Volume Clipping via Per-Fragment Operations in Texture-Based Volume Visualization

1
Volume Clipping via Per-Fragment Operations in
Texture-Based Volume Visualization
Visualization and Interactive Systems
Group University of Stuttgart Germany
2
Approach

• Currently
• Primitive clip planes
• Clip object specified
• Clip geometry using a polygon mesh
• Volumetric 3d textures as a voxel mask

3
Depth-based clipping

• Straightforward approach
• 1 boundary
• Render clip to depth buffer
• Disable writing to depth buffer
• Enable depth test
• Greater or less
• - Render textures

Depth structure for 1 pixel in the frame buffer
4
Depth-based volume probing

• Render front faces of geometry to depth buffer
  to get zfront
• Store depth buffer to a texture (gfx-mem)
• Fragment shader shifts all depth by zfront
• Render back faces
• Render slices with depth shift/test

5
Volumetric Clip Textures

• Voxelise the clip geometry as binary volume
• 1 voxel is inside 0 voxel is clipped
• Per-fragment multiplicaton of volclip texture
• Disadvantages
• Jaggy artifacts due to discretization

6
Results

• Voxel Clipping
• Varying alpha for smooth variation
• Visible jaggy edges trilinear interpolation
  doesnt give sharp edges
• Reuse clip texture cylinder as many translated
  2d circle texture

• Clip Geometry
• High quality (per pixel)
• No aliasing
• Change of geometry is trivial
• slower (x4)
• - Convex geometry (multipass??)