sung-Eui Yoon, and Dinesh Manocha
IEEE Transactions on Visualization and Computer Graphics, 2009, 15(4): 544-557.
Ray Tracing Dynamic Scenes using Selective Restructuring,
Sung-eui Yoon, Sean Curtis, and Dinesh Manocha
Eurographics Symposium on Rendering 2007
Fast Proximity Computation Among Deformable Models using Discrete Voronoi Diagrams ,
Avneesh Sud, Naga Govindaraju, Russell Gayle, Ilknur Kabul, and Dinesh Manocha
ACM SIGGRAPH 2006
Particularly, our benchmarks have high model complexity or non-rigid deformations or breaking objects.
All the dynamic scenes are stored in the ply file format; also, each dynamic frame is stored in a separate ply file with different ID. Sequences of each benchmark are shown in the MPEG4 video of the paper mentioned above. Please contact Sung-eui Yoon or Dinesh Manocha, if you have questions.
2. Dynamic Benchmarks
A. Exploding dragon and bunny model
# of triangles: 252K
# of raw sequences: 16
Zipped file size: 48MB, download
Characteristics: one or more objects break into smaller pieces (i.e. explosions) and thereby change their topologies.
B. Cloth-ball simulation model
# of triangles: 92K
# of raw sequences: 94
Zipped file size: 102MB download
Characteristics: it has non-rigid deformation.
C. N-body simulation
# of triangles: 146K
# of raw sequences: 94
Zipped file size: 78MB download
Characteristics: this benchmark consists of multiple moving objects. Each object may undergo a rigid or deformable motion and the objects collide with each other and the ground.
D. Letters
# of triangles: 5K
# of raw sequences: 475
Zipped file size: 23MB download
Characteristics: Multiple deforming models of alphabet and number shapes that fall into a bowl and break into pieces.
E. Lion
# of triangles: 1.6M
# of raw sequences: 45
Zipped file size: 440MB download
Characteristics: A stone falls on top of a Chinese lion model and the lion gradually breaks into a high number of colliding pieces.
F. Funnel
# of triangles: 18.5K
# of raw sequences: 51
Zipped file size: 95.3M download
Characteristics: A cloth folks into the funnel and passes through it due to the force applied by the ball.
Additional benchmarks corresponding to Fracturing Models are available as KAIST Model Benchmarks .
Acknowledgements
We would like to thank Naga Govindaraju, Ilknur Kabul, and Stephane Redon for the exploding dragon/ bunny model and cloth-ball simulation model; Sean Curtis for the N-body benchmark; Jieyi Zhao for the Lion benchmark; Simon Pabst for the Funnel benchmark.
Other developers include Sung-eui Yoon, Russ Gayle, Avneesh Sud, Sean Curtis, Jieyi Zhao, Min Tang, Ming Lin, and Dinesh Manocha.
Related links
Collision Detection and Proximity Queries
Updated March 2012