Nico Galoppo

LU-GPU: Efficient Algorithms for Solving Dense Linear Systems on Graphics Hardware

We present a novel algorithm to solve dense linear systems using graphics processors (GPUs). We reduce matrix decomposition and row operations to a series of rasterization problems on the GPU. These include new techniques for streaming index pairs, swapping rows and columns and parallelizing the computation to utilize multiple vertex and fragment processors. We also use appropriate data representations to match the rasterization order and cache technology of graphics processors. We have implemented our algorithm on different GPUs and compared the performance with optimized CPU implementations. In particular, our implementation on a NVIDIA GeForce 7800 GPU outperforms a CPU-based ATLAS implementation. Moreover, our results show that our algorithm is cache and bandwidth efficient and scales well with the number of fragment processors within the GPU and the core GPU clock rate. We use our algorithm for fluid flow simulation and demonstrate that the commodity GPU is a useful co-processor for many scientific applications.

Adaptive dynamics of articulated bodies

Forward dynamics is central to physically-based simulation and control of articulated bodies. We present an adaptive algorithm for computing forward dynamics of articulated bodies: using novel motion error metrics, our algorithm can automatically simplify the dynamics of a multi-body system, based on the desired number of degrees of freedom and the location of external forces and active joint forces. We demonstrate this method in plausible animation of articulated bodies, including a large-scale simulation of 200 animated humanoids and multi-body dynamics systems with many degrees of freedom. The graceful simplification allows us to achieve up to two orders of magnitude performance improvement in several complex benchmarks.

Soft Articulated Characters with Fast Contact Handling

Fast contact handling of soft articulated characters is a computationally challenging problem, in part due to complex interplay between skeletal and surface deformation. We present a fast, novel algorithm based on a layered representation for articulated bodies that enables physically‐plausible simulation of animated characters with a high‐resolution deformable skin in real time. Our algorithm gracefully captures the dynamic skeleton‐skin interplay through a novel formulation of elastic deformation in the pose space of the skinned surface. The algorithm also overcomes the computational challenges by robustly decoupling skeleton and skin computations using careful approximations of Schur complements, and efficiently performing collision queries by exploiting the layered representation. With this approach, we can simultaneously handle large contact areas, produce rich surface deformations, and capture the collision response of a character/s skeleton.

Fast simulation of deformable models in contact using dynamic deformation textures

We present an efficient algorithm for simulating contacts between deformable bodies with high-resolution surface geometry using dynamic deformation textures, which reformulate the 3D elastoplastic deformation and collision handling on a 2D parametric atlas to reduce the extremely high number of degrees of freedom in such a computationally demanding simulation. We perform proximity queries for deformable bodies using a two-stage algorithm directly on dynamic deformation textures, resulting in output-sensitive collision detection that is independent of the combinatorial complexity of the deforming meshes. We present a robust, parallelizable formulation for computing constraint forces using implicit methods that exploits the structure of the motion equations to achieve highly stable simulation, while taking large time steps with inhomogeneous materials. The dynamic deformation textures can also be used directly for real-time shading and can easily be implemented using SIMD architecture on commodity hardware. We show that our approach, complementing existing pioneering work, offers significant computational advantages on challenging contact scenarios in dynamic simulation of deformable bodies.

Interactive Virtual Hair Salon

User interaction with animated hair is desirable for various applications but difficult because it requires real-time animation and rendering of hair. Hair modeling, in cluding styling, simulation, and rendering, is computationally challenging due to the enormous number of deformable hair strands on a human head, elevating the computational complexity of many essential steps, such as collision detection and self-shadowing for hair. Using simulation localization techniques, multi-resolution representations, and graphics hardware rendering acceleration, we have developed a physically-based virtual hair salon system that simulates and renders hair at accelerated rates, enabling users to interactively style virtual hair. With a 3D haptic interface, users can directly manipulate and position hair strands, as well as employ real-world styling applications (cutting, blow-drying, etc.) to create hairstyles more intuitively than previous techniques.

Interactive haptic rendering of high-resolution deformable objects

We present an efficient algorithm for haptic rendering of deformable bodies with highly detailed surface geometry using a fast contact handling algorithm. We exploit a layered deformable representation to augment the physically based deformation simulation with efficient collision detection, contact handling and interactive haptic force feedback.

AdaPT: Real-time adaptive pedestrian tracking for crowded scenes

We present a novel realtime algorithm to compute the trajectory of each pedestrian in a crowded scene. Our formulation is based on an adaptive scheme that uses a combination of deterministic and probabilistic trackers to achieve high accuracy and efficiency simultaneously. Furthermore, we integrate it with a multi-agent motion model and local interaction scheme to accurately compute the trajectory of each pedestrian. We highlight the performance and benefits of our algorithm on well-known datasets with tens of pedestrians.

Accelerated wave-based acoustics simulation

We present an efficient technique to model sound propagation accurately in an arbitrary 3D scene by numerically integrating the wave equation. We show that by performing an offline modal analysis and using eigenvalues from a refined mesh, we can simulate sound propagation with reduced dispersion on a much coarser mesh, enabling accelerated computation. Since performing a modal analysis on the complete scene is usually not feasible, we present a domain decomposition approach to drastically shorten the pre-processing time. We introduce a simple, efficient and stable technique for handling the communication between the domain partitions. We validate the accuracy of our approach against cases with known analytical solutions. With our approach, we have observed up to an order of magnitude speedup compared to a standard finite-difference technique.

Visual simulation of shockwaves

We present an efficient method for visual simulations of shock phenomena in compressible, inviscid fluids. Our algorithm is derived from one class of the finite volume method especially designed for capturing shock propagation, but offers improved efficiency through physically-based simplification and adaptation for graphical rendering. Our technique is well suited for parallel implementation on multicore architectures and is also capable of handling complex, bidirectional object-shock interactions stably and robustly. We describe its applications to various visual effects, including explosion, sonic booms and turbulent flows.

A Simulation-based VR System for Interactive Hairstyling

We have developed a physically-based VR system that enables users to interactively style dynamic virtual hair by using multiresolution simulation techniques and graphics hardware rendering acceleration for simulating and rendering hair in real time. With a 3D haptic interface, users can directly manipulate and position hair strands, as well as employ real-world styling applications (cutting, blow-drying, etc.) to create hairstyles more intuitively than previous techniques.