Multi-resolution modeling, visualization
and compression of volumetric data

IEEE Visualization 2003, Tutorial number 3 (full day), Monday, October 20, 2003.

Organized by: Valerio Pascucci (please send email with your comments).

Instructors:

Paolo Cignoni, Italian National Research Council.
Leila De Floriani, University of Genova and University of Maryland.
Valerio Pascucci, Lawrence Livermore National Laboratory.
Jarek Rossignac, Georgia Institute of Technology.
Claudio T. Silva, Oregon Health & Science University.

Course Description.
Volume meshes are widely used in modeling, simulation and visualization of physical objects and natural phenomena and their constant increase in size makes imperative the use of multi-resolution techniques in any interactive application. This course covers the fundamental issues in multi-resolution modeling and visualization of volume meshes, including topics from mesh construction, visualization and compression. The presentation of each topic starts from the basics to be understandable by a novice.
Mesh construction is discussed from the dual points of view of (i) decimation of high-resolution data, or (ii) of refinement of coarse simple models. In the first case, we discuss error metrics and data reduction strategies. In the second case, we discuss techniques that allow using coarse meshes of very general type and provide ways to deal simply with adaptive refinement and sharp features.
The second session discusses multi-resolution data structures. We show how to build a multi-resolution model from a sequence of decimation operations and discuss traversal strategies that allow quick generation of adaptive approximations of the original model. The second half of the session is dedicated to the detailed analysis of a very simple multi-resolution model that allows achieving real-time interaction (slicing and volume rendering) with a very large data set (rectilinear grid of size 2048x2048x2048).
The third portion of the course is devoted to the problem of using efficient visualization techniques to render interactively unstructured volume meshes. The session is divided into two parts; one on iso-surface extraction and the other on direct volume rendering.
The course is concluded with a session devoted to the problem of storing and transmitting efficiently volume meshes. This includes discussion of connectivity and geometry coding as well as different strategies for single resolution encoding versus progressive encoding. The most advanced topics will include compression of 4D meshes.

Syllabus (pdf-tutorial-notes):
-	Introduction (10 minutes)
-	Volumetric Modeling (First Morning Session)
		Simplification of unstructured tetrahedral meshes
			Topics: Error metrics. Boundary effects and other special cases. Decimation techniques. Speaker: Paolo Cignoni. Time: 45 min. Downloads: (pdf-references) (pdf-presentation)
		Subdivision Methods for Volume Meshes
			Topics: One-dimensional B-spline subdivision. Tensor product extension to 2D and 3D meshes. 3D multi-linear subdivision. Slow growing subdivision. Speaker: Valerio Pascucci. Time: 45 min. Downloads: (pdf-references) (pdf-presentation)
-	Multi-resolution Representation Models (Second Morning Session)
		Level-Of-Detail (LOD): data structures and adaptive traversal algorithms.
			Topics: Graph-based data structures for unstructured meshes Tree-based data structure for unstructured meshes Hierarchical representations of regular meshes Selective refinement queries and algorithms Speaker: Leila De Floriani. Time: 55 min. Downloads: (pdf-references) (pdf-presentation)
		Real-time Rendering in External Memory.
			Topics: Cache Oblivious Data Layouts for Out-of-core Refinement. Progressive Slicing and Volume Rendering of Rectilinear Grids. Remote Data Streaming for Real-time Monitoring of Scientific Simulations. Speaker: Valerio Pascucci. Time: 50 min. Downloads: (pdf-references) (pdf-presentation)
-	Visualization Techniques (First Afternoon Session)
		Volume rendering.
			Topics: Foundations of Volume Rendering. Transfer Functions. Ray Tracing. Out-of-core Implementations. Speaker: Claudio Silva. Time: 50 min.
		Isosurface computation.
			Topics: Isosurface Computation for Tetrahedral Meshes. Isosurface Computation for rectilinear Grids with Trilinear Interpolant. Accelerated Techniques with Interval Trees and Seed Sets. Adaptive and Out-of-core Extensions. Speaker: Claudio Silva. Time: 55 min. Downloads: (pdf-references) (pdf-presentation)
-	Compression Techniques (Second Afternoon Session)
		Compression of 3D tetrahedral meshes.
			Topics: Representation and encoding of triangular meshes. Fast compression of triangle mesh connectivity. Encoding of tetrahedral meshes. Progressive encoding of triangular meshes. Progressive transmission of tetrahedral meshes. Speaker: Jarek Rossignac. Time: 45 min. Downloads: (pdf-references) (pdf-presentation)
		Compression of 4D regular data sets.
			Topics: Compression and remote inspection of 4D datasets. The Lorenzo predictor for the out-of-core compression of 4D datasets. Speaker: Jarek Rossignac. Time: 25 min. Downloads: (pdf-references) (pdf-presentation)
		Compression of key-framed animations.
			Topics: Space-Time compression of 3D animations. Speaker: Jarek Rossignac. Time: 25 min. Downloads: (pdf-references) (pdf-presentation)
-	Questions and answers (15 minutes)