Moving Cube Static Vase

1
GPU-Based Voxelization and Its Application in
Flow Modeling Zhe Fan, Wei Li, Xiaoming Wei,
Arie Kaufman Center for Visual Computing and
Department of Computer Science, Stony Brook
University http//www.cs.sunysb.edu/vislab/projec
ts/gpgpu

• Overview
• A voxelization method on the GPU
• Based on nVIDIAs depth peeling
• Application moving boundaries evaluation in flow
  simulation

• Algorithm
• Peel depth layers and generate voxels
• Determine whether done by Occlusion Query
  Extension
• To avoid missing voxels repeat depth peeling
  from 3 orthogonal viewing directions
• Copy generated voxels to vertex array by using
  Pixel Data Range and Vertex Array Range
• Render voxels to Flat3D textures

Swimming Jellyfish on the GPU Lower
Turbulence Higher Turbulence

• GPU-Based Voxelization
• Previous slicing method
• Each rendering pass creates a slice of volume,
  between two z clipping planes
• Our depth peeling method
• Each pass creates voxels on one depth layer
• Fewer rendering passes
• Faster when voxelizing surfaces

• Extend to Flow Boundary Handling
• Generate boundary nodes on-the-fly
• Generate additional boundary information
• Accurate boundary positions
• Velocity of boundary surface
• etc
• Use boundary information to update flow grid
• Useful in flow interaction
• Flow model already runs on GPU lattice Boltzmann
  model
• Moving boundary evaluation runs on GPU
• Dynamics of moving obstacle may also run on GPU
• All inside GPU - no need to read out

Moving Cube Static Vase

• Acknowledgement
• NSF CCR-0306438