Mark A. Duchaineau

ROAMing terrain: real-time optimally adapting meshes

Terrain visualization is a difficult problem for applications requiring accurate images of large datasets at high frame rates, such as flight simulation and ground-based aircraft testing using synthetic sensor simulation. On current graphics hardware, the problem is to maintain dynamic, view-dependent triangle meshes and texture maps that produce good images at the required frame rate. We present an algorithm for constructing triangle meshes that optimizes flexible view-dependent error metrics, produces guaranteed error bounds, achieves specified triangle counts directly and uses frame-to-frame coherence to operate at high frame rates for thousands of triangles per frame. Our method, dubbed Real-time Optimally Adapting Meshes (ROAM), uses two priority queues to drive split and merge operations that maintain continuous triangulations built from pre-processed bintree triangles. We introduce two additional performance optimizations: incremental triangle stripping and priority-computation deferral lists. ROAMs execution time is proportional to the number of triangle changes per frame, which is typically a few percent of the output mesh size; hence ROAMs performance is insensitive to the resolution and extent of the input terrain. Dynamic terrain and simple vertex morphing are supported.

Simulating materials failure by using up to one billion atoms and the world's fastest computer: Brittle fracture.

We describe the first of two large-scale atomic simulation projects on materials failure performed on the 12-teraflop ASCI (Accelerated Strategic Computing Initiative) White computer at Lawrence Livermore National Laboratory. This is a multimillion-atom simulation study of crack propagation in rapid brittle fracture where the cracks travel faster than the speed of sound. Our finding centers on a bilayer solid that behaves under large strain like an interface crack between a soft (linear) material and a stiff (nonlinear) material. We verify that the crack behavior is dominated by the local (nonlinear) wave speeds, which can be in excess of the conventional sound speeds of a solid.

Simulating materials failure by using up to one billion atoms and the world's fastest computer: Work-hardening

We describe the second of two large-scale atomic simulation projects on materials failure performed on the 12-teraflop ASCI (Accelerated Strategic Computing Initiative) White computer at the Lawrence Livermore National Laboratory. This investigation simulates ductile failure by using more than one billion atoms where the true complexity of the creation and interaction of hundreds of dislocations are revealed.

Interactive view-dependent rendering of large isosurfaces

We present an algorithm for interactively extracting and rendering isosurfaces of large volume datasets in a view-dependent fashion. A recursive tetrahedral mesh refinement scheme, based on longest edge bisection, is used to hierarchically decompose the data into a multiresolution structure. This data structure allows fast extraction of arbitrary isosurfaces to within user specified view-dependent error bounds. A data layout scheme based on hierarchical space filling curves provides access to the data in a cache coherent manner that follows the data access pattern indicated by the mesh refinement.

Very High Resolution Simulation of Compressible Turbulence on the IBM-SP System

Understanding turbulence and mix in compressible flows is of fundamental importance to real-world applications such as chemical combustion and supernova evolution. The ability to run in three dimensions and at very high resolution is required for the simulation to accurately represent the interaction of the various length scales, and consequently, the reactivity of the intermixin species. Toward this end, we have carried out a very high resolution (over 8 billion zones) 3-D simulation of the Richtmyer-Meshkov instability and turbulent mixing on the IBM Sustained Stewardship TeraOp (SST) system, developed under the auspices of the Department of Energy (DOE) Accelerated Strategic Computing Initiative (ASCI) and located at Lawrence Livermore National Laboratory. We have also undertaken an even higher resolution proof-of-principle calculation (over 24 billion zones) on 5832 processors of the IBM system, which executed for over an hour at a sustained rate of 1.05 Tflop/s, as well as a short calculation with a modified algorithm that achieved a sustained rate of 1.18Tflop/s. The full production scientific simulation, using a further modified algorithm, ran for 27,000 timesteps in slightly over a week of wall time using 3840 processors of the IBM system, clockin a sustained throughput of roughly 0.6 teraflop per second (32-bit arithmetic). Nearly 300,000 graphics files comprising over three terabytes of data were produced and post-processed. The capability of running in 3-D at high resolution enabled us to get a more accurate and detailed picture of the fluid-flow structure - in particular, to simulate the development of fine scale structures from the interactions of long-and short-wavelength phenomena, to elucidate differences between two-dimensional and three-dimensional turbulence, to explore a conjecture regarding the transition from unstable flow to fully developed turbulence with increasing Reynolds number, and to ascertain convergence of the computed solution with respect to mesh resolution.

Generalized B-spline subdivision-surface wavelets for geometry compression

We present a new construction of lifted biorthogonal wavelets on surfaces of arbitrary two-manifold topology for compression and multiresolution representation. Our method combines three approaches: subdivision surfaces of arbitrary topology, B-spline wavelets, and the lifting scheme for biorthogonal wavelet construction. The simple building blocks of our wavelet transform are local lifting operations performed on polygonal meshes with subdivision hierarchy. Starting with a coarse, irregular polyhedral base mesh, our transform creates a subdivision hierarchy of meshes converging to a smooth limit surface. At every subdivision level, geometric detail is expanded from wavelet coefficients and added to the surface. We present wavelet constructions for bilinear, bicubic, and biquintic B-spline subdivision. While the bilinear and bicubic constructions perform well in numerical experiments, the biquintic construction turns out to be unstable. For lossless compression, our transform is computed in integer arithmetic, mapping integer coordinates of control points to integer wavelet coefficients. Our approach provides a highly efficient and progressive representation for complex geometries of arbitrary topology.

Real-time optimal adaptation for planetary geometry and texture: 4-8 tile hierarchies

The real-time display of huge geometry and imagery databases involves view-dependent approximations, typically through the use of precomputed hierarchies that are selectively refined at runtime. A classic motivating problem is terrain visualization in which planetary databases involving billions of elevation and color values are displayed on PC graphics hardware at high frame rates. This paper introduces a new diamond data structure for the basic selective-refinement processing, which is a streamlined method of representing the well-known hierarchies of right triangles that have enjoyed much success in real-time, view-dependent terrain display. Regular-grid tiles are proposed as the payload data per diamond for both geometry and texture. The use of 4-8 grid refinement and coarsening schemes allows level-of-detail transitions that are twice as gradual as traditional quadtree-based hierarchies, as well as very high-quality low-pass filtering compared to subsampling-based hierarchies. An out-of-core storage organization is introduced based on Sierpinski indices per diamond, along with a tile preprocessing framework based on fine-to-coarse, same-level, and coarse-to-fine gathering operations. To attain optimal frame-to-frame coherence and processing-order priorities, dual split and merge queues are developed similar to the realtime optimally adapting meshes (ROAM) algorithm, as well as an adaptation of the ROAM frustum culling technique. Example applications of lake-detection and procedural terrain generation demonstrate the flexibility of the tile processing framework.

A scalable, hybrid scheme for volume rendering massive data sets

We introduce a parallel, distributed memory algorithm for volume rendering massive data sets. The algorithms scalability has been demonstrated up to 400 processors, rendering one hundred million unstructured elements in under one second. The heart of the algorithm is a hybrid approach that parallelizes over both the elements of the input data and over the pixels of the output image. At each stage of the algorithm, there are strong limits on how much work each processor performs, ensuring good parallel efficiency. The algorithm is sample-based. We present two techniques for calculating the sample points: a 3D rasterization technique and a kernel-based technique, which trade off between speed and generality. Finally, the algorithm is very flexible. It can be deployed in general purpose visualization tools and can also support diverse mesh types, ranging from structured grids to curvilinear and unstructured meshes to point clouds.

Adaptive 4-8 Texture Hierarchies

We address the texture level-of-detail problem for extremely large surfaces such as terrain during realtime, view-dependent rendering. A novel texture hierarchy is introduced based on 4-8 refinements of raster tiles, in which the texture grids in effect rotate 45 degrees for each level of refinement. This hierarchy provides twice as many levels of detail as conventional quadtree-style refinement schemes such as mipmaps, and thus provides per-pixel view-dependent filtering that is twice as close to the ideal cutoff frequency for an average pixel. Because of this more gradual change in low-pass filtering, and due to the more precise emulation of the ideal cutoff frequency, we find in practice that the transitions between texture levels of detail are not perceptible. This allows rendering systems to avoid the complexity and performance costs of per-pixel blending between texture levels of detail. The 4-8 texturing scheme is integrated into a variant of the real-time optimally adapting meshes (ROAM) algorithm for view-dependent multiresolution mesh generation. Improvements to ROAM included here are: the diamond data structure as a streamlined replacement for the triangle bintree elements, the use of low-pass-filtered geometry patches in place of individual triangles, integration of 4-8 textures, and a simple out-of-core data access mechanism for texture and geometry tiles.

Material interface reconstruction

The paper presents an algorithm for material interface reconstruction for data sets where fractional material information is given as a percentage for each element of the underlying grid. The reconstruction problem is transformed to a problem that analyzes a dual grid, where each vertex in the dual grid has an associated barycentric coordinate tuple that represents the fraction of each material present. Material boundaries are constructed by analyzing the barycentric coordinate tuples of a tetrahedron in material space and calculating intersections with Voronoi cells that represent the regions where one material dominates. These intersections are used to calculate intersections in the Euclidean coordinates of the tetrahedron. By triangulating these intersection points, one creates the material boundary. The algorithm can treat data sets containing any number of materials. The algorithm can also create nonmanifold boundary surfaces if necessary. By clipping the generated material boundaries against the original cells, one can examine the error in the algorithm. Error analysis shows that the algorithm preserves volume fractions within an error range of 0.5 percent per material.