Kui Wu
Research Subjects
Kui Wu has a total of 125 co-authors. Here are the top co-authors by number of shared publications:
- Zherong Pan
LightSpeed Studios
23 shared publications - Xifeng Gao
LightSpeed Studios
22 shared publications - Tongtong Wang
LightSpeed Studios, Shen Zhen, China
7 shared publications - Yin Yang
University of Utah, Salt Lake City, USA
7 shared publications - Cem Yuksel
University of Utah, Salt Lake City, UT, USA
5 shared publications - Chenfanfu Jiang
University of California Los Angeles
5 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-20
Vibrational Affine Body Dynamics
Rigid-body and affine body dynamics are default simulation modalities for rigid or near-rigid objects. Their low-rank kinematic models inherently preclude vibrational responses that stiff bodies exhibit during abrupt impacts or contacts. Simulating such behavior with full or reduced finite element method (FEM) requires high spatial resolution and small time steps, which significantly increases computational costs of system solving and collision handling. We propose vibrational affine body dynamics (VABD), a novel simulation framework that captures these missing high-frequency vibrations while preserving excellent efficiency and robustness, even in complex scenarios. VABD introduces a carefully constructed secondary vibration subspace that augments standard affine body dynamics to encode high-frequency deformation on top of the time-varying affine frame, while remaining orthogonal to this evolving affine subspace with respect to the system matrix. This subspace design enables efficient reduced simulation with improved expressiveness. For collision handling, VABD employs medial-axis-based proxies and interleaves medial continuous and discrete collision detection to balance robustness and performance across different motion scales. We train a neural MAT (medial axis transform) mapping from reduced vibrational coordinates to medial primitives, enabling collision responses to be coupled back to the reduced system with well-defined derivatives. To efficiently solve the resulting implicit system, we introduce a Newton-quasi-Newton solver that exploits the fact that collision Hessians remain largely unchanged under small time steps, which effectively suppresses the overhead of repeated Hessian construction while retaining strong second-order convergence. Experiments demonstrate that VABD produces realistic vibrational responses for stiff objects undergoing intensive and extensive collisions. In terms of performance, VABD is orders of magnitude faster than FEM-based or reduced simulation approaches.
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-20
Open AccessVibrational Affine Body Dynamics
Rigid-body and affine body dynamics are default simulation modalities for rigid or near-rigid objects. Their low-rank kinematic models inherently preclude vibrational responses that stiff bodies exhibit during abrupt impacts or contacts. Simulating such behavior with full or reduced finite element method (FEM) requires high spatial resolution and small time steps, which significantly increases computational costs of system solving and collision handling. We propose vibrational affine body dynamics (VABD), a novel simulation framework that captures these missing high-frequency vibrations while preserving excellent efficiency and robustness, even in complex scenarios. VABD introduces a carefully constructed secondary vibration subspace that augments standard affine body dynamics to encode high-frequency deformation on top of the time-varying affine frame, while remaining orthogonal to this evolving affine subspace with respect to the system matrix. This subspace design enables efficient reduced simulation with improved expressiveness. For collision handling, VABD employs medial-axis-based proxies and interleaves medial continuous and discrete collision detection to balance robustness and performance across different motion scales. We train a neural MAT (medial axis transform) mapping from reduced vibrational coordinates to medial primitives, enabling collision responses to be coupled back to the reduced system with well-defined derivatives. To efficiently solve the resulting implicit system, we introduce a Newton-quasi-Newton solver that exploits the fact that collision Hessians remain largely unchanged under small time steps, which effectively suppresses the overhead of repeated Hessian construction while retaining strong second-order convergence. Experiments demonstrate that VABD produces realistic vibrational responses for stiff objects undergoing intensive and extensive collisions. In terms of performance, VABD is orders of magnitude faster than FEM-based or reduced simulation approaches.
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-17
Type-Free Global Intersection Analysis with Linear Displacement Fields
History-independent intersection repair is a critical bottleneck in physical simulation and geometric processing. Existing global intersection analysis (GIA) methods identify invalid regions of relatively small area, but they rely on type-specific heuristics for different contour categories, while local intersection contour minimization (ICM) methods are simpler to deploy but often stall in local minima under deep or nested penetrations. In this paper, we present a novel untangling framework that formulates intersection resolution as a minimum-displacement separation-direction problem. For each intersection contour, we test a set of candidate linear displacement directions using VF/EE/FV ray casting and filter the raw hits using cluster culling based on primitive adjacency to find the valid hits belonging to the penetration region. We rank candidate directions by displacement cost, and refine the most promising direction with projected Newton updates on the sphere. Finally, we convert the valid hit support into standard proximity-style penalty pairs and integrate them into the standard physics simulator. We evaluate our method on diverse contour configurations and self-intersection benchmarks. Compared with the GIA and ICM baselines, our method resolves intersections more consistently, particularly in challenging cases involving deep and nested layers. The results show that our method is a practical, solver-compatible untangling pipeline for complex mesh intersections.
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-17
Open AccessType-Free Global Intersection Analysis with Linear Displacement Fields
History-independent intersection repair is a critical bottleneck in physical simulation and geometric processing. Existing global intersection analysis (GIA) methods identify invalid regions of relatively small area, but they rely on type-specific heuristics for different contour categories, while local intersection contour minimization (ICM) methods are simpler to deploy but often stall in local minima under deep or nested penetrations. In this paper, we present a novel untangling framework that formulates intersection resolution as a minimum-displacement separation-direction problem. For each intersection contour, we test a set of candidate linear displacement directions using VF/EE/FV ray casting and filter the raw hits using cluster culling based on primitive adjacency to find the valid hits belonging to the penetration region. We rank candidate directions by displacement cost, and refine the most promising direction with projected Newton updates on the sphere. Finally, we convert the valid hit support into standard proximity-style penalty pairs and integrate them into the standard physics simulator. We evaluate our method on diverse contour configurations and self-intersection benchmarks. Compared with the GIA and ICM baselines, our method resolves intersections more consistently, particularly in challenging cases involving deep and nested layers. The results show that our method is a practical, solver-compatible untangling pipeline for complex mesh intersections.
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-17
Multi-level Quadrature for Projective Dynamics
Projective dynamics (PD) is a popular framework for efficient physical simulation. However, its factorization-based global solve maps poorly to GPU parallelization, while iterative alternatives often fail to converge effectively for stiff or high-resolution models. This paper reformulates the PD global solve by observing that its primary computational bottleneck lies in vector dot products, which can be reinterpreted as discrete integrations over the deformable body. The resulting integrand exhibits a Yukawa-like kernel structure that peaks sharply at each mesh node and decays exponentially with distance. Exploiting this prior, we partition the domain into near, middle, and far fields around each node, and design a quadrature rule for each that matches the local behavior of the integrand: an anisotropic Gaussian fit with a closed-form integral in the near field, a low-order Gauss-Legendre rule for the smoothed residual in the middle field, and a weighted k-means++ clustering with centroid-based quadrature in the irregular far field. This integration perspective reveals an effective sparsity pattern that enables both excellent parallelism and rapid convergence. Our algorithm also serves as a strong preconditioner for collision-involved PD systems and delivers significant speedup over existing PD solvers across a range of deformable scenarios.
- 9 November 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (6) , 1-17
Open AccessMulti-level Quadrature for Projective Dynamics
Projective dynamics (PD) is a popular framework for efficient physical simulation. However, its factorization-based global solve maps poorly to GPU parallelization, while iterative alternatives often fail to converge effectively for stiff or high-resolution models. This paper reformulates the PD global solve by observing that its primary computational bottleneck lies in vector dot products, which can be reinterpreted as discrete integrations over the deformable body. The resulting integrand exhibits a Yukawa-like kernel structure that peaks sharply at each mesh node and decays exponentially with distance. Exploiting this prior, we partition the domain into near, middle, and far fields around each node, and design a quadrature rule for each that matches the local behavior of the integrand: an anisotropic Gaussian fit with a closed-form integral in the near field, a low-order Gauss-Legendre rule for the smoothed residual in the middle field, and a weighted k-means++ clustering with centroid-based quadrature in the irregular far field. This integration perspective reveals an effective sparsity pattern that enables both excellent parallelism and rapid convergence. Our algorithm also serves as a strong preconditioner for collision-involved PD systems and delivers significant speedup over existing PD solvers across a range of deformable scenarios.
- 11 September 2026
- Published by Springer Nature
- p. 410-428
PoseShield: Neural Collision Fields for Human Self-collision Resolution
- 11 September 2026
- Published by Springer Nature
- p. 410-428
PoseShield: Neural Collision Fields for Human Self-collision Resolution
- 13 August 2026
- Published by Wiley in Computer Graphics Forum
- p. e70514
Spatial Eigenanalysis of 2D Deformation Energies
We present the first closed‐form, per‐element eigensystems for 2D deformation energies in the primary domain of spatial coordinates. Previous analyses have only been able to find such eigensystems in proxy coordinates such as the deformation gradient, but as the spatial positions directly correspond to global degrees of freedom, our analysis opens up avenues for novel algorithms that were not possible before. First, we propose a spectral preconditioner that approximates global near‐nullspace modes using local analytical eigenvectors and clusters global eigenvalues using a low‐rank correction. Second, we derive a unified projected Newton formulation that combines the element eigensystem with the mass and damping terms, yielding a superior eigenvalue filter. We test our preconditioner on surface parameterization problems and our projected Newton framework on dynamic simulation examples. The results show significant efficiency gains over previous methods. In particular, our physics‐aware filtering strategy reduces eigenvalue clamping by up to 30×, which in turn decreases the Newton iteration count by roughly 2×.
- 13 August 2026
- Published by Wiley in Computer Graphics Forum
- p. e70514
Spatial Eigenanalysis of 2D Deformation Energies
We present the first closed‐form, per‐element eigensystems for 2D deformation energies in the primary domain of spatial coordinates. Previous analyses have only been able to find such eigensystems in proxy coordinates such as the deformation gradient, but as the spatial positions directly correspond to global degrees of freedom, our analysis opens up avenues for novel algorithms that were not possible before. First, we propose a spectral preconditioner that approximates global near‐nullspace modes using local analytical eigenvectors and clusters global eigenvalues using a low‐rank correction. Second, we derive a unified projected Newton formulation that combines the element eigensystem with the mass and damping terms, yielding a superior eigenvalue filter. We test our preconditioner on surface parameterization problems and our projected Newton framework on dynamic simulation examples. The results show significant efficiency gains over previous methods. In particular, our physics‐aware filtering strategy reduces eigenvalue clamping by up to 30×, which in turn decreases the Newton iteration count by roughly 2×.
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-3
Fluid Simulation with the Lattice Boltzmann Method
Over the past decade, the Lattice Boltzmann Method (LBM) has matured into a leading framework for high-performance fluid simulation across research and industry alike. Its appeal rests on a set of properties that align unusually well with the demands of modern computing: collision and streaming operators that are strictly local, mapping naturally onto massively parallel architectures; an intrinsic ability to handle complex geometries through simple boundary rules; and a level of physical fidelity that has proven sufficient across a widening range of applications — from real-time simulation and visual effects to large-scale industrial tools such as virtual wind tunnels. The growing body of LBM-based work at SIGGRAPH and SIGGRAPH Asia reflects this trajectory directly: the method has moved from occasional appearance to recurring presence as a core computational framework. Despite this prominence, LBM has never been the subject of a dedicated SIGGRAPH course. To date, it has only been mentioned peripherally, most notably as a small component of the Real-Time Physics course at SIGGRAPH 2004. As a result, there remains a significant gap between the method’s practical importance and its formal coverage within the SIGGRAPH educational program. This lack of structured exposure has made it difficult for practitioners and researchers to fully understand the method’s foundations, strengths, limitations, and best practices for deployment on contemporary hardware. These course notes are an attempt to close that gap. Written by authors with direct experience across multiple LBM publications at SIGGRAPH over the past six years, they aim to provide something the existing literature rarely offers in a single place: a coherent path from first principles to production-oriented practice. The intent is not to survey recent work, but to equip attendees with the concepts, implementation details, and the critical perspective needed to start working with LBM solvers.
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-3
Open AccessFluid Simulation with the Lattice Boltzmann Method
Over the past decade, the Lattice Boltzmann Method (LBM) has matured into a leading framework for high-performance fluid simulation across research and industry alike. Its appeal rests on a set of properties that align unusually well with the demands of modern computing: collision and streaming operators that are strictly local, mapping naturally onto massively parallel architectures; an intrinsic ability to handle complex geometries through simple boundary rules; and a level of physical fidelity that has proven sufficient across a widening range of applications — from real-time simulation and visual effects to large-scale industrial tools such as virtual wind tunnels. The growing body of LBM-based work at SIGGRAPH and SIGGRAPH Asia reflects this trajectory directly: the method has moved from occasional appearance to recurring presence as a core computational framework. Despite this prominence, LBM has never been the subject of a dedicated SIGGRAPH course. To date, it has only been mentioned peripherally, most notably as a small component of the Real-Time Physics course at SIGGRAPH 2004. As a result, there remains a significant gap between the method’s practical importance and its formal coverage within the SIGGRAPH educational program. This lack of structured exposure has made it difficult for practitioners and researchers to fully understand the method’s foundations, strengths, limitations, and best practices for deployment on contemporary hardware. These course notes are an attempt to close that gap. Written by authors with direct experience across multiple LBM publications at SIGGRAPH over the past six years, they aim to provide something the existing literature rarely offers in a single place: a coherent path from first principles to production-oriented practice. The intent is not to survey recent work, but to equip attendees with the concepts, implementation details, and the critical perspective needed to start working with LBM solvers.
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-10
LightOpt: Lights Optimization for Real-time Rendering
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Open AccessLightOpt: Lights Optimization for Real-time Rendering
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-11
Global Discrete Optimization for Quad-Dominant Mesh Reduction
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-11
Open AccessGlobal Discrete Optimization for Quad-Dominant Mesh Reduction
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-11
HairLRM: Strand-based Hair Modeling via Large Reconstruction Models
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-11
Open AccessHairLRM: Strand-based Hair Modeling via Large Reconstruction Models
- Open Access
- 3 July 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (4) , 1-22
Interactive Yarn-level Knitwear with Nested Douglas-Rachford Splitting
While yarn-level garments offer rich dynamic details and compelling visual realism compared to triangle-based models, their wide adoption is hindered by the immense computational cost due to the presence of a large number of degrees of freedom (DOFs). This paper proposes a novel simulation framework designed to enhance the performance and stability for numerical simulation of nonlinear, non-convex, and high-resolution knitwear. Our method generalizes the Douglas-Rachford Splitting (DRS) scheme to resolve the non-convex coupling between stretching, shearing, bending, twisting, and contacting at each yarn thread. A key contribution is a nested decomposition strategy that decouples the non-convex variational energy into independent and convex sub-problems. Such convexification improves solver robustness and removes the necessity for frequent line searches. We provide a theoretically grounded strategy for metric selection for each sub-problem, derived from an analysis of the convergence guarantee of DRS. Consequently, our method achieves close-to-optimal convergence along the nonlinear iterations rather than relying on ad-hoc parameter tuning. The paper also clarifies a formal connection between our generalized DRS and the ADMM (Alternating Direction Method of Multipliers) framework, extending the applicability of our analysis to a broader set of constrained dynamics problems. Experimental results demonstrate that our method robustly handles complex knitwear simulation scenes with superior efficiency, stability, and physical fidelity compared to existing methods. With a matrix-free GPU parallelization, our method allows an interactive simulation rate of knitwear of multi-million DOFs.
- 3 July 2026
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 45 (4) , 1-22
Open AccessInteractive Yarn-level Knitwear with Nested Douglas-Rachford Splitting
While yarn-level garments offer rich dynamic details and compelling visual realism compared to triangle-based models, their wide adoption is hindered by the immense computational cost due to the presence of a large number of degrees of freedom (DOFs). This paper proposes a novel simulation framework designed to enhance the performance and stability for numerical simulation of nonlinear, non-convex, and high-resolution knitwear. Our method generalizes the Douglas-Rachford Splitting (DRS) scheme to resolve the non-convex coupling between stretching, shearing, bending, twisting, and contacting at each yarn thread. A key contribution is a nested decomposition strategy that decouples the non-convex variational energy into independent and convex sub-problems. Such convexification improves solver robustness and removes the necessity for frequent line searches. We provide a theoretically grounded strategy for metric selection for each sub-problem, derived from an analysis of the convergence guarantee of DRS. Consequently, our method achieves close-to-optimal convergence along the nonlinear iterations rather than relying on ad-hoc parameter tuning. The paper also clarifies a formal connection between our generalized DRS and the ADMM (Alternating Direction Method of Multipliers) framework, extending the applicability of our analysis to a broader set of constrained dynamics problems. Experimental results demonstrate that our method robustly handles complex knitwear simulation scenes with superior efficiency, stability, and physical fidelity compared to existing methods. With a matrix-free GPU parallelization, our method allows an interactive simulation rate of knitwear of multi-million DOFs.