BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/Los_Angeles X-LIC-LOCATION:America/Los_Angeles BEGIN:DAYLIGHT TZOFFSETFROM:-0800 TZOFFSETTO:-0700 TZNAME:PDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0700 TZOFFSETTO:-0800 TZNAME:PST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260417T190159Z LOCATION:West Building\, Ballroom AB DTSTART;TZID=America/Los_Angeles:20250810T180000 DTEND;TZID=America/Los_Angeles:20250810T204500 UID:siggraph_SIGGRAPH 2025_sess284@linklings.com SUMMARY:Papers Fast Forward DESCRIPTION:Sponsored by Adobe Research\n\nStressful Tree Modeling: Breaki ng Branches with Strands\n\nA novel approach for the computational modelin g of lignified tissues, such as those found in tree branches and timber, e xtends strand-based representation to describe biophysical processes at sh ort and long time scales. The computationally fast simulation leverages Co sserat rod physics and enables t...\n\n\nBosheng Li (Purdue University), N ikolas Schwarz (Kiel University), Wojtek Palubicki (AMU), Sören Pirk (Kiel University), Dominik L. Michels (King Abdullah University of Science and Technology (KAUST)), and Bedrich Benes (Purdue University)\n-------------- -------\nImage-Space Collage and Packing with Differentiable Rendering\n\n This work introduces an efficient image-space collage technique that optim izes geometric layouts using a differential renderer and hierarchical reso lution strategy. Our approach simplifies complex shape handling in image-s pace optimization, offering fixed computational complexity. Experiments sh ow o...\n\n\nZhenyu Wang and Min Lu (Shenzhen University)\n--------------- ------\nMonetGPT: Solving Puzzles Enhances MLLMs’ Image Retouching Skills\ n\nMonetGPT explores using multimodal large language models (MLLMs) for ph oto retouching by injecting domain knowledge via visual puzzles. These puz zles help MLLMs understand individual operations, visual aesthetics, and generate expert plans. Our procedural pipeline enables explainable edits w ith det...\n\n\nNiladri Shekhar Dutt (University College London (UCL)); Du ygu Ceylan (Adobe); and Niloy Mitra (University College London (UCL), Adob e)\n---------------------\nFeature-Aligned Parametrization in Penner Coord inates\n\nWe extend Penner-coordinate-based methods for seamless parametri zations to surfaces with sharp features to which the parametrization needs to be aligned. We describe a two-phase method to efficiently minimize fe ature constraint residual errors. We demonstrate that the resulting algor ithm works rob...\n\n\nRyan Capouellez and Rodrigo Singh (New York Univers ity), Martin Heistermann and David Bommes (University of Bern), and Denis Zorin (New York University)\n---------------------\nAnymate: A Dataset and Baselines for Learning 3D Object Rigging\n\nWe present the Anymate Datase t, a large-scale dataset of 230K 3D assets paired with expert-crafted rigg ing and skinning information---70 times larger than existing datasets. Usi ng this dataset, we propose a learning-based auto-rigging framework with t hree sequential modules for joint, connectivity, ...\n\n\nYufan Deng, Yuha o Zhang, and Chen Geng (Stanford University); Shangzhe Wu (Stanford Univer sity, University of Cambridge); and Jiajun Wu (Stanford University)\n----- ----------------\nHexHex: Highspeed Extraction of Hexahedral Meshes\n\nWe present HexHex, which extracts a hexahedral mesh from a locally injective integer-grid map. Key contributions include a conservative rasterization t echnique and a novel mesh data structure called propeller. Our algorithm i s significantly faster and uses less memory than the previous state-of-the -...\n\n\nTobias Kohler, Martin Heistermann, and David Bommes (University of Bern)\n---------------------\nPainless Differentiable Rotation Dynamics \n\nThis work introduces a forward and differentiable rigid-body dynamics framework using Lie-algebra rotation derivatives. The approach offers simp lified, compact derivatives, improved conditioning, and higher efficiency compared to traditional methods. Applications include fundamental rigid-bo dy probl...\n\n\nMagí Romanyà (Universidad Rey Juan Carlos) and Miguel A. Otaduy (Universidad Rey Juan Carlos, Meta Reality Labs Research)\n-------- -------------\nUnified Pressure, Surface Tension and Friction for SPH Flui ds\n\nWe present a new SPH approach to replicate the behavior of droplets and other smaller scale fluid bodies. For this, we develop a new implicit surface tension formulation and implement a Coulomb friction force at the fluid-solid interface. A strong coupling between both forces and pressure is achieve...\n\n\nTimo Probst and Matthias Teschner (University of Freibu rg) and Timo Probst\n---------------------\nPARC: Physics-based Augmentati on with Reinforcement Learning for Character Controllers\n\nPARC is a fram ework that enhances terrain traversal with machine learning and physics-ba sed simulation. By iteratively training a kinematic motion generator and s imulated motion tracker, PARC produces a character controller capable of t raversing complex environments using highly agile motor skills, ...\n\n\nM ichael Xu, Yi Shi, and KangKang Yin (Simon Fraser University) and Xue Bin Peng (Simon Fraser University, NVIDIA)\n---------------------\nPredicting Fabric Appearance Through Thread Scattering and Inversion\n\nThis paper pr esents a novel pipeline to digitize physical threads and predict fabric ap pearance before fabricating cloth samples, addressing a real need in the f ashion industry. It enables designers to make more informed material choic es, thereby promoting sustainable production, reducing costs, and...\n\n\n Mengqi Xia, Zhaoyang Zhang, Sumit Chaturvedi, and Yutong Yi (Yale Universi ty); Rundong Wu (ByteDance Inc.); and Holly Rushmeier and Julie Dorsey (Ya le University)\n---------------------\nFast Isotropic Median Filtering\n\n The median filter is a staple of computational image processing. Existing efficient methods share a common flaw, which is that they use a square ker nel, producing visual artifacts. Our method overcomes this limitation, ena bling fast and high-quality circular-kernel median filtering, across multi ple ...\n\n\nBen Weiss (Google Research)\n---------------------\nOn Planar Shape Interpolation With Logarithmic Metric Blending\n\nLogarithmic metri c blending enables smooth interpolation between planar shapes while boundi ng both conformal and area distortions. By blending symmetric positive def inite metrics in the log domain, our method geometrically interpolates dis tortions. This leads to natural transitions that outperform e...\n\n\nAlon Feldman and Mirela Ben-Chen (Technion – Israel Institute of Technology)\n ---------------------\nAppearance-aware Multi-view SVBRDF Reconstruction v ia Deep Reinforcement Learning\n\nThis paper introduces an appearance-awar e adaptive sampling method using deep reinforcement learning to optimize t he reconstruction of spatially-varying BRDFs from minimal images. By model ing the sampling as a sequential decision-making problem, the method ident ifies the next best view-lighting pair...\n\n\nPengfei Zhu, Jie Guo, Yifan Liu, Qi Sun, Yanxiang Wang, and Keheng Xu (Nanjing University); Ligang Li u (University of Science and Technology of China); and Yanwen Guo (Nanjing University)\n---------------------\nDualMS: Implicit Dual-Channel Minimal Surface Optimization for Heat Exchanger Design\n\nDualMS is a novel frame work for designing high-performance heat exchangers by directly optimizing the separation surface of two fluids using dual skeleton optimization and neural implicit functions. It offers greater topological flexibility than TPMS and achieves superior thermal performance with lo...\n\n\nWeizheng Z hang (Shandong University); Hao Pan (School of Software, Tsinghua Universi ty); Lin Lu, Xiaowei Duan, and Xin Yan (Shandong University); and Ruonan W ang and Qiang Du (Institute of Engineering Thermophysics, Chinese Academy of Sciences)\n---------------------\nHoLa: B-Rep Generation using a Holist ic Latent Representation\n\nWe introduce a novel representation for learni ng and generating Computer-Aided Design (CAD) models in the form of bounda ry representations (BReps). Our representation unifies the continuous geom etric properties of BRep primitives in different orders (e.g., surfaces an d curves) and their\ndiscrete top...\n\n\nYilin Liu, Duoteng Xu, and Xingy ao Yu (Shenzhen University); Xiang Xu (Simon Fraser University, Autodesk R esearch); Daniel Cohen-Or (Tel Aviv University, Shenzhen University); Hao Zhang (Simon Fraser University); and Hui Huang (Shenzhen University)\n---- -----------------\nFast Determination and Computation of Self-intersection s for NURBS Surfaces\n\nDetecting surface self-intersections is crucial fo r CAD modeling to prevent issues in simulation and manufacturing. This pap er presents an algebraic signature-based algorithm for fast determining se lf-intersections of NURBS surfaces. This signature is then recursively cro ss-used to compute the self-...\n\n\nKai Li and Xiaohong Jia (State Key La boratory of Mathematical Sciences, Academy of Mathematics and Systems Scie nce, Chinese Academy of Sciences; University of Chinese Academy of Science s); Falai Chen (University of Science and Technology of China); and Kai Li \n---------------------\nReStyle3D: Scene-Level Appearance Transfer with S emantic Correspondences\n\nRedesign spaces effortlessly-ReStyle3D transfor ms indoor scenes by transferring object-specific styles from a single refe rence image, preserving 3D coherence. Combining semantic-aware diffusion a nd depth guidance, it enables photo-realistic virtual staging—faithfully r edecorating furniture, te...\n\n\nLiyuan Zhu and Shengqu Cai (Stanford Uni versity), Shengyu Huang (NVIDIA), Gordon Wetzstein (Stanford University), Naji Khosravan (Zillow), and Iro Armeni (Stanford University)\n----------- ----------\nSynedelica: Mixed Reality Reimagined\n\nSynedelica challenges traditional approaches to mixed reality by transforming physical environme nts through a synesthetic experience. This artwork emphasizes the potentia l for immersive technology to mediate reality itself, fostering social int eraction and shared experiences. By reimagining how we p...\n\n\nJohn Desn oyers-Stewart (Univeristy of British Columbia) and Noah Miller and Bernhar d Riecke (Simon Fraser University)\n---------------------\nLayerFlow: A Un ified Model for Layer-aware Video Generation\n\nWe propose LayerFlow, a u nified framework for layer-aware video generation, enabling seamless creat ion of transparent foregrounds, clean backgrounds, and blended scenes. Wit h multi-stage training and LoRA techniques improving layer-wise video qual ity with limited data, it also supports variants lik...\n\n\nSihui Ji (The University of Hong Kong); Hao Luo (DAMO Academy, Alibaba Group); and Xi C hen, Yuanpeng Tu, Yiyang Wang, and Hengshuang Zhao (The University of Hong Kong)\n---------------------\nDiffusing Winding Gradients (DWG): A Parall el and Scalable Method for 3D Reconstruction from Unoriented Point Clouds\ n\nDiffusing Winding Gradients (DWG) efficiently reconstructs watertight 3 D surfaces from unoriented point clouds. Unlike conventional methods, DWG avoids solving linear systems or optimizing objective functions, enabling simple implementation and parallel execution. Our CUDA implementation on a n NVIDI...\n\n\nWeizhou Liu (Beijing Normal University); Jiaze Li (Nanyang Technological University); Xuhui Chen and Fei Hou (Institute of Software, Chinese Academy of Sciences; University of Chinese Academy of Sciences); Shiqing Xin (Shandong University); Xingce Wang and Zhongke Wu (Beijing Nor mal University); Chen Qian (SenseTime Group); Ying He (Nanyang Technologic al University College of Computing and Data Science); and Ying He\n------ ---------------\nLinear-Time Transport with Rectified Flows\n\nWe propose a simple, parallelizable algorithm inspired by rectified flows to match pr obability distributions. With linear-time complexity, it approximates opti mal transport by employing summed-area tables and direct particle advectio n. We illustrate our applications in stippling, mesh parameterizati...\n\n \nKhoa Do (University of Michigan), David Coeurjolly (CNRS - LIRIS), Poora n Memari (CNRS - LIX), and Nicolas Bonneel (CNRS - LIRIS)\n--------------- ------\nData-Efficient Discovery of Hyperelastic TPMS Metamaterials with E xtreme Energy Dissipation\n\nWe introduce a method for discovering novel m icroscale TPMS structures with high-energy dissipation. By combining a par ametric design space, empirical testing, and uncertainty-aware deep ensemb les with Bayesian optimization, we efficiently explore and discover struct ures with extreme energy dissipat...\n\n\nMaxine Perroni-Scharf (Massachus etts Institute of Technology (MIT)); Zachary Ferguson (Massachusetts Insti tute of Technology (MIT), CLO Virtual Fashion); and Thomas Butruille, Carl os Portela, and Mina Konaković Luković (Massachusetts Institute of Technol ogy (MIT))\n---------------------\nGeometric Contact Potential\n\nWe prese nt a systematic derivation of a continuum potential defined for smooth and piecewise smooth surfaces, by identifying a set of natural requirements f or contact potentials. Our potential is formulated independently of surfac e discretization and addresses the shortcomings of existing potential-...\ n\n\nZizhou Huang and Maxwell Paik (New York University); Zachary Ferguson (Massachusetts Institute of Technology, CLO Virtual Fashion); and Daniele Panozzo and Denis Zorin (New York University)\n---------------------\nA F ast Parallel Median Filtering Algorithm Using Hierarchical Tiling\n\nThis paper introduces a novel median filtering algorithm, using hierarchical ti ling to reduce redundant computations and achieve better complexity than p rior sorting-based methods. The paper discusses two implementations, for b oth small and larger kernel sizes, that outperform the state of the art b. ..\n\n\nLouis Sugy (NVIDIA)\n---------------------\nC5D: Sequential Contin uous Convex Collision Detection Using Cone Casting\n\nWe propose a fast, s ingle-threaded continuous collision detection (CCD) algorithm for convex s hapes under affine motion. By combining conservative advancement with a co ne-casting approach, it avoids primitive-level overhead and enables effici ent integration into intersection-free simulation methods ...\n\n\nXiaodi Yuan (University of California San Diego); Fanbo Xiang (Hillbot Inc.); Yin Yang (University of Utah); and Hao Su (University of California San Diego , Hillbot Inc.)\n---------------------\nTransforming Unstructured Hair Str ands into Procedural Hair Grooms\n\nRecent methods have been developed to reconstruct 3D hair strand geometry from images. We introduce an inverse h air grooming pipeline to transform these unstructured hair strands into pr ocedural hair grooms controlled by a small set of guide strands and artist -friendly grooming operators, enabling e...\n\n\nWesley Chang and Andrew R ussell (University of California San Diego); Stephane Grabli, Matt Chiang, Christophe Hery, and Doug Roble (Meta); Ravi Ramamoorthi and Tzu-Mao Li ( University of California San Diego); and Olivier Maury (Meta)\n----------- ----------\nRectangular Surface Parameterization\n\nThis paper presents a method for mapping curved surfaces to the plane without shear, enabling re ctangular parameterizations. It introduces a novel approach for computing integrable, orthogonal frame fields. The method improves mesh quality, sup ports rich user control, and outperforms existing techni...\n\n\nEtienne C orman (CNRS) and Keenan Crane (Carnegie Mellon University)\n-------------- -------\nAutoKeyframe: Autoregressive Keyframe Generation for Human Motion Synthesis and Editing\n\nWe present AutoKeyframe, a novel framework that simultaneously accepts dense and sparse control signals for motion generat ion by generating keyframes directly. Our method reduces manual efforts fo r keyframing while maintaining precise controllability, using an autoregre ssive diffusion model and a ne...\n\n\nBowen Zheng and Ke Chen (Zhejiang U niversity; State Key Laboratory of CAD&CG, Zhejiang University); Yuxin Yao (University of Cambridge, Department of Engineering); Zijiao Zeng and Xin wei Jiang (Tencent Games Digital Content Technology Center); He Wang (UCL Centre for Artificial Intelligence, Department of Computer Science, Univer sity College London); Joan Lasenby (University of Cambridge, Department of Engineering); and Xiaogang Jin (Zhejiang University; State Key Laboratory of CAD&CG, Zhejiang University)\n---------------------\nPDT: Point Distri bution Transformation with Diffusion Models\n\nPDT is a novel framework th at uses diffusion models to transform unstructured point clouds into seman tically meaningful and structured distributions, such as keypoints, joints , and feature lines. Exploring complex point distribution transformation, PDT captures fine-grained geometry and semantics, o...\n\n\nJionghao Wang (Texas A&M University); Cheng Lin (University of Hong Kong); Yuan Liu (HKU ST); Rui Xu and Zhiyang Dou (University of Hong Kong); Xiaoxiao Long (Nanj ing University); Haoxiang Guo (Skywork AI, Kunlun Inc.); Taku Komura (Univ ersity of Hong Kong); and Xin Li and Wenping Wang (Texas A&M University)\n ---------------------\nIMLS-Splatting: Efficient Mesh Reconstruction from Multi-view Images via Point Representation\n\nWe propose IMLS-Splatting, a n end-to-end multi-view mesh optimization method that leverages point clou ds for surface representation. By introducing a splatting-based differenti able IMLS algorithm, our approach efficiently converts point clouds into S DF and texture field, enabling multi-view mesh opt...\n\n\nKaizhi Yang (Un iversity of Science and Technology of China); Liu Dai and Isabella Liu (Un iversity of California San Diego); Xiaoshuai Zhang (Hillbot Inc.); Xiaoyan Sun and Xuejin Chen (University of Science and Technology of China); Zexi ang Xu (Hillbot Inc.); and Hao Su (University of California San Diego, Hil lbot Inc.)\n---------------------\nStochastic Barnes-Hut Approximation for Fast Summation on the GPU\n\nWe present a novel stochastic version of the Barnes-Hut approximation. Regarding the level-of-detail (LOD) family of a pproximations as control variates, we construct an unbiased estimator of t he kernel sum being approximated. Through several examples in graphics, we demonstrate that our method outpe...\n\n\nAbhishek Madan (University of T oronto); Nicholas Sharp, Francis Williams, and Ken Museth (NVIDIA); and Da vid I.W. Levin (University of Toronto, NVIDIA)\n---------------------\nJGS 2: Near Second-order Converging Jacobi/Gauss-Seidel for GPU Elastodynamics \n\nThis paper presents a new GPU simulation algorithm, which converges as fast as global Newton's method and as efficient as Jacobi method.\n\n\nLe i Lan (University of Utah; State Key Lab of CAD&CG, Zhejiang University); Zixuan Lu and Chun Yuan (University of Utah); Weiwei Xu (State Key Lab of CAD&CG, Zhejiang University, China); Hao Su (UCSD); Huamin Wang (Style3D R esearch); Chenfanfu Jiang (UCLA); and Yin Yang (University of Utah)\n----- ----------------\nElevating 3D Models: High-Quality Texture and Geometry R efinement from a Low-Quality Model\n\nElevate3D transforms low-quality 3D models into high-quality assets through iterative texture and geometry ref inement. At its core, HFS-SDEdit refines textures generatively while prese rving the input’s identity leveraging high-frequency guidance. The resulti ng texture then guides geometry refi...\n\n\nNuri Ryu, Jiyun Won, Jooeun S on, and Minsu Gong (POSTECH); Joo-Haeng Lee (Pebblous); and Sunghyun Cho ( POSTECH)\n---------------------\nSphere Carving: Bounding Volumes for Sign ed Distance Fields\n\nWe introduce Sphere Carving, a method for automatica lly computing bounding volumes for conservative implicit surface. SDF quer ies define a set of spheres, from which we extract intersection points, us ed to compute a bounding volume with guarantees. Sphere Carving is concept ually simple and independe...\n\n\nHugo Schott (INSA, Lyon; Adobe); Théo T honat and Thibaud Lambert (Adobe); Eric Guérin (INSA, Lyon); Eric Galin (U niversité Claude Bernard Lyon 1); and Axel Paris (Adobe)\n---------------- -----\nLarge-Scale Multi-Character Interaction Synthesis\n\nThis work intr oduces a conditional generative framework for large-scale multi-character interaction synthesis by facilitating natural interactive motions and tran sitions where characters are coordinated for new interactive partners, pro posing a coordinatable multi-character interaction space for int...\n\n\nZ iyi Chang (Durham University); He Wang (UCL Centre for Artificial Intellig ence, Department of Computer Science, University College London (UCL)); an d George Koulieris and Hubert Shum (Durham University)\n------------------ ---\nLAM: Large Avatar Model for One-shot Animatable Gaussian Head\n\nLAM is an innovative Large Avatar Model for animatable Gaussian head reconstru ction from a single image in seconds. Our Gaussian heads are immediately a nimatable and renderable without additional networks or post-processing. T his allows seamless integration into existing rendering pipelines, ensurin ...\n\n\nYisheng He, Xiaodong Gu, Xiaodan Ye, Chao Xu, Zhengyi Zhao, Yuan Dong, Weihao Yuan, Zilong Dong, and Liefeng Bo (Alibaba Group)\n---------- -----------\nTransparentGS: Fast Inverse Rendering of Transparent Objects with Gaussians\n\nWe propose TransparentGS, a fast inverse rendering pipel ine for transparent objects based on 3D-GS. The main contributions are thr ee-fold: efficient transparent Gaussian primitives for specular refraction , GaussProbe to encode ambient light and nearby contents, and the IterQuer y algorithm to reduce ...\n\n\nLetian Huang, Dongwei Ye, Jialin Dan, and C hengzhi Tao (State Key Lab for Novel Software Technology, Nanjing Universi ty); Huiwen Liu (TMCC, College of Computer Science, Nankai University); Ku n Zhou (State Key Lab of CAD&CG, Zhejiang University; Institute of Hangzho u Holographic Intelligent Technology); Bo Ren (TMCC, College of Computer S cience, Nankai University); and Yuanqi Li, Yanwen Guo, and Jie Guo (State Key Lab for Novel Software Technology, Nanjing University)\n-------------- -------\nFlexible 3D Cage-based Deformation via Green Coordinates on Bézie r Patches\n\nThis work constructs Green coordinates for cages composed of Bézier patches, which enables flexible deformations with curved boundaries . The high-order structure also allows us to create a compact curved cage for the input models. Additionally, this work proposes a global projection technique for pr...\n\n\nDong Xiao and Renjie Chen (University of Science and Technology of China)\n---------------------\nFacial Microscopic Struc tures Synthesis from a Single Unconstrained Image\n\nOur framework can eff iciently synthesize facial microstructure from an unconstrained facial ima ge via differentiable optimization. We propose neural wrinkle simulation f or differentiable microstructure parameterization, and direction distribut ion similarity to align features with blurry image patche...\n\n\nYouyang Du and Lu Wang (Shandong University) and Beibei Wang (Nanjing University)\ n---------------------\nDeformable Beta Splatting\n\nDeformable Beta Splat ting (DBS) is a novel approach for real-time radiance field rendering that leverages deformable Beta Kernels with adaptive frequency control for bot h geometry and color encoding. DBS captures complex geometries and lightin g with state-of-the-art fidelity, while only using 45% fe...\n\n\nRong Liu (USC Institute for Creative Technologies (ICT)), Dylan Sun (University of Southern California), Meida Chen (USC Institute for Creative Technologies (ICT)), Yue Wang (University of Southern California), and Andrew Feng (US C Institute for Creative Technologies (ICT))\n---------------------\nMomen t Bounds are Differentiable: Efficiently Approximating Measures in Inverse Rendering\n\nMeasures can be compactly represented and approximated using the theory of moments. This work proves that such moment-based representa tions are differentiable, leading to principled and efficient approaches f or approximating transmittance and visibility in differentiable rendering. \n\n\nMarkus Worchel and Marc Alexa (TU Berlin)\n---------------------\nFa st But Accurate: A Real-Time Hyperelastic Simulator with Robust Frictional Contact\n\nThis paper presents a GPU-friendly framework for real-time imp licit simulation of hyperelastic materials with frictional contacts. Utili zing a novel splitting strategy and efficient solver, the approach achieve s robust, high-performance simulation across various stiffness materials, handling large d...\n\n\nZiqiu Zeng (University of Strasbourg; Centre for Artificial Intelligence and Robotics, Hong Kong, CAS); Siyuan Luo and Fan Shi (National University of Singapore); and Zhongkai Zhang (Centre for Art ificial Intelligence and Robotics, Hong Kong, CAS)\n---------------------\ nInverse Design of Discrete Interlocking Materials with Desired Mechanical Behavior\n\nIn this paper, we present a computational approach for design ing Discrete Interlocking Materials (DIM) with desired mechanical properti es. We demonstrate the effectiveness of our method by designing discrete i nterlocking materials with diverse limit profiles for in- and out-of-plane deformation and ...\n\n\nPengbin Tang, Bernhard Thomaszewski, Stelian Cor os, and Bernd Bickel (ETH Zürich)\n---------------------\nSpherical Lighti ng with Spherical Harmonics Hessian\n\nWe introduce spherical harmonics He ssian and solid spherical harmonics, a variant of spherical harmonics, to compute the spherical harmonics Hessian efficiently and accurately to the computer graphics community. These mathematical tools are used to develop an analytical representation of the Hessian...\n\n\nKei Iwasaki (Saitama U niversity, Prometech CG Research) and Yoshinori Dobashi (Hokkaido Universi ty, Prometech CG Research)\n---------------------\nColorSurge: Bringing Vi brancy and Efficiency to Automatic Video Colorization via Dual-Branch Fusi on\n\nWe propose ColorSurge, a lightweight dual-branch network for end-to- end video colorization. It delivers vivid, accurate, and real-time results from grayscale input, and is easily extensible for high-quality performan ce at low computational cost.\n\n\nHongbo Zhao, Jiaxing Li, Peiyi Zhang, P eng Xiao, Jianxin Lin, and Yijun Wang (Hunan University)\n---------------- -----\nMaterialPicker: Multi-Modal DiT-Based Material Generation\n\nMateri alPicker is a multi-modal material generation model that creates high-qual ity material maps from images and/or text by fine-tuning a video diffusion model. It robustly extracts materials from real-world photos, even with d istortion or occlusion, enhancing fidelity, diversity, and efficiency in.. .\n\n\nXiaohe Ma (State Key Lab of CAD&CG, Zhejiang University); Valentin Deschaintre, Milos Hasan, and Fujun Luan (Adobe Research); Kun Zhou (State Key Lab of CAD&CG, Zhejiang University; ZJU-FaceUnity Joint Lab of Intell igent Graphics); Hongzhi Wu (State Key Lab of CAD&CG, Zhejiang University) ; and Yiwei Hu (Adobe Research)\n---------------------\nCMD: Controllable Multiview Diffusion for 3D Editing and Progressive Generation\n\nCMD revol utionizes 3D generation by enabling flexible local editing of 3D models fr om a single rendering, as well as progressive, interactive creation of com plex 3D scenes. At its core, CMD leverages a conditional multiview diffusi on model to seamlessly modify/add new components—enhancing cont...\n\n\nPe ng Li (Hong Kong University of Science and Technology), Suizhi Ma (Johns H opkins University), Jialiang Chen and Yuan Liu (Hong Kong University of Sc ience and Technology), Congyi Zhang (Univeristy of British Columbia), Wei Xue and Wenhan Luo (Hong Kong University of Science and Technology), Alla Sheffer (Univeristy of British Columbia), Wenping Wang (Texas A&M Universi ty), and Yike Guo (Hong Kong University of Science and Technology)\n------ ---------------\nEnd-to-end Surface Optimization for Light Control\n\nDesi gning freeform surfaces to reflect or refract light to achieve target ligh t distributions is a challenging inverse problem. We propose an end-to-end optimization strategy using a novel differentiable rendering model driven by image errors, combined with face-based optimal transport initializatio ...\n\n\nYuou Sun (University of Science and Technology of China), Bailin Deng (Cardiff University), Juyong Zhang (University of Science and Technol ogy of China), and Yuou Sun\n---------------------\nText-based Animatable 3D Avatars with Morphable Model Alignment\n\nAnimPortrait3D is a novel met hod for text-based, realistic, animatable 3DGS avatar generation with morp hable model alignment. To address ambiguities in diffusion predictions dur ing 3D distillation, we introduce key strategies: initializing a 3D avatar with robust appearance and geometry, and leverag...\n\n\nYiqian Wu (ETH Z ürich; State Key Lab of CAD and CG, Zhejiang University); Malte Prinzler ( ETH Zürich); Xiaogang Jin (State Key Lab of CAD and CG, Zhejiang Universit y); and Siyu Tang (ETH Zürich)\n---------------------\nFLoD: Integrating F lexible Level of Detail into 3D Gaussian Splatting for Customizable Render ing\n\nFlexible Level of Detail (FLoD) integrates the concept of LoD into 3DGS using a multi-level representation built with 3D Gaussian scale const raints and level-by-level training strategy. FLoD enables flexible renderi ng through single-level or selective rendering for optimal image quality u nder varyin...\n\n\nYunji Seo, Young Sun Choi, HyunSeung Son, and Youngjun g Uh (Yonsei University)\n---------------------\nGuided Lens Sampling for Efficient Monte Carlo Circle-of-Confusion Rendering\n\nA guided lens sampl ing technique that improves Monte Carlo rendering of depth-of-field by pro jecting a global 3D radiance field into lens space via bipolar-cone projec tion. This method efficiently targets high-contribution regions, significa ntly reducing noise and improving convergence for circle-of...\n\n\nJiawei Huang (International Digital Economy Academy), Shaokun Zheng and Kun Xu ( Tsinghua University), Yoshifumi Kitamura (Tohoku University), and Jiaping Wang (International Digital Economy Academy)\n---------------------\nPolyn omial 2D Biharmonic Coordinates for High-order Cages\n\nWe propose polynom ial 2D biharmonic coordinates for closed high-order cages containing polyn omial curves of any order by extending the classical\n2D biharmonic coordi nates using high-order BEM. When applying our coordinate\nto 2D cage-based deformation, users manipulate the \Bezier\ncontrol points to q...\n\n\nSh ibo Liu, Tielin Dai, Ligang Liu, and Xiao-Ming Fu (University of Science a nd Technology of China)\n---------------------\nDifferentiable Geometric A coustic Path Tracing using Time-Resolved Path Replay Backpropagation\n\nIn troducing differentiable path tracing for geometric acoustics with an effi cient gradient algorithm based on path replay backpropagation. The system computes derivatives of output spectrograms with respect to arbitrary scen e parameters (materials, geometry, emitters, microphones) within the frame wo...\n\n\nUgo Finnendahl, Markus Worchel, Tobias Jüterbock, Daniel Wujeck i, Fabian Brinkmann, Stefan Weinzierl, and Marc Alexa (TU Berlin)\n------- --------------\nMulti Layered Autonomy and AI Ecologies in Robotic Art Ins tallations\n\n“Symbiosis of Agents” merges AI-driven multi-agent robotics with immersive environments, exploring the delicate balance of machine age ncy and artist authorship through emergent behaviors in self-organized AI ecologies. Its layered approach—micro-level strategie, meso-level drives, ...\n\n\nBaoyang Chen (Hong Kong University of Science and Technology, Cen tral Academy of Fine Arts) and Xian Xu and Huamin Qu (Hong Kong University of Science and Technology)\n---------------------\nPiecewise Ruled Approx imation for Freeform Mesh Surfaces\n\nWe propose a method to approximate a rbitrary freeform surface meshes with piecewise ruled surfaces. Our approa ch optimizes mesh shape and ruling direction field simultaneously, extract s patch topology, and refines ruling positions and orientations. The techn ique effectively approximates diverse free...\n\n\nYiling Pan, Zhixin Xu, and Bin Wang (Tsinghua University) and Bailin Deng (Cardiff University)\n- --------------------\nSpatiotemporally Consistent Indoor Lighting Estimati on with Diffusion Priors\n\nWe propose a method for estimating spatiotempo rally varying indoor lighting from videos using a continuous light field r epresented as an MLP. By leveraging 2D diffusion priors fine-tuned to pred ict lighting jointly at multiple locations, our approach achieves superior performance and zero-shot gener...\n\n\nMutian Tong, Rundi Wu, and Changx i Zheng (Columbia University)\n---------------------\nSingle Edge Collapse Quad-Dominant Mesh Reduction\n\nA simple and robust modification to trian gle mesh reduction bridges the gap for what artists want in quad-dominant mesh reduction, preserving symmetry, topology, and joints without sacrific ing geometric quality, allowing for high-quality level-of-detail meshes at no cost compared to what was done be...\n\n\nJulian Knodt (LightSpeed Stu dios)\n---------------------\n3D-Fixup: Advancing Photo Editing with 3D Pr iors\n\n3D-Fixup enables realistic 3D-aware photo editing by leveraging 3D priors and a novel data pipeline that extracts training pairs from real-w orld videos. Its feed-forward architecture supports efficient, high-qualit y edits involving complex 3D transformations while preserving object ident ity, outperf...\n\n\nYen-Chi Cheng (University of Illinois Urbana-Champaig n, Adobe Research); Krishna Kumar Singh and Jae Shin Yoon (Adobe Research) ; Alexander Schwing and Liang-Yan Gui (University of Illinois Urbana-Champ aign); and Matheus Gadelha, Paul Guerrero, and Nanxuan Zhao (Adobe Researc h)\n---------------------\nOne Model to Rig Them All: Diverse Skeleton Rig ging with UniRig\n\nManual 3D rigging is slow. UniRig introduces a unified learning framework for automatic skeletal rigging. Trained on our large, diverse Rig-XL dataset, it uses an autoregressive model and cross-attentio n to accurately rig various characters and objects, significantly outperfo rming prior methods and ...\n\n\nJia-Peng Zhang, Cheng-Feng Pu, and Meng-H ao Guo (CS Dept, Tsinghua University); Yan-Pei Cao (VAST); and Shi-Min Hu (CS Dept, Tsinghua University)\n---------------------\nFluid Simulation on Compressible Flow Maps\n\nWe present a unified compressible flow map fram ework based on Lagrangian path integrals, enabling conservative density-en ergy transport and flexible pressure treatments. Validated on diverse syst ems—from shocks to shallow water—it captures complex flow features like vo rtices and wave int...\n\n\nDuowen Chen and Zhiqi Li (Georgia Institute of Technology); Taiyuan Zhang and Jinjin He (Dartmouth College); Junwei Zhou (University of Michigan, Purdue University); Bart G. van Bloemen Waanders (Sandia National Laboratories); and Bo Zhu (Georgia Institute of Technolo gy)\n---------------------\nClosed-form Generalized Winding Numbers of Rat ional Parametric Curves for Robust Containment Queries\n\nWe derive closed -form expressions for GWNs of rational parametric curves for robust contai nment queries. \nOur closed-form expression enables efficient computation of GWN, even if the query points are located on the rational curve. We als o derive the derivatives of GWN for other applications.\n\n\nShibo Liu, Li gang Liu, and Xiao-Ming Fu (University of Science and Technology of China) \n---------------------\n3DGS2: Near Second-order Converging 3D Gaussian S platting\n\nThis paper introduces a nearly second-order convergent trainin g algorithm for 3D Gaussian Splatting that exploits independent kernel att ributes and sparse coupling across images. By constructing and solving sma ll Newton systems for parameter groups, it achieves about an-order faster training while m...\n\n\nLei Lan (University of Utah; State Key Lab of CAD and CG, Zhejiang University); Tianjia Shao (Zhejiang University); Zixuan Lu and Yu Zhang (University of Utah); Chenfanfu Jiang (UCLA); and Yin Yang (University of Utah)\n---------------------\nIP-Prompter: Training-Free T heme-Specific Image Generation via Dynamic Visual Prompting\n\nThis paper presents T-Prompter, a method for visually prompting generative models to enable continuous image generation for specific themes, characters, and sc enes. It introduces Dynamic Visual Prompting to enhance generation accurac y and quality, outperforming existing methods in maintaining charac...\n\n \nYuxin Zhang, Minyan Luo, and Weiming Dong (MAIS, Institute of Automation , Chinese Academy of Sciences; School of Artificial Intelligence, Universi ty of Chinese Academy of Sciences); Xiao Yang, Haibin Huang, and Chongyang Ma (ByteDance Inc.); Oliver Deussen (University of Konstanz); Tong-Yee Le e (National Cheng-Kung University); and Changsheng Xu (MAIS, Institute of Automation, Chinese Academy of Sciences; School of Artificial Intelligence , University of Chinese Academy of Sciences)\n---------------------\nColla borative On-Sensor Array Cameras\n\nWe introduce a collaborative metalens array comprising over 100-million nanopillars for broadband imaging. The p roposed array camera is only a few millimeters flat and employs a non-gene rative reconstruction method, which performs favorably and without halluci nations, irrespective of the scene illum...\n\n\nJipeng Sun (Princeton Uni versity), Kaixuan Wei (King Abdullah University of Science and Technology (KAUST)), Thomas Eboli (Université Paris-Saclay), Congli Wang and Cheng Zh eng (Princeton University), Zhihao Zhou and Arka Majumdar (University of W ashington), Wolfgang Heidrich (King Abdullah University of Science and Tec hnology (KAUST)), and Felix Heide (Princeton University)\n---------------- -----\nC-Tubes: Design and Optimization of Tubular Structures Composed of Developable Strips\n\nC-tubes are 3D tubular structures made of developabl e strips. We introduce an algorithm to construct C-tubes while guaranteein g exact surface developability and an optimization method for design explo ration. Applications span architecture, engineering, and product design. W e present prototypes showc...\n\n\nMichele Vidulis, Klara Mundilova, Quent in Becker, Florin Isvoranu, and Mark Pauly (EPFL)\n---------------------\n Generative detail enhancement for physically based materials\n\nWe present a tool for enhancing the detail of physically based materials using an of f-the-shelf diffusion model and inverse rendering. Our goal is to enhance the visual fidelity of materials with detail that is often tedious to auth or, by adding signs of wear, aging, weathering, etc.\n\n\nSaeed Hadadan (U niversity of Maryland College Park, NVIDIA); Benedikt Bitterli, Tizian Zel tner, Jan Novák, Fabrice Rousselle, Jacob Munkberg, Jon Hasselgren, and Ba rtlomiej Wronski (NVIDIA); and Matthias Zwicker (University of Maryland Co llege Park)\n---------------------\nMultiple Importance Reweighting for Pa th Guiding\n\nWe combine the estimates generated in each guiding iteration , leveraging the importance distributions from multiple guiding iterations . We demonstrate that our path-level reweighting makes guiding algorithms less sensitive to noise and overfitting in distributions.\n\n\nZhimin Fan, Yiming Wang, and Chenxi Zhou (Nanjing University); Ling-Qi Yan (Universit y of California Santa Barbara); and Yanwen Guo and Jie Guo (Nanjing Univer sity)\n---------------------\nReenact Anything: Semantic Video Motion Tran sfer Using Motion-Textual Inversion\n\nReenact Anything introduces a unifi ed framework for semantic motion transfer, covering applications from full -body and face reenactment to controlling the motion of inanimate objects and the camera. Thereby, motions are represented using text/image embeddin gs of an image-to-video diffusion model and...\n\n\nManuel Kansy (ETH Züri ch, Disney Research Studios); Jacek Naruniec and Christopher Schroers (Dis ney Research Studios); Markus Gross (ETH Zürich, Disney Research Studios); and Romann Weber (Disney Research Studios)\n---------------------\nGenera tive Video Matting\n\nLimited high-quality ground-truth data hinders tradi tional video matting's real-world application. This work tackles this by a dvocating for large-scale training with diverse synthetic segmentation and matting data. A novel generative pipeline is also introduced to predict t emporally consistent alpha...\n\n\nYongtao Ge (The University of Adelaide, Zhejiang University); Kangyang Xie, Guangkai Xu, and Mingyu Liu (Zhejiang University); Li Ke, Longtao Huang, and Hui Xue (Alibaba Group); Hao Chen (Zhejiang University); and Chunhua Shen (Zhejiang University of Technology , Zhejiang University)\n---------------------\nwave2weave: A Procedural We ave Data Generation\n\nThis paper presents a procedural data generation me thod for Jacquard weaving that uses matrix computations to create textiles with complex shaded patterns and a triple-layer structure. Employing this method, the authors creatively applied noise functions to weave design an d produced a textile artwor...\n\n\nTatsuki Hayama (Keio University) and K otaro Uchibe (Hosoo Co.,Ltd)\n---------------------\n3D Stylization via La rge Reconstruction Model\n\nGiven a 3D object representing the source cont ent and a reference style image, our method performs 3D stylization with a large pre-trained reconstruction model. This is achieved in a zero-shot m anner, with no training or test time optimization required, while deliveri ng superior visual fidelity and ...\n\n\nIpek Oztas (Bilkent University), Duygu Ceylan (Adobe Research), and Aysegul Dundar (Bilkent University)\n-- -------------------\nFluid Simulation on Vortex Particle Flow Maps\n\nWe p resent the Vortex Particle Flow Map (VPFM) method, which revitalizes the t raditional Vortex-In-Cell approach for computer graphics. By evolving vort icity and higher-order quantities along particle flow maps, our method ach ieves significantly improved long-term stability and vorticity preservatio ...\n\n\nSinan Wang (Georgia Institute of Technology); Junwei Zhou (Univer sity of Michigan Ann Arbor, Purdue University); Fan Feng (Dartmouth Colleg e); and Zhiqi Li, Yuchen Sun, Duowen Chen, Greg Turk, and Bo Zhu (Georgia Institute of Technology)\n---------------------\nVariable Shared Template for Consistent Non-rigid ICP\n\nWe propose a novel ICP framework that join tly optimizes a shared template and instance-wise deformations. Our approa ch automatically captures common shape features from input shapes, achievi ng state-of-the-art accuracy and consistency while eliminating the need to carefully select a preset template ...\n\n\nYucheol Jung, Hyomin Kim, Hye jeong Yoon, Yoonha Hwang, and Seungyong Lee (POSTECH)\n------------------- --\nMotionCanvas: Cinematic Shot Design with Controllable Image-to-Video G eneration\n\nMotionCanvas enables intuitive cinematic shot design in image -to-video generation by letting users control both camera movements and ob ject motions in a 3D-aware scene. Combining classical graphics with modern diffusion models, it translates motion intentions into spatiotemporal sig nals—withou...\n\n\nJinbo Xing (The Chinese University of Hong Kong, Adobe Research); Long Mai, Cusuh Ham, Jiahui Huang, and Aniruddha Mahapatra (Ad obe Research); Chi-Wing Fu (The Chinese University of Hong Kong); Tien-Tsi n Wong (Monash University); and Feng Liu (Adobe Research)\n--------------- ------\nFast Physics-Based Modeling of Knots and Ties using Templates\n\nW e propose a physics-based modeling system for knots and ties using pipe-li ke parametric templates, defined by Bézier curves and adaptive radii for f lexible, intersection-free shapes. Our method maps cloth regions from UV s pace into 3D knot forms via a penetration-free initialization and supports qu...\n\n\nDewen Guo (Peking University, Style3D Research); Zhendong Wang and Zegao Liu (Style3D Research); Sheng Li and Guoping Wang (Peking Unive rsity); Yin Yang (University of Utah); and Huamin Wang (Style3D Research)\ n---------------------\nDesignManager: An Agent-Powered Copilot for Design ers to Integrate AI Design Tools into Creative Workflows\n\nDesignManager is an AI-powered design support system that functions as an interactive co pilot throughout the creative workflow. With node-based visualization of d esign evolution and conversational interaction modes, it helps designers t rack, modify, and branch their processes while providing context...\n\n\nW eitao You, Yinyu Lu, Zirui Ma, Nan Li, Mingxu Zhou, Xue Zhao, Pei Chen, an d Lingyun Sun (Zhejiang University)\n---------------------\nInstantRestore : Single-Step Personalized Face Restoration with Shared-Image Attention\n\ nInstantRestore is a fast, personalized face restoration framework that us es a single-step diffusion model with an extended self-attention mechanism to match low-quality image patches to high-quality reference patches. Lev eraging implicit correspondences in the denoising network, we efficiently trans...\n\n\nHoward Zhang (Snap, University of California Los Angeles); Y uval Alaluf (Tel Aviv University); Sizhuo Ma (Snap); Achuta Kadambi (Unive rsity of California Los Angeles); and Jian Wang and Kfir Aberman (Snap)\n- --------------------\nPhysics-inspired Estimation of Optimal Cloth Mesh Re solution\n\nWe propose a method to estimate optimal cloth mesh resolution based on material stiffness and boundary conditions like shirring or stitc hing, and dynamic wrinkles from motion-induced collisions. To ensure smoot h resolution transitions, we calculate transition distances and generate a mesh sizing map...\n\n\nDiyang Zhang, Zhendong Wang, and Zegao Liu (Style 3D Research); Xinming Pei (State Key Laboratory of CAD & CG, Zhejiang Univ ersity; Style3D Research); Weiwei Xu (State Key Laboratory of CAD & CG, Zh ejiang University); and Huamin Wang (Style3D Research)\n------------------ ---\nEncoded Marker Clusters for Auto-Labeling in Optical Motion Capture\n \nMarker-based optical motion capture (MoCap) is critical for virtual prod uction and movement sciences. We propose a novel framework for MoCap auto- labeling and matching using uniquely coded clusters of reflective markers (AEMCs). Compared to commercial software, our method achieves higher label ing ac...\n\n\nHao Wang (Beihang University, Beijing Jiaotong University); Taogang Hou, Tianhui Liu, and Jiaxin Li (Beijing Jiaotong University); Ti anmiao Wang (Beihang University); and Hao Wang and Taogang Hou\n---------- -----------\nLifting the Winding Number: Precise Discontinuities in Neural Fields for Physics Simulation\n\nWe designed a neural field capable of ca pturing a diverse family of discontinuities, enabling the simulation of cu ts in thin-walled deformable structures. By lifting input coordinates usin g generalized winding numbers, our approach models discontinuities explici tly and flexibly, supporting accurate,...\n\n\nYue Chang, Mengfei Liu, and Zhecheng Wang (University of Toronto); Peter Yichen Chen (MIT CSAIL); and Eitan Grinspun (University of Toronto)\n---------------------\nMotion Inv ersion for Video Customization\n\nWe propose Motion Embeddings for video g eneration, enabling precise motion in video transfer across diverse scenes and objects. These embeddings disentangle motion from appearance, preserv ing original dynamics while adapting to new prompts. Experiments show that our method achieved high-quality, pro...\n\n\nLuozhou Wang, Ziyang Mai, a nd Guibao Shen (Hong Kong University of Science and Technology, Guangzhou) ; Yixun Liang (Hong Kong University of Science and Technology); Xin Tao, P engfei Wan, and Di Zhang (Kuaishou Technology); Yijun Li (Adobe Research); and Yingcong Chen (Hong Kong University of Science and Technology, Guangz hou)\n---------------------\nNoise-Coded Illumination for Forensic and Pho tometric Video Analysis\n\nVideo forensics, which focuses on identifying f ake or manipulated video, is becoming increasingly difficult with the deve lopment of more advanced video editing techniques. We show how coding near -imperceptible, noise-like modulations into the illumination of a scene ca n create information asymmetry ...\n\n\nPeter Michael (Cornell University) , Zekun Hao (Cornell Tech), Serge Belongie (University of Copenhagen), Abe Davis (Cornell University), and Peter Michael\n---------------------\nPra ctical Inverse Rendering of Textured and Translucent Appearance\n\nThis wo rk addresses recovering textured materials using inverse rendering. Our La placian mipmapping improves the reconstruction of high-resolution textures . We also propose a novel gradient computation that enables efficiently re constructing textured, path-traced subsurface scattering. The methods a... \n\n\nPhilippe Weier (Saarland University, Google); Jérémy Riviere, Ruslan Guseinov, and Stephan Garbin (Google); Philipp Slusallek (Saarland Univer sity, DFKI); Bernd Bickel (Google, ETH Zürich); and Thabo Beeler and Delio Vicini (Google)\n---------------------\nGaussian Fluids: A Grid-Free Flui d Solver based on Gaussian Spatial Representation\n\nWe present a grid-fre e fluid simulator featuring a novel Gaussian spatial representation (GSR) for velocity field. The advantages of GSR over traditional Lagrangian/Eule rian data structures are 4-folded: memory compactness, spatial adaptivity, vorticity preservation and continuous differentiability....\n\n\nJingrui Xing (School of Intelligence Science and Technology, Peking University); B in Wang (Independent); and Mengyu Chu and Baoquan Chen (Peking University, State Key Laboratory of General Artificial Intelligence)\n--------------- ------\nHybrid Tours: A Clip-based System for Authoring Long-take Touring Shots\n\nWe propose Hybrid Tours, a hybrid approach to creating long-take shots by combining short video clips in a virtual interface. We show that Hybrid Tours makes capturing long-take touring shots much easier, and that clip-based authoring and reconstruction lead to higher-fidelity results a t lower compu...\n\n\nXinrui Liu, Longxiulin Deng, and Abe Davis (Cornell University)\n---------------------\nHyborg Agency: Fostering AI Agents Thr ough Community Conversations in a Digital Forest\n\nHyborg Agency proposes an artistic perspective on AI agents: We can design AI agents that mainta in their distinct non-human nature while meaningfully participating in hum an social contexts.\nPresenting AI agents as mechanical deer nurtured by c ommunity conversations, this computational ecosystem demo...\n\n\nYuqian S un (Computer Science Research Centre, Royal College of Art); Chenhang Chen g and Chuyan Xu (Individual); Chang Hee Lee (Korea Advanced Institute of S cience and Technology (KAIST)); and Ali Asadipour (Computer Science Resear ch Centre, Royal College of Art)\n---------------------\nDesigning Pin-pre ssion Gripper and Learning its Dexterous Grasping with Online In-hand Adju stment\n\nWe introduce a pin-pression gripper featuring parallel-jaw finge rs with 2D arrays of independently extendable pins, allowing instant shape adaptation to target object geometry and dynamic in-hand re-orientation f or enhanced grasp stability. Reinforcement learning with curriculum-based training enabl...\n\n\nHewen Xiao and Xiuping Liu (Dalian University of Te chnology), Hang Zhao (Wuhan University), Jian Liu (Shenyang University of Technology), and Kai Xu (National University of Defense Technology (NUDT)) \n---------------------\nLearning to Draw Is Learning to See: Analyzing Ey e Tracking Patterns for Assisted Observational Drawing\n\nWe present an im age-to-image drawing setup capturing eye tracking and stroke data across 1 56 drawings from 10 artists. Our findings reveal consistent fixation patte rns, strong gaze–stroke correlations, and structured drawing sequences, of fering new insights into professional observation strate...\n\n\nFengqi LI U, Longji Huang, and Zhengyu Huang (The Hong Kong University of Science an d Technology (Guangzhou)) and Zeyu Wang (The Hong Kong University of Scien ce and Technology (Guangzhou), The Hong Kong University of Science and Tec hnology)\n---------------------\nPLT: Part-Wise Latent Tokens as Adaptable Motion Priors for Physically Simulated Characters\n\nWe introduce a physi cally-based character animation framework that exploits part-wise latent t okens. The novel structured decomposition enables dynamic exploration to s tably adapt to diverse unseen scenarios. Additional refinement networks im prove overall motion quality. We show superior performance...\n\n\nJinseok Bae, Younghwan Lee, Donggeun Lim, and Young Min Kim (Seoul National Unive rsity)\n---------------------\nin(A)n(I)mate - AI-Mediated Conversations w ith Inanimate Objects\n\n2025, rumored to be the "year of AI agents," the artwork in(A)n(I)mate envisions a future where AI systems act behind the s cenes of objects, providing them agency and performativity, animating them , and bringing them closer to human awareness. By inviting conversations w ith everyday objects, in(A)n(...\n\n\nAvital Meshi (University of Californ ia Davis) and Adam Wright (UC Davis)\n---------------------\nRags2Riches: Computational Garment Reuse\n\nWe present the first algorithm to automatic ally compute sewing patterns for\nupcycling existing garments into new des igns. Our algorithm takes as input\ntwo garment designs along with their c orresponding sewing patterns and\ndetermines how to cut one of them to mat ch the other by following garment\nreus...\n\n\nAnran Qi (INRIA, Universit é Côte d'Azur); Nico Pietroni (University of Technology Sydney); Maria Kor osteleva (ETH Zurich, Meshcapade); Olga Sorkine-Hornung (ETH Zurich); and Adrien Bousseau (INRIA, Université Côte d'Azur)\n---------------------\nRi gAnything: Template-Free Autoregressive Rigging for Diverse 3D Assets\n\nR igAnything is a transformer-based model that autoregressively generates 3D rigging without templates. It sequentially predicts joints and skeleton t opology while assigning skinning weights, working on objects in any pose. It’s 20× faster than existing methods, completing rigging in under 2 se... \n\n\nIsabella Liu (University of California San Diego); Zhan Xu, Yifan Wa ng, and Hao Tan (Adobe Research); Zexiang Xu (Hillbot Inc.); Xiaolong Wang (University of California San Diego); Hao Su (University of California Sa n Diego, Hillbot Inc.); and Zifan Shi (Adobe Research)\n------------------ ---\nDynamic Concepts Personalization from Single Videos\n\nPersonalizing text-to-video models is challenging because dynamic concepts require captu ring both appearance and motion. We propose Set-and-Sequence, a framework that personalizes DiT-based video models by first learning an identity LoR A basis from unordered frames, then fine-tuning coefficients wit...\n\n\nR ameen Abdal, Or Patashnik, Ivan Skorokhodov, Willi Menapace, Aliaksandr Si arohin, Sergey Tulyakov, Daniel Cohen-Or, and Kfir Aberman (Snap)\n------- --------------\nOn-the-fly Reconstruction for Large-Scale Novel View Synth esis from Unposed Images\n\nWe propose a fast, on-the-fly 3D Gaussian Spla tting method that jointly estimates poses and reconstructs scenes. Through fast pose initialization, direct primitive sampling, and scalable cluster ing and merging, it efficiently handles diverse ordered image sequences of arbitrary length.\n\n\nAndreas Meuleman, Ishaan Shah, and Alexandre Lanvi n (INRIA, Université Côte d'Azur); Bernhard Kerbl (TU Wien); and George Dr ettakis (INRIA, Université Côte d'Azur)\n---------------------\nCineMaster : A 3D-Aware and Controllable Framework for Cinematic Text-to-Video Genera tion\n\nA 3D-aware and controllable text-to-video generation method allows users to manipulate objects and camera jointly in 3D space for high-quali ty cinematic video creation.\n\n\nQinghe Wang (Dalian University of Techno logy); Yawen Luo (The Chinese University of Hong Kong); Xiaoyu Shi (Kuaish ou Technology); Xu Jia and Huchuan Lu (Dalian University of Technology); T ianfan Xue (The Chinese University of Hong Kong); and Xintao Wang, Pengfei Wan, Di Zhang, and Kun Gai (Kuaishou Technology)\n---------------------\n VideoAnydoor: High-fidelity Video Object Insertion with Precise Motion Con trol\n\nWe propose VideoAnydoor, a zero-shot video object insertion framew ork with high-fidelity detail preservation and precise motion control, whe re a pixel warper and a image-video mix-training strategy are designed to warp the pixel details according to the trajectories. VideoAnydoor demonst rates signif...\n\n\nYuanpeng Tu (The University of Hong Kong); Luo Hao (D AMO Academy, Alibaba Group); Chen Xi and Sihui Ji (The University of Hong Kong); Xiang Bai (Huazhong University of Science and Technology); and Zhao Hengshuang (The University of Hong Kong)\n---------------------\nDivide-a nd-Conquer Embedding\n\nThe paper proposes a construction algorithm based on a divide-and-conquer strategy to map a disk-topology triangular mesh on to any convex polygon., which supports arbitrary numerical precision and e xact arithmetic. Under exact arithmetic, it strictly guarantees a bijectio n for any mesh and convex po...\n\n\nYuan-Yuan Cheng, Qing Fang, Ligang Li u, and Xiao-Ming Fu (University of Science and Technology of China)\n----- ----------------\nDreamMask: Boosting Open-vocabulary Panoptic Segmentatio n with Synthetic Data\n\nTo address a lack of generalization to novel clas ses, we propose DreamMask, which systematically explores data generation i n the open-vocabulary setting, and how to train the model with both real a nd synthetic data. It significantly simplifies the collection of large-sca le training data, serving as ...\n\n\nYuanpeng Tu and Xi Chen (The Univers ity of Hong Kong), Ser-Nam Lim (UCF), and Hengshuang Zhao (The University of Hong Kong)\n---------------------\n4D Gaussian Videos with Motion Layer ing\n\nWe present 4D Gaussian Video (4DGV) for high-quality, low-storage v olumetric video reconstruction and real-time streaming. Our method effecti vely handles complex motion and enables effective motion compression, achi eving superior performance in both reconstruction quality and storage effi ciency.\n\n\nPinxuan Dai, Peiquan Zhang, and Zheng Dong (Zhejiang Universi ty); Ke Xu (City University of Hong Kong); Yifan Peng (The University of H ong Kong); Dandan Ding (Hangzhou Normal University); Yujun Shen (Ant Group ); Yin Yang (The University of Utah); Xinguo Liu (Zhejiang University); Ry nson W.H. Lau (City University of Hong Kong); and Weiwei Xu (State Key Lab CAD&CG, Zhejiang University)\n---------------------\nMAGNET: Muscle Activ ation Generation Networks for Diverse Human Movement\n\nWe introduce MAGNE T (Muscle Activation Generation Networks), a scalable framework for recons tructing full-body muscle activations across diverse human movements, whic h also includes distilled models for solving downstream tasks or generatin g real-time muscle activations—even on edge devices. T...\n\n\nJungnam Par k, Euikyun Jung, Jehee Lee, and Jungdam Won (Seoul National University)\n- --------------------\nMiSo: A DSL for robust and efficient MINIMIZE and SO LVE problems\n\nMany problems in graphics can be formulated as a non-linea rly constrained global minimization (MINIMIZE), or solution of a system of non-linear constraints (SOLVE). We introduce MiSo, a domain-specific lang uage and compiler for generating efficient code for low-dimensional MINIMI ZE and SOLVE problem...\n\n\nFederico Sichetti and Enrico Puppo (Universit à di Genova), Zizhou Huang (New York University), Marco Attene (CNR IMATI) , and Denis Zorin and Daniele Panozzo (New York University)\n------------- --------\nVR-Doh: Hands-on 3D Modeling in Virtual Reality\n\nVR-Doh, an in tuitive VR-based 3D modeling system that lets you sculpt and manipulate so ft objects and edit 3D Gaussian Splatting scenes in real time. Combining p hysics-based simulation and expressive interaction, VR-Doh empowers both n ovices and experts to create rich, deformable, simulation-ready m...\n\n\n Zhaofeng Luo (Carnegie Mellon University, Peking University); Zhitong Cui (Carnegie Mellon University, Zhejiang University); Shijian Luo (Zhejiang U niversity); Mengyu Chu (Peking University, State Key Laboratory of General Artificial Intelligence); and Minchen Li (Carnegie Mellon University)\n-- -------------------\nRadiance Surfaces: Optimizing Surface Representations with a 5D Radiance Field Loss\n\nWe present a simple and fast method to r econstruct radiance surfaces by directly supervising the radiance field vi a image projection. \nUnlike volumetric approaches, we move alpha blending and ray marching from image formation into loss computation. \nThis simpl e modification enables high-quality surf...\n\n\nZiyi Zhang and Nicolas Ro ussel (EPFL); Thomas Muller, Tizian Zeltner, Merlin Nimier-David, and Fabr ice Rousselle (NVIDIA); and Wenzel Jakob (EPFL)\n---------------------\nGe nerating Past and Future in Digital Painting Processes\n\nA framework to g enerate past and future processes for drawing process videos.\n\n\nLvmin Z hang and Chuan Yan (Stanford University), Yuwei Guo and Jinbo Xing (CUHK), and Maneesh Agrawala (Stanford University)\n---------------------\nNeural ly Integrated Finite Elements for Differentiable Elasticity on Evolving Do mains\n\nWe train a network to map signed distance fields to the quadratur e points and weights of non-conforming numerical integration rule in a Mix ed Finite Element formulation, enabling differentiable elastic simulation over evolving domains. We demonstrate applications to image-guided materia l and topolog...\n\n\nGilles Daviet, Tianchang Shen, Nicholas Sharp, and D avid Levin (NVIDIA) and Gilles Daviet\n---------------------\nCobra: Effic ient Line Art COlorization with BRoAder References\n\nCobra is a novel eff icient long-context fine-grained ID preservation framework for line art co lorization, achieving high precision, efficiency, and flexible usability f or comic colorization. By effectively integrating extensive contextual ref erences, it transforms black-and-white line art into vibra...\n\n\nJunhao Zhuang (Tsinghua University); Lingen Li, Xuan Ju, and Zhaoyang Zhang (Chin ese University of Hong Kong); Chun Yuan (Tsinghua University); and Ying Sh an (Tencent)\n---------------------\nMobius: Text to Seamless Looping Vide o Generation via Latent Shift\n\nMobius is a novel method to generate seam lessly looping videos from text descriptions directly without any user ann otations, thereby creating new visual materials for the multi-media presen tation.\n\n\nXiuli Bi, Jianfei Yuan, and Bo Liu (Chongqing University of P ost and Telecommunications); Yong Zhang (Meituan); Xiaodong Cun (Great Bay University); Chi-Man Pun (University of Macau); and Bin Xiao (Chongqing U niversity of Post and Telecommunications)\n---------------------\nA Versat ile Quaternion-Based Constrained Rigid Body Dynamics\n\nWe present an impl icitly-integrated, quaternion-based constrained Rigid Body Dynamics (RBD) that guarantees satisfaction of kinematic constraints, unifying the soluti on strategy for complex mechanical systems with arbitrary kinematic struct ures, by navigating subspaces spanned by constraint forces a...\n\n\nGuire c Maloisel, Ruben Grandia, Christian Schumacher, Espen Knoop, and Moritz B ächer (Disney Research)\n---------------------\nOptimal r-Adaptive In-Time step Remeshing for Elastodynamics\n\nWe propose a coupled mesh-adaptation model and physical simulation algorithm to jointly generate, per timestep, optimal adaptive remeshings and implicit solutions for the simulation of frictionally contacting, large-deformation elastica.\n\n\nJiahao Wen (Univ ersity of Southern California, Adobe Research); Jernej Barbic (University of Southern California); and Danny Kaufman (Adobe Research)\n------------- --------\n3DGH: 3D Head Generation with Composable Hair and Face\n\nWe pre sent 3DGH, a generative model that creates realistic 3D human heads with c omposable hair and face components. By modeling both the separation and co rrelation between hair and face in a generative paradigm, it enables high- quality, full-head synthesis and flexible 3D hairstyle editing with stro.. .\n\n\nChengan He (Yale University); Junxuan Li (Meta Codec Avatars Lab); Tobias Kirschstein and Artem Sevastopolskiy (Technical University of Munic h); Shunsuke Saito, Qingyang Tan, Javier Romero, and Chen Cao (Meta Codec Avatars Lab); Holly Rushmeier (Yale University); and Giljoo Nam (Meta Code c Avatars Lab)\n---------------------\nHigh-Fidelity Novel View Synthesis via Splatting-Guided Diffusion\n\nWe present SplatDiff, a pixel-splatting- guided diffusion model for single-image novel view synthesis (NVS). Levera ging pixel splatting and video diffusion, SplatDiff generates high-quality novel views with consistent geometry and high-fidelity details. SplatDiff achieves state-of-the-art results in ...\n\n\nXiang Zhang (ETH Zürich, Di sneyResearch|Studios); Yang Zhang and Lukas Mehl (DisneyResearch|Studios); Markus Gross (ETH Zürich, DisneyResearch|Studios); and Christopher Schroe rs (DisneyResearch|Studios)\n---------------------\nVector-Valued Monte Ca rlo Integration Using Ratio Control Variates\n\nVariance reduction techniq ues are widely used to reduce the noise of Monte Carlo integration. Howeve r, these techniques are typically designed with the assumption that the in tegrand is scalar-valued. To address this, we introduce ratio control vari ations, an estimator that leverages a ratio-based ap...\n\n\nHaolin Lu (UC San Diego, MPI for Informatics); Delio Vicini (Google Inc.); and Wesley C hang and Tzu-Mao Li (UC San Diego)\n---------------------\nHigh-performanc e CPU Cloth Simulation Using Domain-decomposed Projective Dynamics\n\nThis paper presents a CPU-based cloth simulation algorithm that partitions gar ment models into domains that can be processed by each individual CPU core . Using projective dynamics with domain-level parallelization, this method achieves high performance comparable to GPU methods and runs about an or ...\n\n\nZixuan Lu, Ziheng Liu, and Lei Lan (University of Utah); Huamin W ang (Style3D Research); Yuko Ishiwaka (SoftBank); Chenfanfu Jiang (UCLA); Kui Wu (LightSpeed Studios); and Yin Yang (University of Utah)\n---------- -----------\nReal-Time Knit Deformation and Rendering\n\nIn this work, we introduce the first real-time framework that integrates yarn-level simulat ion with fiber-level rendering. The whole system provides real-time perfor mance and has been evaluated through various application scenarios, includ ing knit simulation for small patches and full garments and y...\n\n\nTao Huang (LightSpeed Studios), Haoyang Shi (University of Utah), Mengdi Wang and Yuxing Qiu (LightSpeed Studios), Yin Yang (University of Utah), and Ku i Wu (LightSpeed Studios)\n---------------------\nOffset Geometric Contact \n\nWe introduce Offset Geometric Contact (OGC), a groundbreaking method o ffering "penetration-free for free" simulations of codimensional objects. OGC efficiently constructs offset volumetric shapes to ensure stable, art ifact-free collisions. Leveraging parallel GPU computations, it delivers r eal-time...\n\n\nAnka H. Chen (University of Utah, NVIDIA); Jerry Hsu and Ziheng Liu (University of Utah); Miles Macklin (NVIDIA); and Yin Yang and Cem Yuksel (University of Utah)\n---------------------\nFlexiAct: Towards Flexible Action Control in Heterogeneous Scenarios\n\nWe propose FlexiAct, an image animation framework that transfers actions from a reference vide o to any target image, enabling variations in layout, viewpoint, and skele tal structure while maintaining identity consistency.\n\n\nShiyi Zhang (Ts inghua University); Junhao Zhuang, Zhaoyang Zhang, and Ying Shan (Tencent) ; and Yansong Tang (Tsinghua University)\n---------------------\nTeGA: Tex ture Space Gaussian Avatars for High-Resolution Dynamic Head Modeling\n\nB y combining a continuous, UVD tangent space 3DGS model with a UNet deforma tion network while maintaining adaptive densification, we present a novel high-detail 3D head avatar model that preserves even finer detail like por es and eyelashes at 4K resolution.\n\n\nGengyan S. Li (ETH Zürich, Google) and Paulo Gotardo, Timo Bolkart, Stephan Garbin, Kripasindhu Sarkar, Abhi mitra Meka, Alexandros Lattas, and Thabo Beeler (Google)\n---------------- -----\nDrag Your Gaussian: Effective Drag-Based Editing with Score Distil lation for 3D Gaussian Splatting\n\nDYG is a 3D drag-based scene editing m ethod for Gaussian Splatting that enables precise, multi-view consistent g eometric edits using 3D masks and control points. It combines implicit tri plane representation and a drag-based diffusion model for high-quality, fi ne-grained results. Visit our project pa...\n\n\nYansong Qu, Dian Chen, an d Xinyang Li (Xiamen University); Xiaofan Li (Baidu Inc.); and Shengchuan Zhang, Liujuan Cao, and Rongrong Ji (Xiamen University)\n----------------- ----\nEscher Tile Deformation via Closed-Form Solution\n\nA real-time defo rmation method for Escher tiles --- interlocking organic forms that seamle ssly tessellate the plane following symmetry rules. Rather than treating t iles as mere boundaries, we consider them as textured shapes, ensuring tha t both the boundary and interior deform simultaneously. The de...\n\n\nCra ne He Chen (Industrial Light & Magic, Northeastern University) and Vladimi r Kim (Adobe)\n---------------------\nSobol' Sequences with Guaranteed-Qua lity 2D Projections\n\nIn the context of quasi-Monte Carlo rendering, we i ntroduce a new Sobol' construction and demonstrate that particular pairs o f polynomials of the form p and p^2+p+1 in Sobol'-based sampling lead to ( 1, 2)-sequences. They can be combined to form high-dimensional low discrep ancy sequences with good 2D...\n\n\nNicolas Bonneel and David Coeurjolly ( CNRS - LIRIS) and Jean-Claude Iehl and Victor Ostromoukhov (Université Cla ude Bernard Lyon 1, CNRS - LIRIS)\n---------------------\nDiscipline Toget her with the Self in Kendo: Exploring "Qi" through Mixed Reality and Autoe thnography\n\nThis work explores “qi” in kendo through mixed reality and a utoethnography, blending tradition and technology. By animating digital hu mans with “qi”, it frames martial arts as art. The project invites reflect ion on selfhood and offers fresh insights at the intersection of cul...\n\ n\nKaren Furuta, Jingjing Li, Tatsuki Fushimi, and Yoichi Ochiai (Universi ty of Tsukuba)\n---------------------\nTowards Understanding Depth Percept ion in Foveated Rendering\n\nWe demonstrate that stereoacuity is remarkabl y resilient to foveated rendering and remains unaffected with up to 2× str onger foveation than commonly used. To this end, we design a psychovisual experiment and derive a simple perceptual model that determines the amount of foveation that does not affec...\n\n\nSophie Kergaßner, Taimoor Tariq, and Piotr Didyk (Università della Svizzera italiana)\n------------------- --\nDon’t Splat your Gaussians: Volumetric Ray-Traced Primitives for Model ing and Rendering Scattering and Emissive Media\n\nWe formalize the path-t racing of volumes composed of anisotropic kernel mixture models. Our work enables computing physically-based light transport on complex volumetric a ssets efficiently, on tiny memory budgets. We further introduce Epanechnik ov kernels as an efficient alternative in kernel-based ...\n\n\nJorge Cond or (Universita della Svizzera Italiana); Sébastien Speierer, Lukas Bode, B ožič Aljaž, and Simon Green (Meta Reality Labs); Piotr Didyk (Universita d ella Svizzera Italiana); Adrián Jarabo (Meta Reality Labs); and Jorge Cond or\n---------------------\nGenerative Neural Materials\n\nWe present the f irst generative model for neural BTFs, enabling single-shot generation fro m arbitrary text or image prompts. To achieve this, we introduce a univers al neural material basis and train a conditional diffusion model to genera te materials in this basis from flash images, natural images a...\n\n\nNit hin Raghavan (University of California San Diego), Krishna Mullia (Adobe R esearch), Alexander Trevithick (University of California San Diego), Fujun Luan and Miloš Hašan (Adobe Research), and Ravi Ramamoorthi (University o f California San Diego)\n---------------------\nEditDuet: A Multi-Agent Sy stem for Video Non-Linear Editing\n\nWe automate video nonlinear editing u sing a multi-agent system. An Editor agent uses tools to create sequences from clips and instructions, while a Critic agent provides feedback in nat ural language. Our learning-based approach enhances agent communication. E valuations with an LLM-as-a-judge metric ...\n\n\nMarcelo Sandoval-Castañe da (TTIC, Adobe Research); Bryan Russell and Josef Sivic (Adobe Research); Gregory Shakhnarovich (TTIC); and Fabian David Caba Heilbron (Adobe Resea rch)\n---------------------\nFaraday Cage Estimation of Normals for Point Clouds and Ribbon Sketches\n\nWe propose a novel method for normal estimat ion of unoriented point clouds and VR ribbon sketches that leverages a mod eling of the Faraday cage effect. Our method is uniquely robust to the pre sence of interior structures and artifacts, producing superior surfacing o utput when combined with Poisson S...\n\n\nDaniel Scrivener, Daniel Cui, a nd Ellis Coldren (Boston University); Mazdak Abulnaga (Massachusetts Insti tute of Technology (MIT), Harvard Medical School); Mikhail Bessmeltsev (Un iversite de Montreal); and Edward Chien (Boston University)\n------------- --------\nImage-space Adaptive Sampling for Fast Inverse Rendering\n\nOur goal is to accelerate inverse rendering by reducing the sampling budget wi thout sacrificing overall performance. We introduce a novel image-space ad aptive sampling framework to accelerate inverse rendering by dynamically a djusting pixel sampling probabilities based on gradient variance and contr ...\n\n\nKai Yan (University of California Irvine); Cheng Zhang (Reality L abs Research, Meta); Sébastien Speierer (Reality Labs, Meta); Guangyan Cai (University of California Irvine); Yufeng Zhu and Zhao Dong (Reality Labs , Meta); and Shuang Zhao (University of California Irvine)\n-------------- -------\nInterspatial Attention for Efficient 4D Human Video Generation\n\ nWe introduce a novel interspatial attention (ISA) for diffusion transform ers, which maintains identity and ensures motion consistency while allowin g precise control of camera and body poses. Combined with a custom video v ariation autoencoder, our model achieves state-of-the-art performance for photo...\n\n\nRuizhi Shao (Tsinghua University), Yinghao Xu (Stanford Univ ersity), Yujun Shen (Alibaba Group), Ceyuan Yang (ByteDance Inc.), Yang Zh eng and Changan Chen (Stanford University), Yebin Liu (Tsinghua University ), and Gordon Wetzstein (Stanford University)\n---------------------\nQuad tree Tall Cells for Eulerian Liquid Simulation\n\nThis paper introduces a novel grid structure that extends tall cell methods for efficient deep wat er simulation. Unlike previous methods, our approach subdivides tall cells horizontally, allowing for more aggressive adaptivity. We demonstrate tha t this novel form of adaptivity delivers superior perf...\n\n\nFumiya Nari ta (The University of Tokyo, GAME FREAK Inc.); Nimiko Ochiai (GAME FREAK I nc.); Takashi Kanai (The University of Tokyo); and Ryoichi Ando (Unaffilia ted)\n---------------------\nMatCLIP: Light- and Shape-Insensitive Assignm ent of PBR Material Models\n\nMatCLIP assigns realistic PBR materials to 3 D models using shape- and lighting-invariant descriptors derived from imag es, including LDM outputs and photos. It outperforms prior methods by over 15%, enabling consistent material predictions across varied geometry and lighting, with applications to lar...\n\n\nMichael Birsak (King Abdullah U niversity of Science and Technology (KAUST)), John Femiani (Miami Universi ty), and Biao Zhang and Peter Wonka (King Abdullah University of Science a nd Technology (KAUST))\n---------------------\nGenAnalysis: Joint Shape An alysis by Learning Man-Made Shape Generators with Deformation Regularizati ons\n\nWe present GenAnalysis, an implicit shape generation framework enab ling joint shape matching and consistent segmentation by enforcing as-affi ne-as-possible (AAAP) deformations via regularization loss in latent space . It enables shape analysis via extracting and analysing shape variations in the tang...\n\n\nYuezhi Yang and Haitao Yang (University of Texas at Au stin), George Kiyohiro Nakayama (Stanford University), Xiangru Huang (West lake University), Leonidas Guibas (Stanford University), and Qixing Huang (University of Texas at Austin)\n---------------------\nEVA: Expressive Vi rtual Avatars from Multi-view Videos\n\nIn this work, we introduce Express ive Virtual Avatars (EVA), an actor-specific, fully controllable and expre ssive human avatar framework that achieves high-fidelity, lifelike renderi ngs in real-time, while enabling independent control of facial expressions , body movements, and hand gestures.\n\n\nHendrik Junkawitsch, Guoxing Sun , and Heming Zhu (Max Planck Institute for Informatics) and Christian Theo balt and Marc Habermann (Max Planck Institute for Informatics; Saarbrücken Research Center for Visual Computing, Interaction and Artificial Intellig ence)\n---------------------\nStable Cosserat Rods\n\nCosserat rods have b ecome increasingly popular for simulating complex thin elastic rods. Howev er, traditional approaches often encounter significant challenges in robus tly and efficiently solving for valid quaternion orientations. We introduc e Stable Cosserat rods, which can achieve high accuracy wi...\n\n\nJerry H su (University of Utah), Tongtong Wang and Kui Wu (LightSpeed Studios), an d Cem Yuksel (University of Utah)\n---------------------\nSplat and Replac e: 3D Reconstruction with Repetitive Elements\n\nWe leverage repetitions i n 3D scenes to improve reconstruction in low-quality parts due to poor cov erage and occlusions. Our methods segments the repetitions, registers them together, and optimizes a shared representation with multi-view informati on flowing from all repetitions, improving the recons...\n\n\nNicolas Viol ante, Andréas Meuleman, and Alban Gauthier (INRIA, Université Côte d'Azur) ; Fredo Durand (Massachusetts Institute of Technology (MIT)); Thibault Gro ueix (Adobe Research); and George Drettakis (INRIA, Université Côte d'Azur )\n---------------------\nStreamME: Simplify 3D Gaussian Avatar within Liv e Stream\n\nThe StreamME takes live stream video as input to enable rapid 3D head avatar reconstruction. It achieves impressive speed, capturing the basic facial appearance within 10-seconds and reaching high-quality fidel ity within 5-minutes. StreamME reconstructs facial features through on-the -fly training, a...\n\n\nLuchuan Song (University of Rochester, Adobe Rese arch); Yang Zhou, Zhan Xu, Yi Zhou, and Deepali Aneja (Adobe Research); an d Chenliang Xu (University of Rochester)\n---------------------\nSwiftSket ch: A Diffusion Model for Image-to-Vector Sketch Generation\n\nSwiftSketch , a diffusion-based model with a transformer-decoder, generates high-quali ty vector sketches from images in under a second. It progressively denoise s stroke coordinates sampled from a Gaussian distribution, effectively gen eralizing across various object classes. Training uses the ControlS...\n\n \nEllie Arar, Yarden Frenkel, and Daniel Cohen-Or (Tel Aviv University); A riel Shamir (Reichman University); and Yael Vinker (Computer Science and A rtificial Intelligence Laboratory (CSAIL), Massachusetts Institute of Tech nology (MIT))\n---------------------\nLearning to Assemble with Alternativ e Plans\n\nWe introduce a reinforcement learning framework for assembling structures composed of rigid parts. A pre-trained policy generates alterna tive assembly plans, enabling rapid adaptation to unexpected disruptions. Our approach supports efficient and robust planning for multi-robot assemb ly tasks.\n\n\nZiqi Wang (HKUST, EPFL); Wenjun Liu (HKUST); Jingwen Wang a nd Gabriel Vallat (EPFL); Fan Shi (National University of Singapore); and Stefana Parascho and Maryam Kamgarpour (EPFL)\n---------------------\nInte rsection-Free Garment Retargeting\n\nWe introduce an automatic tool to ret arget artist-designed garments on a standard mannequin to possibly non-hum an avatars with unrealistic characteristics, which widely appear in games and animations. We preserve the geometrical features in the original desig n, guarantee intersection-free, and fit t...\n\n\nZizhou Huang (New York U niversity, Roblox); Chrystiano Araújo and Andrew Kunz (Roblox); Denis Zori n and Daniele Panozzo (New York University); and Victor Zordan (Roblox, Cl emson University)\n---------------------\nHand-Shadow Poser\n\nWe solve an inverse hand-shadow problem: finding poses of left and right hands that t ogether produce a shadow resembling the target 2D input, e.g., animals, le tters, and everyday objects. Our three-stage pipeline decouples the anatom ical constraints and semantic constraints, and our benchmark provid...\n\n \nHao Xu and Yinqiao Wang (The Chinese University of Hong Kong), Niloy Mit ra (University College London (UCL)), Shuaicheng Liu (University of Electr onic Science and Technology of China), Pheng Ann Heng (Chinese University of Hong Kong), and Chi-Wing Fu (The Chinese University of Hong Kong)\n---- -----------------\nAnyTop: Character Animation Diffusion with Any Topology \n\nAnyTop generates motion for diverse character skeletons using only ske letal structure as input. This diffusion model incorporates topology infor mation and textual joint descriptions to learn semantic correspondences ac ross different skeletons. It generalizes with minimal training examples an d suppor...\n\n\nInbar Gat, Sigal Raab, Guy Tevet, Yuval Reshef, Amit Haim Bermano, and Daniel Cohen-Or (Tel Aviv University)\n--------------------- \nVirtualized 3D Gaussians: Flexible Cluster-based Level-of-Detail System for Real-Time Rendering of Composed Scenes\n\nV3DG achieves real-time rend ering of massive 3D Gaussians in large, composed scenes through a novel LO D approach.\nInspired by Nanite, V3DG processes detailed 3D assets into cl usters at various granularities offline, and selectively renders 3D Gaussi ans at runtime—flexibly balancing rendering s...\n\n\nXijie Yang (Zhejiang University, Shanghai Artificial Intelligence Laboratory); Linning Xu (The Chinese University of Hong Kong); Lihan Jiang (University of Science and Technology of China, Shanghai Artificial Intelligence Laboratory); Dahua L in (The Chinese University of Hong Kong, Shanghai Artificial Intelligence Laboratory); and Bo Dai (University of Hong Kong)\n---------------------\n Appearance-Preserving Scene Aggregation for Level-of-Detail Rendering\n\nW e present a novel volumetric representation for the aggregated appearance of complex scenes and a pipeline for level-of-detail generation and render ing. Our representation preserves accurate far-field appearance and spatia l correlation from scene geometry. Our method faithfully reproduces appear anc...\n\n\nYang Zhou and Tao Huang (University of California Santa Barbar a), Ravi Ramamoorthi (University of California San Diego), Pradeep Sen and Ling-Qi Yan (University of California Santa Barbara), and Ling-Qi Yan\n-- -------------------\nStochastic Preconditioning for Neural Field Optimizat ion\n\nStochastic preconditioning adds spatial noise to query locations du ring neural field optimization; it can be formalized as a stochastic estim ate for a blur operator. This simple technique eases optimization and sign ificantly improves quality for neural fields optimization, matching or out performing ...\n\n\nSelena Ling (NVIDIA, University of Toronto); Merlin Ni mier-David (NVIDIA); Alec Jacobson (University of Toronto); and Nicholas S harp (NVIDIA)\n---------------------\nTopological Offsets\n\nTopological O ffsets is a method for generating offset surfaces that are topologically e quivalent to an offset infinitesimally close to the surface. By constructi on, the offsets are manifold, watertight, self-intersection-free, and stri ctly enclose the input. Tested on Thingi10k, it supports applicat...\n\n\n Daniel Zint, Zhouyuan Chen, Yifei Zhu, and Denis Zorin (New York Universit y/Courant); Teseo Schneider (University of Victoria); and Daniele Panozzo (New York University/Courant)\n---------------------\nLearning to Move, Le arning to Play, Learning to Animate: a Multimedia Exploration of the More- than-human Intelligence\n\n"Learning to Move, Learning to Play, Learning t o Animate" is an interdisciplinary multimedia performance, merging real-ti me AI visuals, plant biofeedback, and found object robotics to explore mor e-than-human intelligence. Challenging anthropocentrism, it envisions co-c reative agency among humans, ma...\n\n\nMingyong Cheng, Sophia Sun, Han Zh ang, and Yuemeng Gu (University of California San Diego)\n---------------- -----\nDeepMill: Neural Accessibility Learning for Subtractive Manufacturi ng\n\nThe proposed neural network, DeepMill, can efficiently predict inacc essible and occlusion regions in subtractive manufacturing. By utilizing a cutter-aware dual-head octree-based convolutional architecture, it overco mes the computational inefficiency of traditional geometric methods and is capable o...\n\n\nFanchao Zhong and Yang Wang (Shandong University), Peng -Shuai Wang (Peking University), and Lin Lu and Haisen Zhao (Shandong Univ ersity)\n---------------------\nSOAP: Style-Omniscient Animatable Portrait s\n\nSOAP awakens the 3D princess from 2D stylized photos. Unlike other wo rks that directly drive the 2D photos, SOAP reconstructs well-rigged 3D av atars, with detailed geometry and all-around texture, from just a single s tylized picture.\n\n\nTingting Liao and Yujian Zheng (Mohamed bin Zayed Un iversity of Artificial Intelligence); Yuliang Xiu (Westlake University); A dilbek Karmanov (Mohamed bin Zayed University of Artificial Intelligence); Liwen Hu (Pinscreen); Leyang Jin (Mohamed bin Zayed University of Artific ial Intelligence); and Hao Li (Mohamed bin Zayed University of Artificial Intelligence, Pinscreen)\n---------------------\nPhysicsFC: Learning User- Controlled Skills for a Physics-Based Football Player Controller\n\nPhysic sFC introduces a breakthrough in interactive football simulation—enabling real-time control of physically simulated players that perform complex ski lls with smooth transitions. It combines skill-specific learning, physics- informed rewards, latent-guided training, and transition-aware sta...\n\n\ nMinsu Kim, Eunho Jung, and Yoonsang Lee (Hanyang University)\n----------- ----------\nFaceExpressions-70k: A Dataset of Perceived Expression Differe nces\n\nWe introduce FaceExpressions-70k, a large-scale dataset comprising 70,500 crowdsourced comparisons of facial expressions collected from over 1,000 participants. It supports the training of perceptual models for exp ression differences and helps guide decisions on acceptable latency and sa mpling rates...\n\n\nAvinab Saha and Yu-Chih Chen (University of Texas Aus tin); Jean-Charles Bazin, Christian Häne, Ioannis Katsavounidis, and Alexa ndre Chapiro (Reality Labs, Meta); and Alan Bovik (University of Texas Aus tin)\n---------------------\nAssetDropper: Asset Extraction via Diffusion Models with Reward-Driven Optimization\n\nAssetDropper is a novel framewor k for extracting standardized assets from reference images, addressing cha llenges such as occlusion and distortion. Leveraging both synthetic and re al-world datasets, along with a reward-driven feedback mechanism, it achie ves state-of-the-art performance in asset extr...\n\n\nLanjiong Li (The Ho ng Kong University of Science and Technology (Guangzhou)); Guanhua Zhao (S chool of Electronic and Computer Engineering, Peking University); Lingting Zhu (The University of Hong Kong); Zeyu Cai (The Hong Kong University of Science and Technology (Guangzhou)); Lequan Yu (The University of Hong Kon g); Jian Zhang (School of Electronic and Computer Engineering, Peking Univ ersity); and Zeyu Wang (The Hong Kong University of Science and Technology (Guangzhou), The Hong Kong University of Science and Technology)\n------- --------------\nA Deep Learning-based Virtual Oculoplastic Surgery Simulat or\n\nA novel deep learning system enhances realistic virtual oculoplastic surgery simulations.\n\n\nSeonghyeon Kim (KAIST, Visual Media Lab; Anigma Technologies); Chang Wook Seo (Anigma Technologies); Kwanggyoon Seo (KAIS T, Visual Media Lab); Seung Han Song (Chungnam National University Hospita l, Chungnam National University); and Junyong Noh (KAIST, Visual Media Lab )\n---------------------\nInverse Geometric Locomotion\n\nWe present a com putational framework for optimizing shape sequences to achieve user-define d motion objectives in deformable bodies undergoing geometric locomotion. Through a reduced spatiotemporal parameterization of the shape sequences, our method is able to efficiently capture the complex coupling...\n\n\nQue ntin Becker (EPFL); Oliver Gross (University of California San Diego, EPFL ); and Mark Pauly (EPFL)\n---------------------\nMulti-Dimensional Procedu ral Wave Noise\n\nWe introduce a fast, wave-based procedural noise model e nabling precise spectral control in any dimension. Using precomputed wave functions and inverse Fourier transforms, it supports Gaussian and non-Gau ssian noises—including Gabor, Phasor, and novel recursive cellular pattern s—making i...\n\n\nPascal Guehl (LIX - Ecole Polytechnique/CNRS, Institut Polytechnique de Paris); Rémi Allègre (ICube, Université de Strasbourg; CN RS); Guillaume Gilet (Université de Sherbrooke); Basile Sauvage (ICube, Un iversité de Strasbourg; CNRS); Marie-Paule Cani (LIX - Ecole Polytechnique /CNRS, Institut Polytechnique de Paris); and Jean-Michel Dischler (ICube, Université de Strasbourg; CNRS)\n---------------------\nxADA: Controllable and Expressive Audio-Driven Animation\n\nWe introduce xADA, a generative model for creating expressive and realistic animation of the face, tongue, and head directly from speech audio. \nThe animation maps directly onto M etaHuman compatible rig controls enabling integration into industry-standa rd content creation pipelines. \nxADA generalize...\n\n\nSarah Taylor, Sal vador Medina, Jonathan Windle, Erica Alcusa Sáez, and Iain Matthews (Epic Games)\n---------------------\nTiny is not small enough: High quality, low -resource facial animation models through hybrid knowledge distillation\n\ nThe goal of this work is to train lip sync animation models that can run in real-time and on-device. We design a two-stage knowledge distillation f ramework to distill large, high-quality models. Our results show that we c an train small models with low latency and a comparatively small loss in q ualit...\n\n\nZhen Han, Mattias Teye, Derek Yadgaroff, and Judith Bütepage (Electronic Arts)\n---------------------\nPocket Time-Lapse\n\nPocket Tim e-Lapse is a system to record, explore and visualize long-term changes in the environment, based on data that a user can capture with the phone they carry. Our contributions include a process to conveniently capture a scen e, and novel techniques for registering and visualizing panoramic ti...\n\ n\nEric Chen (Cornell University; Computer Science and Artificial Intellig ence Laboratory (CSAIL), Massachusetts Institute of Technology (MIT)) and Žiga Kovačič, Madhav Aggarwal, and Abe Davis (Cornell University)\n------- --------------\nControllable Complex Freezing Dynamics Simulation on Thin Films\n\nWe present a physics-based method for simulating intricate freezi ng dynamics on thin films. Our novel Phase Map method integrated with MELP particles reproduces Marangoni freezing dynamics and the "Snow-Globe Effe ct". The framework captures soap bubble freezing dynamics while ensuring s tability in c...\n\n\nYijie Liu (TMCC, College of Computer Science, Nankai University); Taiyuan Zhang (Dartmouth College; TMCC, College of Computer Science, Nankai University); and Xiaoxiao Yan, Nuoming Liu, and Bo Ren (TM CC, College of Computer Science, Nankai University)\n--------------------- \nCueTip: An Interactive and Explainable Physics-aware Pool Assistant\n\nC ueTip is an interactive and explainable automated coaching assistant for a variant of pool/billiards. CueTip has a natural-language interface, the a bility to perform contextual, physics-aware reasoning, and its explanation s are rooted in a set of predetermined guidelines developed by domain expe rts...\n\n\nSean Memery and Kevin Denamganaï (University of Edinburgh), Ji axin Zhang and Zehai Tu (Lightspeed Studios), Yiwen Guo (Independent), and Kartic Subr (University of Edinburgh)\n---------------------\nHyper-Dimen sional Deformation Simulation\n\nEver feel like three dimensions isn't qui te enough? We performed the analysis necessary to simulate the motion of d eformables in four spatial dimensions! Along the way, we developed techniq ues for generating simulation-ready hyper-meshes, analyzing hyper-dimensio nal deformation energies, and detecti...\n\n\nAlvin Shi, Haomiao Wu, and T heodore Kim (Yale University)\n---------------------\nStable-Makeup: When Real-World Makeup Transfer Meets Diffusion Model\n\nStable-Makeup is a dif fusion-based makeup transfer method. It leverages a Detail-Preserving make up encoder, and content-structure control modules to preserve facial conte nt and structure during transfer. Extensive experiments show that Stable-M akeup outperforms existing methods, offering robust, gen...\n\n\nYuxuan Zh ang (Shanghai Jiao Tong University), Yirui Yuan (Shanghai Tech University) , Yiren Song (National University of Singapore), and Jiaming Liu (Tiamat A I)\n---------------------\nFloating Strokes: A Spatial Interpretation and Modeling Method of Chinese Calligraphy\n\nTransforming 2D Chinese calligra phy into 3D forms deepens how traditional art is understood and experience d, combining cultural heritage with modern technology. This approach adds spatial depth, opening new possibilities for digital art, preservation, an d interactive design.\n\nBy merging computationa...\n\n\nTroy TianYu LIN ( Hong Kong University of Science and Technology, Guangzhou); Boyan Zheng (I ndependent Researcher); and Haichuan Lin, Wen You, Kang Zhang, and Chen Li ang (Hong Kong University of Science and Technology, Guangzhou)\n--------- ------------\nDiscrete Torsion of Connection Forms on Simplicial Meshes\n\ nAlthough discrete connections are ubiquitous in vector field design, thei r torsion remains unstudied. We extend the existing toolbox to control the torsion of discrete connections: we introduce a new discrete Levi-Civita connection and define torsion as a measure of deviation from this referenc e, so...\n\n\nTheo Braune (Centre National de la Recherche Scientifique - Laboratoire d'informatique de l'École Polytechnique (LIX), Inria Saclay); Mark Gillespie (Inria Saclay); Yiying Tong (Michigan State University); an d Mathieu Desbrun (Inria Saclay)\n---------------------\nCurl Quantization for Automatic Placement of Knit Singularities\n\nWe present a method for the automatic placement of knit singularities based on curl quantization. Our method generates knit graphs that maintain all structural manufacturin g constraints as well as any additional user constraints. This approach al lows for simulation-free previews of rendered knits an...\n\n\nRahul Mitra (Boston University, LightSpeed Studios); Mattéo Couplet (Boston Universit y); Tongtong Wang (LightSpeed Studios); Megan Hoffman (Northeastern Univer sity); Kui Wu (LightSpeed Studios); and Edward Chien (Boston University)\n ---------------------\nPatch-Grid: An Efficient and Feature-Preserving Neu ral Implicit Surface Representation\n\nWe introduce Patch-Grid, a unified neural implicit representation that efficiently represents complex shapes, preserves sharp features, and handles open boundaries and thin geometric details. By decomposing shapes into patches encapsulated by adaptive featu re grids and merging them through localized...\n\n\nGuying Lin (Carnegie M ellon University); Lei Yang (The University of Hong Kong); Congyi Zhang (T he University of British Columbia); Hao Pan (Tsinghua University); Yuhan P ing, Guodong Wei, and Taku Komura (The University of Hong Kong); John Keys er and Wenping Wang (Texas A&M University); and Guying Lin\n-------------- -------\nCirrus: Adaptive Hybrid Particle-Grid Flow Maps on GPU\n\nWe intr oduce an adaptive octree-based GPU simulator for large-scale fluid simulat ion. Our hybrid particle-grid flow map advection scheme effectively preser ves vortex details, enabling high-resolution and high-quality results. The source code has been made publicly available at: https://wang-mengdi.g... \n\n\nMengdi Wang (Georgia Institute of Technology), Fan Feng (Dartmouth C ollege), and Junlin Li and Bo Zhu (Georgia Institute of Technology)\n----- ----------------\nColor Matching and Biomimicry for Multi-material Dental 3D Printing\n\nWe propose a practical method for dental layer biomimicry a nd multi-spot shade matching using multi-material 3D printing. It integrat es seamlessly into workflows combining dental CAD tools and industrial mul ti-material slicers. \nWe validated it by printing multiple dentures and t eeth with varying in...\n\n\nAndrás Simon (Fraunhofer IGD, Technical Unive rsity of Darmstadt); Danwu Chen (Fraunhofer IGD); Philipp Urban (Fraunhofe r IGD, Norwegian University of Science and Technology NTNU); and Vincent D uveiller and Henning Lübbe (VITA Zahnfabrik H. Rauter GmbH & Co. KG)\n---- -----------------\nBANG: Dividing 3D Assets via Generative Exploded Dynami cs\n\nBANG introduces Generative Exploded Dynamics, a novel method that dy namically decomposes 3D objects into meaningful, volumetric parts through smooth, controllable exploded views. Bridging intuitive human understandin g and generative AI, it enables precise part-level manipulation, semantic comprehens...\n\n\nLongwen Zhang, Qixuan Zhang, and Haoran Jiang (Shanghai Tech University, Deemos Technology); Yinuo Bai (ShanghaiTech University); Wei Yang (Huazhong University of Science and Technology); and Lan Xu and J ingyi Yu (ShanghaiTech University)\n---------------------\nGaussian Wave S platting for Computer-Generated Holography\n\nWe develop novel and efficie nt computer-generated holography algorithms, dubbed Gaussian Wave Splattin g, that transform Gaussian-based scene representations into holograms. We derive a closed-form 2D Gaussian-to-hologram transform supporting occlusio ns and alpha blending, along with an efficient, ea...\n\n\nSuyeon Choi, Br ian Chao, Jacqueline Yang, Manu Gopakumar, and Gordon Wetzstein (Stanford University)\n---------------------\nANIME-Rod: Adjustable Nonlinear Isotro pic Materials for Elastic Rods\n\nWe derive a nonlinear elastic rod energy , starting from a general 3D volumetric isotropic material. Validated agai nst FEM, we accurately capture rod stretching, bending and twisting, under finite deformations. We also propose how to separately control linear/non linear stretchability/bendability/twis...\n\n\nHuanyu Chen, Jiahao Wen, an d Jernej Barbič (University of Southern California)\n--------------------- \nFast Subspace Fluid Simulation with a Temporally-Aware Basis\n\nWe intro duce a novel reduced-order fluid simulation technique leveraging Dynamic M ode Decomposition (DMD) to enable fast, memory-efficient, and user-control lable subspace simulation. Combining spatial ROM compression with spectral physical insights, our method excels in animation, real-time interact...\ n\n\nSiyuan Chen (University of Toronto, Shanghai Jiao Tong University) an d Yixin Chen, Jonathan Panuelos, Otman Benchekroun, Yue Chang, Eitan Grins pun, and Zhecheng Wang (University of Toronto)\n---------------------\nSki llMimic-V2: Learning Robust and Generalizable Interaction Skills from Spar se and Noisy Demonstrations\n\nThis work addresses the challenge of learni ng robust interaction skills from limited demonstrations. By introducing n ovel data augmentation techniques for skill transitions and recovery patte rns, combined with enhanced reinforcement imitation learning methods, we a chieve superior performance in lear...\n\n\nRunyi Yu (Hong Kong University of Science and Technology, Shanghai Aritificial Intelligence Laboratory); Yinhuai Wang, Qihan Zhao, and Hok Wai Tsui (Hong Kong University of Scien ce and Technology); Jingbo Wang (Shanghai Aritificial Intelligence Laborat ory); and Ping Tan and Qifeng Chen (Hong Kong University of Science and Te chnology)\n---------------------\nTexture Size Reduction Through Symmetric Overlap and Texture Carving\n\nWe develop a method to compress textures a nd UVs for meshes in a content-aware way. We combine this with overlapping and folding symmetric UV charts, and demonstrate our approach on a datase t from Sketchfab. We outperform prior work in visual similarity to the ori ginal mesh.\n\n\nJulian Knodt and Xifeng Gao (LightSpeed Studios) and Juli an Knodt\n---------------------\nNested Attention: Semantic-aware Attentio n Values for Concept Personalization\n\nThis paper introduces Nested Atten tion, a mechanism that improves text-to-image personalization by injecting query-dependent subject features into cross-attention layers, achieving s trong identity preservation and prompt alignment. The method maintains the model’s prior, enabling multi-subject...\n\n\nOr Patashnik (Tel Aviv Univ ersity, Snap); Rinon Gal (Tel Aviv University); Daniil Ostashev, Sergey Tu lyakov, and Kfir Aberman (Snap); and Daniel Cohen-Or (Tel Aviv University, Snap)\n---------------------\nRelightable Full-Body Gaussian Codec Avatar s\n\nWe present the first drivable full-body avatar model that reconstruct s perceptually realistic relightable appearance.\n\n\nShaofei Wang (ETH Zü rich); Tomas Simon, Igor Santesteban, Timur Bagautdinov, Junxuan Li, Vasu Agrawal, Fabian Prada, Shoou-I Yu, Pace Nalbone, Matt Gramlich, Roman Luba chersky, Chenglei Wu, Javier Romero, Jason Saragih, and Michael Zollhoefer (Reality Labs Research, Meta); Andreas Geiger (University of Tübingen, Tü bingen AI Center); Siyu Tang (ETH Zürich); and Shunsuke Saito (Reality Lab s Research, Meta)\n---------------------\nQUASAR: Quad-based Adaptive Stre aming And Rendering\n\nThis paper introduces an improved quad-based geomet ry streaming method for remote rendering that reduces bandwidth demands th rough temporal compression and supports QoE-driven adaptation. It achieves high-quality visuals, captures disocclusion events, uses 15× less data th an SOTA, and reduces bandwi...\n\n\nEdward Lu and Anthony Rowe (Carnegie M ellon University)\n---------------------\nSketch2Anim: Towards Transferrin g Sketch Storyboards into 3D Animation\n\nThis paper presents a novel and first approach - Sketch2Anim, to automatically translate 2D storyboard ske tches into high-quality 3D animations through multi-conditional motion gen eration.\n\n\nLei Zhong (University of Edinburgh), Chuan Guo (Snap Inc.), Yiming Xie (Northeastern University), and Jiawei Wang and Changjian Li (Un iversity of Edinburgh)\n---------------------\nNeural Co-Optimization of S tructural Topology, Manufacturable Layers, and Path Orientations for Fiber -Reinforced Composites\n\nWe present a computational framework that co-opt imizes structural topology, curved layers, and fiber orientations for manu facturable, high-strength composites. Using implicit neural fields, our me thod integrates design and fabrication objectives into a unified optimizat ion process, achieving up to 3...\n\n\nTao Liu, Tianyu Zhang, Yongxue Chen , Weiming Wang, Yu Jiang, Yuming Huang, and Charlie C.L. Wang (University of Manchester)\n---------------------\nQuadric-Based Silhouette Sampling f or Differentiable Rendering\n\nPhysically based differentiable rendering c omputes gradients of the rendering equation. The task is made difficult by discontinuities in the integrand at object silhouettes. To address this c hallenge, we propose a novel edge sampling approach that outperforms the s tate-of-the-art among unidirectiona...\n\n\nMariia Soroka (Cornell Univers ity, Intel); Christoph Peters (Delft University of Technology, Intel); and Steve Marschner (Cornell University)\n---------------------\nA Neural Par ticle Level Set Method for Dynamic Interface Tracking\n\nWe propose Neural PLS, a neural particle level-set method for tracking and evolving dynamic neural representations. Oriented particles serve as interface trackers an d sampling seeders, enabling efficient evolution on a multi-resolution gri d-hash structure. Our approach integrates traditional PLS and...\n\n\nDuow en Chen (Georgia Institute of Technology), Junwei Zhou (University of Mich igan), Bo Zhu (Georgia Institute of Technology), and Duowen Chen\n-------- -------------\nLight Pipe Holographic Display: Bandwidth-preserved Kaleido scopic Guiding for AR Glasses\n\nWe present and analyze a holographic augm ented reality display with the bandwidth-preserved guiding method using a light pipe. We propose the use of light pipe to spatially relocate the lig ht engine from the image combiner at the front-module, enabling enhanced w eight distribution and obstruction-fr...\n\n\nMinseok Chae, Chun Chen, Seu ng-Woo Nam, and Yoonchan Jeong (Seoul National University)\n-------------- -------\nThe Posthumous World: Our selves, Our Bodies and a World that Wi ll Continue Without Us\n\nHow would our attitudes to death and dying chang e if we could see how a human body is reabsorbed into the environment? The Posthumous World is a project about death and our relationship with the p lanet. At its centre will be a new artwork - a poetic meditation on a body ’s journey to re-join th...\n\n\nRichard Wright (Royal Holloway University of London)\n---------------------\nMeschers: Geometry Processing of Impos sible Objects\n\nMeschers are a mesh representation for Escheresque geomet ry. They allow us to solve partial differential equations on the surface o f an impossible object, meaning that we can find impossible shortest paths , perform mescher smoothing, and even inverse render a mescher from an ima ge.\n\n\nAna Dodik, Isabella Yu, Kartik Chandra, and Jonathan Ragan-Kelley (Computer Science and Artificial Intelligence Laboratory (CSAIL), Massach usetts Institute of Technology (MIT)); Joshua Tenenbaum (Massachusetts Ins titute of Technology (MIT)); and Vincent Sitzmann and Justin Solomon (Comp uter Science and Artificial Intelligence Laboratory (CSAIL), Massachusetts Institute of Technology (MIT))\n---------------------\nAutomated Task Sch eduling for Cloth and Deformable Body Simulations in Heterogeneous Computi ng Environments\n\nThis paper introduces an automated scheduling framework to optimize cloth and deformable simulations across heterogeneous computi ng devices. Using an enhanced HEFT algorithm and asynchronous iteration me thods, our approach minimizes communication delays and maximizes paralleli sm. our experiments dem...\n\n\nChengzhu He (Xiamen University, Style3D Re search); Zhendong Wang (Style3D Research); Zhaorui Meng, Junfeng Yao, and Shihui Guo (Xiamen University); and Huamin Wang (Style3D Research)\n------ ---------------\nAlgorithmic Miner: Humanity in Service - An AI-Driven VR Journey into Machine Logic\n\nAlgorithmic Miner uses VR to reveal the hidd en labor behind AI systems. By immersing participants in data annotation t asks, it critically reflects on exploitation, automation, and techno-capit alism, prompting new discussions on ethical, human-centered design in inte ractive systems.\n\n\nJia SUN (The Hong Kong University of Science and Tec hnology (Guangzhou)); Zheng WEI (The Hong Kong University of Science and T echnology); and Pan HUI (The Hong Kong University of Science and Technolog y (Guangzhou), The Hong Kong University of Science and Technology)\n------ ---------------\nSketch3DVE: Sketch-based 3D-Aware Scene Video Editing\n\n We propose Sketch3DVE, a sketch-based 3D-aware video editing method to ena ble detailed local manipulation of videos with significant viewpoint chang es. Our approach leverages detailed analysis and editing of underlying 3D scene representations, combined with a diffusion model to synthesize reali stic...\n\n\nFeng-Lin Liu (Institute of Computing Technology, Chinese Acad emy of Sciences; University of Chinese Academy of Sciences); Shi-Yang Li ( Institute of Computing Technology, Chinese Academy of Sciences); Yan-Pei C ao (VAST); Hongbo Fu (Hong Kong University of Science and Technology); and Lin Gao (Institute of Computing Technology, Chinese Academy of Sciences; University of Chinese Academy of Sciences)\n---------------------\nSemanti cally Consistent Text-to-Motion with Unsupervised Styles\n\nWe introduce a novel method that integrates unsupervised style from arbitrary references into a text-driven diffusion model to generate semantically consistent st ylized human motion. We leverage text as a mediator to capture the tempora l correspondences between motion and style, enabling the seamles...\n\n\nL injun Wu and Xiangjun Tang (Zhejiang University; State Key Laboratory of C AD&CG, Zhejiang University); Jingyuan Cong (University of California San D iego); He Wang (UCL Centre for Artificial Intelligence, Department of Comp uter Science, University College London (UCL)); Bo Hu, Xu Gong, Songnan Li , and Yuchen Liao (Tencent Technology (Shenzhen) Co., Ltd.); Yiqian Wu (Zh ejiang University); Chen Liu (State Key Lab of CAD and CG, Zhejiang Univer sity); and Xiaogang Jin (Zhejiang University; State Key Laboratory of CAD& CG, Zhejiang University)\n---------------------\nUltraMeshRenderer: Effici ent Structure and Management of GPU Out-of-core Memory for Real-time Rende ring of Gigantic 3D Meshes\n\nThis paper presents UltraMeshRenderer, a GPU out-of-core method for real-time rendering of 3D scenes with billions of vertices and triangles. It features a balanced hierarchical mesh, coherenc e-based LOD selection, and parallel in-place GPU memory management, achiev ing efficient data transfer and me...\n\n\nHuadong Zhang (Rochester Instit ute of Technology); Lizhou Cao (Rochester Institute of Technology, Univers ity of Maryland Eastern Shore); and Chao Peng (Rochester Institute of Tech nology)\n---------------------\nArenite: A Physics-based Sandstone Simulat or\n\nArenite is a novel, physics-based simulation method for generating r ealistic sandstone structures. It combines fabric interlocking, multi-fact or erosion, and particle-based deposition. Our GPU-based implementation pr oduces detailed 3D shapes such as arches, alcoves, hoodoos, and buttes in minutes an...\n\n\nZhanyu Yang (Purdue University); Aryamaan Jain and Guil laume Cordonnier (Inria, Université Côte d'Azur); Marie-Paule Cani (Centre National de la Recherche Scientifique - Laboratoire d'informatique de l'É cole Polytechnique (LIX), Institut Polytechnique de Paris); and Zhaopeng W ang and Bedrich Benes (Purdue University)\n---------------------\nA Polyhe dral Construction of Empty Spheres in Discrete Distance Fields\n\nSpheres that are disjoint from a given union of spheres can be computing by solvin g a convex hull problem. This can be exploited for contouring discretely s ampled signed distance functions.\n\n\nMax Kohlbrenner and Marc Alexa (Tec hnical University of Berlin)\n---------------------\nDigital Crazing: Appl ying Quadtree Structures to Simulate Time and Human Intervention\n\nThis w ork offers an innovative approach to digitally replicating crazing pattern s, which are aesthetic crazing found on ceramics. By using a quadtree stru cture, the method captures the time dependent and user-interaction aspects of these patterns, providing a novel perspective in digital material de.. .\n\n\nSzu-Han Lu and Hsiao-Ching Chou (National Yang Ming Chiao Tung Univ ersity, Institute of Applied Arts)\n---------------------\nNeural BRDF Imp ortance Sampling by Reparameterization\n\nWe introduce a reparameterizatio n-based formulation of neural BRDF importance sampling. Comparing to previ ous methods that construct a probability transform to the BRDF through mul ti-step invertible neural networks, our BRDF sampling is in single step wi thout needing network invertibility, achieving...\n\n\nLiwen Wu (Universit y of California San Diego); Sai Bi (Adobe Research); Zexiang Xu (Hillbot); Hao Tan, Kai Zhang, and Fujun Luan (Adobe Research); Haolin Lu (Max Planc k Institute for Informatics); and Ravi Ramamoorthi (University of Californ ia San Diego)\n---------------------\nTransformer IMU Calibrator: Dynamic On-body IMU Calibration for Inertial Motion Capture\n\nThis work proposes a dynamic calibration system for inertial motion capture, which can dynami cally remove non-static IMU drift and sensor-body offset during usage, ena ble user-friendly calibration (without T-pose and IMU heading reset), and ensure long-term robustness.\n\n\nChengxu Zuo, Jiawei Huang, Xiao Jiang, a nd Yuan Yao (Xiamen University); Xiangren Shi (Bournemouth University); Ru i Cao (Xiamen University); Xinyu Yi and Feng Xu (Tsinghua University); Shi hui Guo (Xiamen University); and Yipeng Qin (Cardiff University)\n-------- -------------\nCompensating Spatiotemporally Inconsistent Observations for Online Dynamic 3D Gaussian Splatting\n\nWe reveal that existing online re construction of dynamic scenes with 3D Gaussian Splatting produces tempora lly inconsistent results, led by inevitable noise in real-world recordings . To address this, we decompose the rendered images into the ideal signal and the errors during optimization, achieving...\n\n\nYoungsik Yun, Jeongm in Bae, and Hyunseung Son (Yonsei University); Seoha Kim, Hahyun Lee, and Gun Bang (Electronics and Telecommunications Research Institute); and Youn gjung Uh (Yonsei University)\n---------------------\nStroke Transfer for P articipating Media\n\nWe present a stroke-based method for transforming dy namic 3D scenes with smoke, fire, or clouds into painterly animations. Lea rning from user-provided exemplars, our system transfers stroke styles—col or, width, length, and orientation—while preserving motion and structure. This enables e...\n\n\nNaoto Shirashima (AGU); Hideki Todo (Takushoku Univ ersity); Yuki Yamaoka (AGU); Shizuo Kaji (Kyushu University, Kyoto Univers ity); and Kunihiko Kobayashi, Haruna Shimotahira, and Yonghao Yue (AGU)\n- --------------------\nSegment-based Light Transport Simulation\n\nWe intro duce a novel segment-based framework for light transport simulation, effic iently assembling paths from disconnected segments. Our method includes in novative segment sampling techniques and corresponding estimation strategi es. To demonstrate its strengths, we propose a robust bidirectional pa...\ n\n\nWenyou Wang (University of Waterloo), Rex West (Aoyama Gakuin Univers ity), and Toshiya Hachisuka (University of Waterloo)\n-------------------- -\nInteractive Optimization of Scaffolded Procedural Patterns\n\nWe introd uce a method for interactive design of procedural patterns, allowing users to sketch content incrementally in a level-by-level fashion. Each level, or scaffold, builds on the previous one, making optimization more responsi ve and controllable. A comprehensive validation demonstrates improved...\n \n\nDavide Sforza (Sapienza University of Rome); Marzia Riso (Sapienza Uni versity of Rome; INRIA, Université Côte d'Azur); and Filippo Muzzini, Nico la Capodieci, and Fabio Pellacini (University of Modena and Reggio Emilia) \n---------------------\nCLR-Wire: Towards Continuous Latent Representatio ns for 3D Curve Wireframe Generation\n\nCLR-Wire is a unified generative f ramework for 3D curve-based wireframes, jointly modeling geometry and topo logy in a continuous latent space. Using attention-driven VAEs and flow ma tching, it enables high-quality, diverse generation from noise, images, or point clouds—advancing CAD design, sh...\n\n\nXueqi Ma, Yilin Liu, Tianlo ng Gao, Qirui Huang, and Hui Huang (Shenzhen University)\n---------------- -----\nA Monte Carlo Rendering Framework for Simulating Optical Heterodyne Detection\n\nWe present a general spectral-domain simulation framework fo r optical heterodyne detection (OHD), extending path integral rendering to capture power spectral density of OHD. Unlike existing domain-specific to ols, our approach supports diverse scenes and applications. We validate it against real-worl...\n\n\nJuhyeon Kim (Dartmouth College); Craig Benko, M agnus Wrenninge, Ryusuke Villemin, and Zeb Barber (Aurora Innovation); and Wojciech Jarosz and Adithya Pediredla (Dartmouth College)\n-------------- -------\nBecoming Space: Exploring Agential Materiality Through AI-Generat ed Metamorphosis in Artistic Practice\n\n"Becoming Space" is an installati on that explores the agency of AI, discourse, and material intersections t hrough AI-generated forms and 3D printing. Inspired by Ovid's Metamorphose s, it explores human-animal transformations using CLIP-guided diffusion mo dels and stereolithography. The installation ...\n\n\nXinyu Ma and Hengyu Meng (The Hong Kong University of Science and Technology (Guangzhou)); Ziw ei Wu (The Hong Kong University of Science and Technology); and Zeyu Wang and Clea von Chamier-Waite (The Hong Kong University of Science and Techno logy (Guangzhou), The Hong Kong University of Science and Technology)\n--- ------------------\nStreaming-Aware Neural Monte Carlo Rendering Framework with Unified Denoising-Compression and Client Collaboration\n\nTo reduce the high rendering costs and transmission bandwidth requirements of path t racing-based cloud rendering, we propose a novel streaming-aware rendering framework that is able to learn a joint optimal model integrating two pat h-tracing acceleration (adaptive sampling and denoising) and video c...\n\ n\nHangming Fan, Yuchi Huo, Chuankun Zheng, and Chonghao Hu (State Key Lab oratory of CAD & CG, Zhejiang University); Yazhen Yuan (Game Engine Depart ment, CROS, Tencent, China); and Rui Wang (State Key Laboratory of CAD & C G, Zhejiang University)\n---------------------\nA Hybrid Near-wall Model f or Kinetic Simulation of Turbulent Boundary Layer Flows\n\nWe present an i nnovative hybrid near-wall model for the multi-resolution lattice Boltzman n solver to effectively enable simulations of high Reynolds number turbule nt boundary layer flows. For the first time, it strikes an excellent balan ce between the precision demanded by industrial computational d...\n\n\nMe ngyun Liu, Kai Bai, and Xiaopei Liu (ShanghaiTech University)\n----------- ----------\nCreating Fluid-Interactive Virtual Agents by an Efficient Simu lator with Local-domain Control\n\nWe introduce a novel local-domain fluid -solid interaction simulator grounded in a lattice Boltzmann solver. By le veraging an MPC-based domain-tracking approach and an improved convective boundary condition, it offers enhanced stability and efficiency for derivi ng control policies of virtual agents, ...\n\n\nWenbin Song, Heng Zhang, Y ang Wang, and Xiaopei Liu (ShanghaiTech University)\n--------------------- \nProgressive Dynamics++: A Framework for Stable, Continuous, and Consiste nt Animation Across Resolution and Time\n\nWe propose a general framework, Progressive Dynamics++, for constructing a family of progressive dynamics integration methods that advance physical simulation states forward in bo th time and spatial resolution. We analyze requirements for stable, contin uous, and consistent level-of-detail animation ...\n\n\nJiayi Eris Zhang ( Stanford University, Adobe); Doug James (Stanford University); and Danny K aufman (Adobe)\n---------------------\nDual-Band Feature Fusion for Neural Global Illumination with Multi-Frequency Reflections\n\nWe present a neur al global illumination method capable of capturing multi-frequency reflect ions in dynamic scenes by leveraging object-centric feature grids and a no vel dual-band fusion module. Our approach produces high-quality, realistic rendering effects and outperforms state-of-the-art technique...\n\n\nShao hua Mo, Chuankun Zheng, Zihao Lin, and Dianbing Xi (State Key Lab of CAD&C G, Zhejiang University); Qi Ye (Zhejiang University); Rui Wang and Hujun B ao (State Key Lab of CAD&CG, Zhejiang University); and Yuchi Huo (State Ke y Lab of CAD&CG, Zhejiang University; Zhejiang Lab)\n--------------------- \nAdvancing GPU IPC for Stiff Affine-Deformable Simulation\n\nWe present a GPU-optimized IPC framework achieving up to 10× speedup across soft, stif f, and hybrid simulations. Key innovations include a connectivity-enhanced MAS preconditioner, a parallel-friendly inexact strain limiting energy, a nd a hash-based two-level reduction strategy for fast Hes-\nsian as...\n\n \nKemeng Huang (Carnegie Mellon University, The University of Hong Kong); Xinyu Lu (TransGP); Huancheng Lin (Carnegie Mellon University, The Univers ity of Hong Kong); Taku Komura (The University of Hong Kong); Minchen Li ( Carnegie Mellon University); and Kemeng Huang\n---------------------\nOrde r Matters: Learning Element Ordering for Graphic Design Generation\n\nWe p ropose a Generative Order Learner (GOL) that optimizes element ordering fo r graphic design generation. Our approach learns a content-aware neural or der, which can significantly improve graphic generation quality, generaliz e across different types of generative models and help design generators s ...\n\n\nBo Yang and Ying Cao (ShanghaiTech University)\n----------------- ----\nInstance Segmentation of Scene Sketches Using Natural Image Priors\n \nINKi enables instance segmentation for scene sketches by adapting image segmentation models with class-agnostic tuning and depth-based refinement. We introduce a new dataset INK-scene with diverse styles and demonstrate layered sketch organization for advanced editing, including inpainting occ luded ...\n\n\nMia Tang (Stanford University, Carnegie Mellon University); Yael Vinker (Computer Science and Artificial Intelligence Laboratory (CSA IL), Massachusetts Institute of Technology (MIT)); Chuan Yan and Lvmin Zha ng (Stanford University); and Maneesh Agrawala (Stanford University, Roblo x Research)\n---------------------\nComputational Modeling of Gothic Micro architecture\n\nGothic microarchitecture—a prevalent feature of late medie val art—comprises sculptural works that replicate monumental Gothic forms, though its original construction techniques remain historically undocumen ted. Leveraging insights from 15th-century Basel goldsmith drawings, we pr esent an...\n\n\nAviv Segall and Jing Ren (ETH Zurich), Martin Schwarz (Un iversity of Basel), and Olga Sorkine-Hornung (ETH Zurich)\n--------------- ------\nVideoPainter: Any-length Video Inpainting and Editing with Plug-an d-Play Context Control\n\nVideoPainter introduces a dual-branch framework for video inpainting with a lightweight context encoder that integrates wi th pre-trained diffusion transformers. Its ID resampling strategy maintain s identity consistency across any-length videos, while VPData and VPBench provide the largest segmentati...\n\n\nYuxuan Bian (The Chinese University of Hong Kong, Tencent); Zhaoyang Zhang (Tencent); Xuan Ju (The Chinese Un iversity of Hong Kong); Mingdeng Cao (The University of Tokyo); Liangbin X ie (University of Macau); Ying Shan (Tencent); and Qiang Xu (The Chinese U niversity of Hong Kong)\n---------------------\nPainting with 3D Gaussian Splat Brushes\n\nWe present the first interactive system for painting with 3D Gaussian splat brushes. With our tool, artists can sample volumetric f ragments from real-world Gaussian splat captures and paint with them in re al time. Our tool seamlessly deforms sampled splats along painted strokes, introducing realisti...\n\n\nKarran Pandey (University of Toronto); Anita Hu, Clement Fuji Tsang, and Or Perel (NVIDIA); Karan Singh (University of Toronto); and Maria Shugrina (NVIDIA)\n---------------------\nImage-GS: C ontent-Adaptive Image Representation via 2D Gaussians\n\nWe introduce Imag e-GS, a content-adaptive image representation based on colored 2D Gaussian s. Image-GS achieves remarkable rate-distortion performance across diverse images and textures while supporting hardware-friendly fast random access and flexible quality control through a smooth level-of-detai...\n\n\nYunx iang Zhang and Bingxuan Li (New York University), Alexandr Kuznetsov (Adva nced Micro Devices (AMD)), Akshay Jindal and Stavros Diolatzis (Intel Corp oration), Kenneth Chen (New York University), Anton Sochenov and Anton Kap lanyan (Intel Corporation), and Qi Sun (New York University)\n------------ ---------\nScaffoldAvatar: High-Fidelity Gaussian Avatars with Patch Expre ssions\n\nScaffoldAvatar presents a novel approach for reconstructing ultr a-high fidelity animatable head avatars, which can be rendered in real-tim e. Our method operates on patch-based local expression features and synthe sizes 3D Gaussians dynamically by leveraging tiny scaffold MLPs. We employ color-based d...\n\n\nShivangi Aneja (Technical University of Munich, Dis neyResearch|Studios); Sebastian Weiss, Irene Baeza, Prashanth Chandran, an d Gaspard Zoss (DisneyResearch|Studios); Matthias Niessner (Technical Univ ersity Munich); and Derek Bradley (DisneyResearch|Studios)\n-------------- -------\nVariational Elastodynamic Simulation\n\nThis paper shows how to e xpress variational time integration for a large class of elastic energies as an optimization problem with a “hidden” convex substructure. Our integr ator improves the performance of elastic simulation tasks, while conservin g physical invariants up to tolerance/num...\n\n\nLeticia Mattos Da Silva (Massachusetts Institute of Technology (MIT)); Silvia Sellán (Massachusett s Institute of Technology (MIT), Columbia University); and Natalia Pacheco -Tallaj and Justin Solomon (Massachusetts Institute of Technology (MIT))\n ---------------------\nStitch-A-Shape: Bottom-up Learning for B-Rep Genera tion\n\nStitch-A-Shape introduces a novel framework for generating B-Rep m odels by directly addressing both topology and geometry. Using a sequentia l stitching approach, it assembles 3D shapes from vertices through curves to faces, effectively managing topological and geometric complexities. The framework d...\n\n\nPu Li (Institute of Automation, Chinese Academy Of Sc iences) and Wenhao Zhang, Jinglu Chen, and Dongming Yan (Institute of Auto mation, Chinese Academy of Sciences)\n---------------------\nMonocular Onl ine Reconstruction with Enhanced Detail Preservation\n\nWe propose a high- quality online reconstruction pipeline for monocular input streams, recons tructing environments with detail across multiple levels while maintaining high speed.\n\n\nSongyin Wu (Meta Reality Labs Research, University of Ca lifornia Santa Barbara); Zhaoyang Lv, Yufeng Zhu, Duncan Frost, and Zhengq in Li (Meta Reality Labs Research); Ling-Qi Yan (University of California Santa Barbara); and Carl Ren, Richard Newcombe, and Zhao Dong (Meta Realit y Labs Research)\n---------------------\nAdaptive Phase-Field-FLIP for Ver y Large Scale Two-Phase Fluid Simulation\n\nWe present an algorithm for si mulating large-scale, violently turbulent two-phase flows—such as breaking ocean waves, tsunamis, and asteroid impacts—at extreme resolutions of the coupled water-air velocity field. This is achieved by integrating a new m ultiphase FLIP variant with highly e...\n\n\nBernhard Braun (Technical Uni versity Munich), Jan Bender (RWTH Aachen University), and Nils Thuerey (Te chnical University Munich)\n---------------------\nSpline Deformation Fiel d\n\nWe combine splines, a classical tool from applied mathematics, with implicit Coordinate Neural Networks to model deformation fields, achieving strong performance across multiple datasets. The explicit regularization from spline interpolation enhances spatial coherency in challenging scenar ios. We f...\n\n\nMingyang Song (Disney Research Studios, ETH Zürich); Yan g Zhang (Disney Research Studios); Marko Mihajlovic and Siyu Tang (ETH Zür ich); Markus Gross (ETH Zürich, Disney Research Studios); and Tunc Ozan Ay din (Disney Research Studios)\n---------------------\nCK-MPM: A Compact-Ke rnel Material Point Method\n\nWe introduce a compact, C2-continuous kernel for MPM that reduces numerical diffusion and improves efficiency—without sacrificing stability. Built on a dual-grid framework and compatible with APIC and MLS, our method enables high-fidelity, large-scale simulations, f urther pushing the limits of...\n\n\nMichael Liu (Carnegie Mellon Universi ty), Xinlei Wang (NetEase Games Messiah Engine), and Minchen Li (Carnegie Mellon University)\n---------------------\nAdaptive Algebraic Reuse of Reo rdering in Cholesky Factorizations with Dynamic Sparsity Patterns\n\nParth delivers adaptive fill-reducing ordering to accelerate Cholesky solvers i n simulations with dynamic sparsity patterns, such as contact modelling, a chieving up to 255× ordering speedups. With seamless, three-line integrati on into popular solvers like MKL and Accelerate, Parth ensures reliable, . ..\n\n\nBehrooz Zarebavani (University of Toronto), Danny M. Kaufman (Adob e Research), and David I. W. Levin and Maryam Mehri Dehnavi (University of Toronto)\n---------------------\nMIND: Microstructure INverse Design with Generative Hybrid Neural Representation\n\nWe introduce MIND, a novel gen erative framework for inverse-designing diverse, tileable 3D microstructur es. Leveraging latent diffusion and our hybrid neural representation, MIND precisely achieves targeted physical properties, ensures geometric validi ty, and enables seamless boundary compatibility&...\n\n\nTianyang Xue, Lon gdu Liu, and Lin Lu (Shandong University); Paul Henderson (University of G lasgow); Pengbin Tang (ETH Zürich); Haochen Li, Jikai Liu, and Haisen Zhao (Shandong University); Hao Peng (CrownCAD); and Bernd Bickel (ETH Zürich) \n---------------------\nLeapfrog Flow Maps for Real-Time Fluid Simulation \n\nWe present Leapfrog Flow Maps (LFM), a fast hybrid velocity-impulse sc heme with leapfrog integration. The computations are further accelerated b y a matrix-free AMGPCG solver optimized for GPUs. As a result, LFM achieve s high performance and fidelity across diverse examples, including firebal ls and w...\n\n\nYuchen Sun, Junlin Li, Ruicheng Wang, Sinan Wang, and Zhi qi Li (Georgia Institute of Technology); Bart G. van Bloemen Waanders (San dia National Laboratories); and Bo Zhu (Georgia Institute of Technology)\n ---------------------\nBernstein Bounds for Caustics\n\nWe derive vertex p osition and irradiance bounds for each triangle tuple, introducing a bound ing property of rational functions on the Bernstein basis, to significantl y reduce the search domain when systematically simulating specular light t ransport.\n\n\nZhimin Fan, Chen Wang, Yiming Wang, Boxuan Li, and Yuxuan G uo (Nanjing University); Ling-Qi Yan (University of California Santa Barba ra); and Yanwen Guo and Jie Guo (Nanjing University)\n-------------------- -\nKinematic Motion Retargeting for Contact-Rich Anthropomorphic Manipulat ions\n\nWe present a simple, but effective framework for kinematically ret argeting contact-rich anthropomorphic hand-object manipulations by exploit ing contact areas. We reliably retarget contact area data between diverse hands using a novel non-isometric shape matching process and generate high quality res...\n\n\nArjun Lakshmipathy, Jessica Hodgins, and Nancy Pollar d (Carnegie Mellon University) and Arjun Lakshmipathy\n------------------- --\nClebsch Gauge Fluid on Particle Flow Maps\n\nWe present a Clebsch PFM fluid solver that accurately transports wave functions using particle flow maps. Key innovations include a new gauge transformation, improved veloci ty reconstruction on coarse grids, and better fine-scale structure preserv ation. Benchmarks show superior performance over impu...\n\n\nZhiqi Li, Ca ndong Lin, Duowen Chen, and Xinyi Zhou (Georgia Institute of Technology); Shiying Xiong (Zhejiang University); and Bo Zhu (Georgia Institute of Tech nology)\n---------------------\nMASH: Masked Anchored SpHerical Distances for 3D Shape Representation and Generation\n\nWe introduce Masked Anchored SpHerical Distances (MASH), a novel multi-view and parametrized represent ation of 3D shapes. MASH is versatilefor multiple applications including s urface reconstruction, shape generation, completion, and blending, achievi ng superior performance thanks to its unique repre...\n\n\nChanghao Li and Yu Xin (University of Science and Technology of China); Xiaowei Zhou (Sta te Key Laboratory of CAD & CG, Zhejiang University); Ariel Shamir (Reichma n University); Hao Zhang (Simon Fraser University); Ligang Liu (University of Science and Technology of China); and Ruizhen Hu (Shenzhen University) \n---------------------\nGAIA: Generative Animatable Interactive Avatars w ith Expression-conditioned Gaussians\n\nWe present GAIA (Generative Animat able Interactive Avatars) for high-fidelity 3D head avatar generation. GAI A learns dynamic details with expression-conditioned Gaussians, while bein g animatable consistently with an underlying morphable model. With a novel two-branch architecture, GAIA disentangles ...\n\n\nZhengming Yu (Texas A &M University); Tianye Li (NVIDIA); Jingxiang Sun (Tsinghua University, NV IDIA); Omer Shapira, Seonwook Park, Michael Stengel, and Matthew Chan (NVI DIA); Xin Li and Wenping Wang (Texas A&M University); and Koki Nagano and Shalini De Mello (NVIDIA)\n---------------------\nAlignTex: Pixel-Precise Texture Generation from Multi-view Artwork\n\nAlignTex is a novel framewor k for generating high-quality textures from 3D meshes and multi-view artwo rk. It improves texture generation by ensuring both appearance detail and geometric consistency, outpacing traditional methods in quality and effici ency, making it a valuable tool for 3D asset creat...\n\n\nYuqing Zhang, H ao Xu, and Yiqian Wu (Zhejiang University; State Key Laboratory of CAD&CG, Zhejiang University); Sirui Chen and Sirui Lin (Zhejiang University); Xia ng Li (Shenzhen University); Xifeng Gao (Lightspeed Studios, Tencent Ameri ca); and Xiaogang Jin (Zhejiang University; State Key Laboratory of CAD&CG , Zhejiang University)\n---------------------\nMotion Control via Metric-A ligning Motion Matching\n\nMetric-Aligning Motion Matching (MAMM) is a nov el method for controlling motion sequences using sketches, labels, audio, or another motion sequence without requiring training or annotations. By a ligning within-domain distances, MAMM provides a flexible and efficient so lution for motion control acros...\n\n\nNaoki Agata and Takeo Igarashi (Th e University of Tokyo)\n---------------------\nVariational Surface Reconst ruction Using Natural Neighbors\n\nWe introduced a new surface reconstruct ion method from points without normals. The method robustly handles unders ampled regions and scales to large input sizes.\n\n\nJianjun Xia and Tao J u (Washington University in St. Louis)\n---------------------\nCora: Corre spondence-aware image editing using few step diffusion\n\nCora is a novel diffusion-based image editing method that achieves complex edits, such as object insertion, background changes, and non-rigid transformations, in on ly four diffusion steps. By leveraging pixel-wise semantic correspondences between source and target, it preserves key elements of the o...\n\n\nAmi rhossein Alimohammadi, Aryan Mikaeili, and Sauradip Nag (Simon Fraser Univ ersity); Negar Hassanpour (Huawei Canada); and Andrea Tagliasacchi and Ali Mahdavi-Amiri (Simon Fraser University)\n---------------------\nCAST: Com ponent-Aligned 3D Scene Reconstruction from an RGB Image\n\nWe introduce C AST, an innovative method for reconstructing high-quality 3D scenes from a single RGB image. Supporting open-vocabulary reconstruction, CAST excels in managing occlusions, aligning objects accurately, and ensuring physica l consistency with the input, unlocking new possibilities in vir...\n\n\nK aixin Yao, Longwen Zhang, Xinhao Yan, Yan Zeng, and Qixuan Zhang (Shanghai Tech University, Deemos); Jiayuan Gu (ShanghaiTech University); Wei Yang ( Huazhong University of Science and Technology); and Lan Xu and Jingyi Yu ( ShanghaiTech University)\n---------------------\nPrimitiveAnything: Human- Crafted 3D Primitive Assembly Generation with Auto-Regressive transformer\ n\nWe present PrimitiveAnything, a novel framework that reformulates shape primitive abstraction as a primitive assembly generation task. PrimitiveA nything can generate 3D high-quality primitive assemblies that better alig n with human perception while maintaining geometric fidelity across divers e shape...\n\n\nJingwen Ye (Tencent AIPD); Yuze He (Tencent AIPD, Tsinghua University); Yanning Zhou, Yiqin Zhu, and Kaiwen Xiao (Tencent AIPD); Yon g-Jin Liu (Tsinghua University); and Wei Yang and Xiao Han (Tencent AIPD)\ n---------------------\nMulti-Person Interaction Generation from Two-Perso n Motion Priors\n\nGenerate exciting multi-character interactions, such as team fights, with our training-free method! Multi-character interactions can be decomposed into multiple two-person interactions using a directed g raph, which enables repurposing large pre-trained two-character motion syn thesis models without a...\n\n\nWenning Xu, Shiyu Fan, Paul Henderson, and Edmond S. L. Ho (University of Glasgow)\n---------------------\nSpeculati ve AI Re-enactment of the Figurists' Encounters With the I Ching\n\nInstea d of pursuing the concern of AI displacing artists, we emphasise a role fo r artists in reshaping technology and branching it in new directions. A ro le that places us less as a user of AI technology, waiting for its creativ e outputs, but as a maker of what AI can be, perhaps leading us towards .. .\n\n\nIsaac Joseph Clarke (Hong Kong University of Science and Technology (Guangzhou)); Raul Masu (Hong Kong University of Science and Technology, Guangzhou; Conservatorio F.A. Bonporti, Italy); and Theo Papatheodorou (Ho ng Kong University of Science and Technology, Guangzhou)\n---------------- -----\nFeeling Blue or Seeing Red? Investigating the effect of light color , shadow and realism on the perception of emotion of real and virtual huma ns\n\nThis study explores how light color influences the perception of emo tion of virtual characters. By analyzing various lighting conditions, incl uding red and blue hues, we reveal how light affects emotion intensity, re cognition, and genuineness. Findings show that lighting, realism, and shad ows are ke...\n\n\nRachel McDonnell and Bharat Vyas (Trinity College Dubli n), Uros Sikimic (Epic Games), and Pisut Wisessing (CMKL University)\n---- -----------------\nAsymptotic analysis and design of linear elastic shell lattice metamaterials\n\nThis paper introduces a novel asymptotic directio nal stiffness (ADS) metric to analyze the contribution of middle surface g eometry on the stiffness of shell lattice metamaterials, focusing on Tripl y Periodic Minimal Surfaces (TPMS). It provides a theoretical framework an d optimization techniques, ad...\n\n\nDi Zhang and Ligang Liu (University of Science and Technology of China)\n---------------------\nAMOR: Adaptive Character Control through Multi-Objective Reinforcement Learning\n\nPrese nting AMOR, a policy conditioned on context and a linear combination of re ward weights, trained using multi-objective reinforcement learning. Once t rained, AMOR allows for on-the-fly adjustments of reward weights, unlockin g new possibilities in physics-based and robotic character control.\n\n\nL ucas N. Alegre (Instituto de Informática - Universidade Federal do Rio Gra nde do Sul, Disney Research) and Agon Serifi, Ruben Grandia, David Müller, Espen Knoop, and Moritz Bächer (Disney Research)\n---------------------\n Reimagining Beckett’s Not I in Virtual Reality: The MetaHuman as a Digital Double of the Actor\n\nThis practice-based project reimagines Beckett’s N ot I in virtual reality, marrying minimalist theatre with immersive techno logy. A lone, disembodied Metahuman mouth exploits VR’s intense presence w hile subverting customary embodiment and audience agency. Integrating perf orming avatars, ...\n\n\nNéill O'Dwyer, Enda Bates, and Nicholas Johnson ( Trinity College Dublin)\n---------------------\nBoolean Operation for CAD Models Using a Hybrid Representation\n\nWe propose a novel algorithm for e fficient and accurate Boolean operations on B-Rep models by mapping them b ijectively to controllable-error triangle meshes. Using conservative inter section detection on the mesh to locate all surface intersection curves an d carefully handling degeneration and topolo...\n\n\nYingyu Yang and Xiaoh ong Jia (State Key Laboratory of Mathematical Sciences, Academy of Mathema tics and Systems Science, Chinese Academy of Sciences; University of Chine se Academy of Sciences); Bolun Wang (Visual Computing Institute, RWTH Aach en University); Jieyin Yang (State Key Laboratory of Mathematical Sciences , Academy of Mathematics and Systems Science, Chinese Academy of Sciences; University of Chinese Academy of Sciences); Shiqing Xin (Shandong Univers ity); and Dong-Ming Yan (MAIS, Institute of Automation, Chinese Academy of Sciences; University of Chinese Academy of Sciences)\n------------------- --\nGaVS: 3D-Grounded Video Stabilization via Temporally-Consistent Local Reconstruction and Rendering\n\nGaVS: Transform unstable shaky videos into smooth, professional-quality footage. We design novel 3D rednering techno logy that preserves the motion intent while eliminating shakes and distort ions—no cropping, no distortion and workable under dynamics and intense mo tions. GaVS delivers natural-l...\n\n\nZinuo You (ETH Zürich, Huawei Resea rch Zürich); Stamatios Georgoulis (Huawei Research Zürich); Anpei Chen (ET H Zürich, University of Tuebingen); Siyu Tang (ETH Zürich); and Dengxin Da i (Huawei Research Zürich)\n---------------------\nDress-1-to-3: Single Im age to Simulation-Ready 3D Outfit with Diffusion Prior and Differentiable Physics\n\nWe introduce Dress-1-to-3 to reconstruct physics-plausible, sim ulation-ready separated garments from an in-the-wild image. Starting with the image, our approach combines a pre-trained image-to-sewing pattern gen eration model with a pre-trained multi-view diffusion model. The sewing pa ttern is refine...\n\n\nXuan Li, Chang Yu, Wenxin Du, Ying Jiang, Tianyi X ie, and Yunuo Chen (University of California Los Angeles); Yin Yang (Unive rsity of Utah); and Chenfanfu Jiang (University of California Los Angeles) \n---------------------\nNeural Importance Sampling of Many Lights\n\nNeur al approach for estimating spatially varying light selection distributions to improve importance sampling in Monte Carlo rendering. To efficiently m anage hundreds or thousands of lights, we integrate our neural approach wi th light hierarchy techniques, where the network predicts cluster-level di ...\n\n\nPedro Figueiredo and Qihao He (Texas A&M University), Steve Bako (Aurora Innovation), and Nima Khademi Kalantari (Texas A&M University)\n-- -------------------\nNAM: Neural Adjoint Maps for refinement of shape corr espondences\n\nWe introduce Neural Adjoint Maps, a novel representation fo r correspondences between 3D shapes. Built on and extending the functional map framework, our approach enables accurate, non-linear refinement of sh ape matching across meshes and point clouds, setting a new standard in div erse scenarios and ...\n\n\nGiulio Viganò (Università di Milano Bicocca), Maks Ovsjanikov (Centre National de la Recherche Scientifique - Laboratoir e d'informatique de l'École Polytechnique (LIX)), and Simone Melzi (Univer sità di Milano Bicocca)\n---------------------\nDigital Animation of Powde r-Snow Avalanches\n\nPowder-snow avalanches are natural phenomena that res ult from an instability in the snow cover on a mountain relief. This paper introduces a physically-based framework to simulate powder-snow avalanche s under complex terrains, allowing us to animate the turbulent snow cloud dynamics within the avala...\n\n\nFilipe Nascimento, Fabricio S. Sousa, an d Afonso Paiva (Universidade de São Paulo - USP)\n---------------------\nA erial Path Online Planning for Urban Scene Updation\n\nWe present the firs t scene-update aerial path planning algorithm specifically designed for de tecting and updating change areas in urban environments, which paves the w ay for efficient, scalable, and adaptive UAV-based scene updates in comple x urban environments.\n\n\nMingfeng Tang (Shenzhen University), Ningna Wan g (University of Texas at Dallas), Ziyuan Xie (Shenzhen University), Jianw ei Hu (QiYuan Lab), Ke Xie (Shenzhen University), Xiaohu Guo (University o f Texas at Dallas), and Hui Huang (Shenzhen University)\n----------------- ----\nSqueezeMe: Mobile-Ready Distillation of Gaussian Full-Body Avatars\n \nExisting Gaussian Splatting avatars require desktop GPUs, limiting mobil e device use. SqueezeMe converts these avatars into a lightweight represen tation, enabling real-time animation and rendering on mobile devices. By d istilling the corrective decoder into an efficient linear model, SqueezeMe achie...\n\n\nForrest Iandola, Stanislav Pidhorskyi, Igor Santesteban, Di vam Gupta, Anuj Pahuja, Nemanja Bartolovic, Frank Yu, Emanuel Garbin, Toma s Simon, and Shunsuke Saito (Meta)\n---------------------\nPS-CAD: Local G eometry Guidance via Prompting and Selection for CAD Reconstruction\n\nWe propose an iterative prompt-and-select architecture to progressively recon struct the CAD modeling sequence of a target point cloud. We propose the c oncept of local geometric guidance and come up with three ways to integrat e this guidance into iterative reconstruction. Experiments demonstrate the ...\n\n\nBingchen Yang and Haiyong Jiang (School of Artificial Intelligen ce, University of Chinese Academy of Sciences); Hao Pan (Tsinghua Universi ty); Guosheng Lin (Nanyang Technological University); Jun Xiao (School of Artificial Intelligence, University of Chinese Academy of Sciences); Peter Wonka (KAUST); and Bingchen Yang\n---------------------\nTowards Comprehe nsive Neural Materials: Dynamic Structure-Preserving Synthesis with Accura te Silhouette at Instant Inference Speed\n\nWe challenge the comprehensive neural material representation by thoroughly considering the essential as pects of the complete appearance. We introduce an int8-quantized model tha t keeps high fidelity while achieving an order of magnitude speedup compar ed to previous methods, and a controllable struc...\n\n\nZilin Xu (Univers ity of California Santa Barbara); Xiang Chen (Shandong University); Chen L iu (Zhejiang Lingdi Digital Technology Co.,Ltd); Beibei Wang (Nanjing Univ ersity); Lu Wang (Shandong University); Zahra Montazeri (University of Man chester); and Ling-Qi Yan (University of California Santa Barbara)\n------ ---------------\nGSHeadRelight: Fast Relightability for 3D Gaussian Head S ynthesis\n\nGSHeadRelight enables fast, high-quality relightability for 3D Gaussian head synthesis. A linear light model based on learnable radiance transfer is integrated into the native 3DGS rasterization process and sup ports colored illumination. Without requiring expensive light stage data, our method achie...\n\n\nHenglei Lv (Institute of Computing Technology, Ch inese Academy of Sciences; University of Chinese Academy of Sciences); Bai lin Deng (Cardiff University); Jianzhu Guo, Xiaoqiang Liu, Pengfei Wan, an d Di Zhang (Kuaishou Technology); and Lin Gao (Institute of Computing Tech nology, Chinese Academy of Sciences)\n---------------------\nDiffuse-CLoC: Guided Diffusion for Physics-based Character Look-ahead Control\n\nMeet D iffuse-CLoC—a powerful unification of intuitive steering in kinematic moti on generation and physics-based character control. By guiding diffusion ov er joint state-action spaces, it enables agile, steerable, and physically realistic motions across diverse downstream tasks—from obsta...\n\n\nXiaoy u Huang (University of California Berkeley, Robotics and AI Institute); Ta kara Truong (Stanford University, Robotics and AI Institute); Yunbo Zhang, Fangzhou Yu, Jean Pierre Sleiman, and Jessica Hodgins (Robotics and AI In stitute); Koushil Sreenath (Robotics and AI Institute, University of Calif ornia Berkeley); and Farbod Farshidian (Robotics and AI Institute)\n------ ---------------\nPhotoreal Scene Reconstruction from an Egocentric Device\ n\nThis paper investigates photorealistic scene reconstruction using video s captured from an egocentric device in high dynamic range. It presents a novel system utilizing visual-inertial bundle adjustment and a physical im age formation model that handles camera motion artifacts. The experiments using P...\n\n\nZhaoyang Lv, Maurizio Monge, Ka Chen, Yufeng Zhu, Michael Goesele, Jakob Engel, Zhao Dong, and Richard Newcombe (Reality Labs Resear ch, Meta)\n---------------------\nTokenVerse: Versatile Multi-concept Pers onalization in Token Modulation Space\n\nTokenVerse extracts complex visua l elements from images by identifying semantic directions in per-token mod ulation space of DiT models for each word in the image caption. It's capab le of combining concepts from multiple sources by adding corresponding dir ections, enabling flexible generation of new ...\n\n\nDaniel Garibi (Tel A viv University, DeepMind); Shahar Yadin (Technion - Israel Institute of Te chnology, DeepMind); Roni Paiss, Omer Tov, Shiran Zada, and Ariel Ephrat ( DeepMind); Tomer Michaeli (Technion - Israel Institute of Technology, Deep Mind); Inbar Mosseri (DeepMind); and Tali Dekel (Weizmann Institute of Sci ence, DeepMind)\n---------------------\nDC-VSR: Spatially and Temporally C onsistent Video Super-Resolution with Video Diffusion Prior\n\nWe propose DC-VSR, a novel video super-resolution approach based on a video diffusion prior. DC-VSR leverages Spatial and Temporal Attention Propagation (SAP a nd TAP) to ensure spatio-temporally consistent results and Detail-Suppress ion Self-Attention Guidance (DSSAG) to enhance high-frequency detai...\n\n \nJanghyeok Han, Gyujin Sim, and Geonung Kim (POSTECH); Hyun-Seung Lee, Ky uha Choi, and Youngseok Han (Samsung Electronics); and Sunghyun Cho (POSTE CH)\n---------------------\nReservoir Splatting for Temporal Path Resampli ng and Motion Blur\n\nWe introduce reservoir splatting, a technique preser ving exact primary hits during temporal ReSTIR. This approach makes tempor al path resampling more robust under motion, especially for regions with h igh-frequency detail. We further demonstrate how reservoir splatting natur ally enables ReSTIR suppor...\n\n\nJeffrey Liu (University of Illinois Urb ana-Champaign); Daqi Lin, Markus Kettunen, and Chris Wyman (NVIDIA); and R avi Ramamoorthi (NVIDIA, University of California San Diego)\n------------ ---------\nPosition-Normal Manifold for Efficient Glint Rendering on High- Resolution Normal Maps\n\nAccurate modeling of normal distribution functio ns (NDF) over a high-resolution normal map enables intriguing glinty appea rance but is inefficient. We present a manifold-based glint formulation, t ransferring the glint NDF computation to mesh intersections. This framewor k accelerates glint rendering,...\n\n\nLiwen Wu (University of California San Diego), Fujun Luan and Miloš Hašan (Adobe Research), and Ravi Ramamoor thi (University of California San Diego)\n---------------------\nPutting R igid Bodies to Rest\n\nWe identify stable orientations of any rigid shape, and the probability that it will rest at these orientations if randomly d ropped on the ground. We use a differentiable inverse version of our metho d to design and fabricate shapes with target resting behavior, such as dic e with target, nonuniform p...\n\n\nHossein Baktash (Carnegie Mellon Unive rsity), Nicholas Sharp (NVIDIA), Qingnan Zhou and Alec Jacobson (Adobe Res earch), and Keenan Crane (Carnegie Mellon University)\n------------------- --\nELGAR: Expressive Cello Performance Motion Generation for Audio Rendit ion\n\nGenerating string instrument performances with intricate movements and complex interactions poses significant challenges. To address these, w e present ELGAR—the first diffusion-based framework for whole-body instrum ent performance motion generation solely from audio. We further contribute inno...\n\n\nZhiping Qiu and Yitong Jin (Central Conservatory of Music, T singhua University); Yuan Wang (Central Conservatory of Music); Yi Shi (Ce ntral Conservatory of Music, Tsinghua University); Chao Tan (Weilan Tech); Chongwu Wang, Xiaobing Li, and Feng Yu (Central Conservatory of Music); a nd Tao Yu and Qionghai Dai (Tsinghua University)\n---------------------\nE DGE: Epsilon-Difference Gradient Evolution for Buffer-Free Flow Maps\n\nTh is paper presents Epsilon Difference Gradient Evolution (EDGE), a novel me thod for accurate flow-map computation on grids without velocity buffers. EDGE enables large-scale, efficient and high-fidelity fluid simulations th at capture and preserve complex vorticity structures while significantly r ed...\n\n\nZhiqi Li, Ruicheng Wang, Junlin Li, Duowen Chen, Sinan Wang, an d Bo Zhu (Georgia Institute of Technology)\n---------------------\nBe Deci sive: Noise-Induced Layouts for Multi-Subject Generation\n\nText-to-image diffusion models struggle with multi-subject generation due to subject lea kage. Prior methods impose external layouts that conflict with the model’s prior, harming alignment and natural composition. We introduce a method t hat leverages the layout encoded in the initial noise, pro...\n\n\nOmer Da hary (Tel Aviv University, Snap Research); Yehonathan Cohen (Tel Aviv Univ ersity); Or Patashnik (Tel Aviv University, Snap Research); Kfir Aberman ( Snap Research); and Daniel Cohen-Or (Tel Aviv University, Snap Research)\n ---------------------\nConformal First Passage for Epsilon-free Walk-on-Sp heres\n\nWe present a novel Monte Carlo approach to solve boundary integra l equations with Dirichlet boundary conditions in two dimensions. While Wa lk-on-Spheres uses largest empty circles, which touch the boundary in only one point, we utilize semicircles and circle sectors that share one or tw o boundary ed...\n\n\nPaul Himmler and Tobias Günther (Friedrich-Alexander -Universität Erlangen-Nürnberg (FAU))\n---------------------\nDuetGen: Mus ic Driven Two-Person Dance Generation via Hierarchical Masked Modeling\n\n We present a framework for generating music-driven synchronized two-person dance animations with close interactions. Our system represents the two-p erson motion sequence as a cohesive entity, performs hierarchical encoding of the motion sequence into discrete tokens, and utilizes dual generative mas...\n\n\nAnindita Ghosh (DFKI, Max Planck Institute for Informatics); Bing Zhou (Snap Inc.); Rishabh Dabral (Max Planck Institute for Informatic s); Jian Wang (Snap Inc.); Vladislav Golyanik and Christian Theobalt (Max Planck Institute for Informatics); Philipp Slusallek (DFKI, Saarland Unive rsity); and Chuan Guo (Snap Inc.)\n---------------------\nFashionComposer: Compositional Fashion Image Generation\n\nFashionComposer is a flexible m odel for compositional fashion image generation, with a universal framewor k that handles diverse input modalities such as text, human models, and ga rment images. It personalizes appearance, pose, and human figure, using su bject-binding attention to integrate reference ...\n\n\nSihui Ji, Yiyang W ang, and Xi Chen (The University of Hong Kong); Xiaogang Xu (The Chinese U niversity of Hong Kong); Hao Luo (DAMO Academy, Alibaba Group); and Hengsh uang Zhao (The University of Hong Kong)\n---------------------\nClaycode: Stylable and Deformable 2D Scannable Codes\n\nWe introduce a novel scannab le 2D code where the payload is stored in the topology of nested color reg ions, abandoning traditional matrix-based approaches (e.g., QRCodes). Clay codes can be largely deformed, styled, and animated. We present a mapping between bits and topologies, shape-constrained ren...\n\n\nMarco Maida, Al berto Crescini, Marco Perronet, and Elena Camuffo (Independent Researcher) \n---------------------\nSplat4D: Diffusion-Enhanced 4D Gaussian Splatting for Temporally and Spatially Consistent Content Creation\n\nSplat4D gener ates high-fidelity 4D content from monocular videos by integrating multi-v iew rendering, inconsistency identification, a video diffusion model, and asymmetric U-Net refinement. Our framework maintains spatial-temporal cons istency while preserving details and following user guidance, ach...\n\n\n Minghao Yin (University of Hong Kong); Yukang Cao (Nanyang Technological U niversity, Singapore); Songyou Peng (Google DeepMind); and Kai Han (Univer sity of Hong Kong)\n---------------------\nFeature-Preserving Mesh Repair via Restricted Power Diagram\n\nWe present a unified mesh repair framework using a manifold wrap surface to fix diverse imperfections while preservi ng sharp features. By optimizing projected samples and leveraging adaptive weighting, our method ensures watertightness, manifoldness, and high geom etric fidelity, outperforming existi...\n\n\nHuibiao Wen (Shandong Univers ity, University of Health and Rehabilitation Sciences); Guilong He (Shando ng University); Rui Xu (University of Hong Kong); Shuangmin Chen (Qingdao University of Science and Technology); Shiqing Xin (Shandong University); Zhenyu Shu (NingboTech University); Taku Komura (University of Hong Kong); Jieqing Feng (State Key Laboratory of CAD & CG, Zhejiang University); Wen ping Wang (Texas A&M University); and Changhe Tu (Shandong University)\n-- -------------------\nLightLab: Controlling Light Sources in Images with Di ffusion Models\n\nLightLab is a diffusion-based method for parametric cont rol over light sources in an image. Leveraging the linearity of light we c reate a dataset of controled illumniation changes from a small set of real image pairs and synthetic renders, which is used to fine-tune a model to enable physically plau...\n\n\nNadav Magar (Tel Aviv University, Google); Amir Hertz, Eric Tabellion, Yael Pritch, and Alex Rav-Acha (Google); Ariel Shamir (Reichman University, Google); and Yedid Hoshen (Hebrew University of Jerusalem, Google)\n---------------------\nThe Mokume Dataset and Inve rse Modeling of Solid Wood Textures\n\nWe present the Mokume dataset for s olid wood texturing, comprising nearly 190 samples from various species. U sing this dataset, we propose an inverse modeling pipeline to infer volume tric wood textures from surface photographs, employing inverse procedural texturing and neural cellular automata (NCA...\n\n\nMaria Larsson (The Uni versity of Tokyo); Hodaka Yamaguchi (Gifu Prefecture Research Institute fo r Human Life Technology, Nihon University); Ehsan Pajouheshgar (École Poly technique Féderale de Lausanne (EPFL)); I-Chao Shen and Kenji Tojo (The Un iversity of Tokyo); Chia-Ming Chang (National Taiwan University of Arts); Lars Hansson (Luleå University of Technology, Norwegian University of Scie nce and Technology); Olof Broman (Luleå University of Technology); Takashi Ijiri (Shibaura Institute of Technology); Ariel Shamir (Reichman Universi ty); Wenzel Jakob (The University of Tokyo, École Polytechnique Féderale d e Lausanne (EPFL)); and Takeo Igarashi (The University of Tokyo)\n-------- -------------\nA Closest Point Method for PDEs on Manifolds with Interior Boundary Conditions for Geometry Processing\n\nGeometry processing often r equires the solution of PDEs with boundary conditions on the manifold’s in terior. However, input manifolds can take many forms, each requiring speci alized discretizations. Instead, we develop a unified framework for genera l manifold representations by extending the c...\n\n\nNathan King (Univers ity of Waterloo), Haozhe Su (LightSpeed Studios), Mridul Aanjaneya (Rutger s University), Steven Ruuth (Simon Fraser University), Christopher Batty ( University of Waterloo), and Nathan King\n---------------------\nCageNet: A Meta-Framework for Learning on Wild Meshes\n\nWe propose a framework for learning on in-the-wild meshes containing non-manifold elements, multiple components, and interior structures. Our approach uses cages and generali zed barycentric coordinates to parametrize and learn volumetric functions, demonstrated by segmentation and skinning weights, ...\n\n\nMichal Edelst ein (Technion – Israel Institute of Technology); Hsueh-Ti Derek Liu (Roblo x, University of British Columbia); and Mirela Ben-Chen (Technion - Israel Institute of Technology)\n---------------------\nHumanRAM: Feed-forward H uman Reconstruction and Animation Model using Transformers\n\nExisting ava tar methods typically require sophisticated dense-view capture and/or time -consuming per-subject optimization processes. HumanRAM proposes a feed-fo rward approach for generalizable human reconstruction and animation from m onocular or sparse human images. Experiments show that HumanRAM ac...\n\n\ nZhiyuan Yu (Department of Mathematics, Hong Kong University of Science an d Technology); Zhe Li (Huawei); Hujun Bao (State Key Laboratory of CAD&CG, Zhejiang University); Can Yang (Department of Mathematics, Hong Kong Univ ersity of Science and Technology); and Xiaowei Zhou (State Key Laboratory of CAD&CG, Zhejiang Univerisity)\n---------------------\nPaRas: A Rasteriz er for Large-Scale Parametric Surfaces\n\nHigher-order surfaces enable com pact, smooth geometry but require efficient rendering. We introduce PaRas, a GPU-based rasterizer that directly renders parametric surfaces, avoidin g costly tessellation. It integrates seamlessly into existing pipelines, o utperforming traditional methods for quartic t...\n\n\nKechun Wang and Ren jie Chen (University of Science and Technology of China)\n---------------- -----\nWhat is HDR? Perceptual Impact of Luminance and Contrast in Immersi ve Displays\n\nWe studied preferences for different contrasts and peak lum inances in HDR. To do this, we collected a new HDR video dataset, develope d tone mappers, and built an HDR haploscope that can reproduce high lumin ance and contrast. Data was fit to a model which is used for applications like display design...\n\n\nKenneth Chen (New York University; Reality Lab s Research, Meta); Nathan Matsuda, Jon McElvain, Yang Zhao, and Thomas Wan (Reality Labs Research, Meta); Qi Sun (New York University); and Alexandr e Chapiro (Reality Labs Research, Meta)\n---------------------\nMGPBD: A M ultigrid Accelerated Global XPBD Solver\n\nIn high-stiffness, high-resolut ion simulations, while primal space methods typically fail, the dual-space XPBD method produces unphysical softening artifacts due to convergence st all. We design an innovative Algebraic Multigrid method to enhance XPBD, u tilizing lazy-update prolongators and near-kern...\n\n\nChunlei Li and Pen g Yu (Beihang University); Tiantian Liu (Taichi Graphics); Siyuan Yu (Zenu stech); and Yuting Xiao, Shuai Li, Aimin Hao, Yang Gao, and Qinping Zhao ( Beihang University)\n---------------------\nPoeSpin: A Human-AI Dance to P oetry System for Movement-Based Verse Generation\n\nPoeSpin is a human-AI cocreating system. By transforming pole dance movements into poetry throug h AI, we challenge both traditional prejudices against this art form and c onventional approaches to human-AI creativity. This work demonstrates how computational systems can preserve the deeply human aspe...\n\n\nYihua Li (Interactive Telecommunications Program, New York University); Hongyue Che n (English Literature and Literary Theory, University of Freiburg); Yiqing Li (Interactive Telecommunications Program, New York University); and Yet ong Xin (Graduate School of Design, Harvard University)\n----------------- ----\nPhysically Controllable Relighting of Photographs\n\nWe present a ph otograph relighting method that enables explicit control over light source s akin to CG pipelines. We achieve this in a pipeline involving mid-level computer vision, physically-based rendering, and neural rendering. We intr oduce a self-supervised training methodology to train our neura...\n\n\nCh ris Careaga and Yağız Aksoy (Simon Fraser University)\n------------------- --\nBrepDiff: Single-Stage B-rep Diffusion Model\n\nWe present BrepDiff, a simple, single-stage diffusion model for generating Boundary Representati ons (B-reps). Our approach generates B-reps by denoising point-based face samples with a dedicated noise schedule. Unlike multi-stage methods, BrepD iff enables intuitive, editable geometry creation, inclu...\n\n\nMingi Lee and Dongsu Zhang (Seoul National University), Clément Jambon (Massachuset ts Institute of Technology (MIT)), and Young Min Kim (Seoul National Unive rsity)\n---------------------\nIntrinsicEdit: Precise generative image man ipulation in intrinsic space\n\nA generative workflow for precise image ed iting using an intrinsic-image latent space. Built on RGB-X diffusion, it enables diverse edits—like relighting, color changes, and object manipulat ion—while preserving identity and ameliorating intrinsic-channel entanglem ent. All this is done wi...\n\n\nLinjie Lyu (Max-Planck-Institute for Info rmatics & Saarland Informatics Campus, Adobe Research); Valentin Deschaint re, Yannick Hold-Geoffroy, Milos Hasan, and Jae Shin Yoon (Adobe Research) ; Thomas Leimkuehler and Christian Theobalt (Max-Planck-Institute for Info rmatics & Saarland Informatics Campus); and Iliyan Georgiev (Adobe Researc h)\n---------------------\nShape Space Spectra\n\nWe introduce shape-space eigenanalysis to compute eigenfunctions across continuously-parameterized shape families. These eigenfunctions are obtained by minimizing a variati onal principle. To handle eigenvalue dominance swaps at points of multipli city, we incorporate dynamic reordering during optimiz...\n\n\nYue Chang a nd Otman Benchekroun (University of Toronto), Maurizio M. Chiaramonte (Met a Reality Labs Research), Peter Yichen Chen (MIT CSAIL), and Eitan Grinspu n (University of Toronto)\n---------------------\nInstanceGen: Image Gener ation with Instance-level Instructions\n\nWe propose InstanceGen - a new t echnique for improving Text-to-Image models ability to generate images for prompts describing multiple objects, attributes and spatial relationships . InstanceGen requires no training or additional user inputs and achieves state-of-the art results in terms of both accu...\n\n\nEtai Sella (Tel Avi v University, Meta); Yanir Kleiman (Meta); and Hadar Averbuch-Elor (Cornel l Tech)\n---------------------\nUnsupervised Decomposition of 3D Shapes in to Expressive and Editable Extruded Profile Primitives\n\n3D2EP transforms 3D shapes into expressive, editable primitives by extruding 2D profiles a long 3D curves. This approach creates compact, interpretable representatio ns that support intuitive editing and flexible redesign. It delivers high fidelity and efficiency, outperforming existing methods across...\n\n\nChu nyi Sun (Australian National University); Junlin Han and Runjia Li (Univer sity of Oxford); and Weijian Deng, Dylan Campbell, and Stephen Gould (Aust ralian National University)\n---------------------\nDigital F(r)ictions: R eimagining Colombian Art and its Territory\n\nBoth a critique and celebrat ion of digital representation, this project offers multiple perspectives b eyond technological homogenization. Through exploring digital f(r)ictions and multiplicities, we reject singular viewpoints in favor of interconnect ed truths. Our work with AI and Colombian art rais...\n\n\nAna María Zapat a Guzmán, Ludovica Schaerf, Darío Negueruela del Castillo, and Iacopo Neri (University of Zurich, Max Planck Society)\n---------------------\nA Plat form for Interactive AI Character Experiences\n\nWe present a platform for creating believable, conversational digital characters that combine conve rsational AI, speech, animation, memory, personality, and emotions. Demons trated through Digital Einstein, our system enables interactive, story-dri ven experiences and generalizes to any character, mak...\n\n\nRafael Wampf ler, Chen Yang, Dillon Elste, Nikola Kovacevic, Philine Witzig, and Markus Gross (ETH Zürich)\n---------------------\nModel See Model Do: Speech-Dri ven Facial Animation with Style Control\n\nModelSeeModelDo presents a spee ch-driven 3D facial animation method using a latent diffusion model condit ioned on a reference clip to capture nuanced performance styles. A novel " style basis" mechanism extracts key poses to guide generation, achieving e xpressive, temporally coherent animations with ...\n\n\nYifang Pan (Univer sity of Toronto; Jali Research, Canada); Karan Singh (University of Toront o); and Luiz Gustavo Hafemann (Ubisoft)\n---------------------\nUncertaint y for SVBRDF Acquisition using Frequency Analysis\n\nWe quantify uncertain ty for SVBRDF acquisition from multi-view captures using entropy. The othe rwise heavy computation is accelerated in the frequency domain, yielding a practical, efficient method. We apply uncertainty to improve SVBRDF captu re by guiding camera placement, inpainting uncertain regi...\n\n\nRuben Wi ersma (ETH Zürich); Julien Philip (Netflix Eyeline Studios); Miloš Hašan, Krishna Mullia, and Fujun Luan (Adobe Research); Elmar Eisemann (Delft Uni versity of Technology); and Valentin Deschaintre (Adobe Research)\n------- --------------\nRenderFormer: Transformer-based Neural Rendering of Triang le Meshes with Global Illumination\n\nWe present RenderFormer, a neural re ndering pipeline that directly renders an image from a triangle-based repr esentation of scene with full global illumination effects, and that does n ot require per-scene training or finetuning.\n\n\nChong Zeng (State Key La b of CAD and CG, Zhejiang University; Microsoft Research Asia); Yue Dong ( Microsoft Research Asia); Pieter Peers (College of William & Mary); Hongzh i Wu (State Key Lab of CAD and CG, Zhejiang University); and Xin Tong (Mic rosoft Research Asia)\n---------------------\nKernel Predicting Neural Sha dow Maps\n\nWe present a novel shadow method named kernel predicting neura l shadow mapping. By modeling soft shadow values as pixelwise local filter ing from basic hard shadow values, we trained a neural network to predict local filter weights, achieving accurate and temporally-stable soft shadow s with good gene...\n\n\nXuejun Hu, Jinfan Lu, and Kun Xu (Tsinghua Univer sity)\n---------------------\nBuildingBlock: A Hybrid Approach for Structu red Building Generation\n\nWe propose BuildingBlock, a hybrid approach int egrating generative models, PCG, and LLMs for diverse and structured 3D bu ilding generation. A Transformer-based diffusion model generates layouts, which LLMs refine into hierarchical designs. PCG then constructs high-qual ity buildings, achieving state-...\n\n\nJunming Huang, Chi Wang, Letian Li , and Changxin Huang (State Key Laboratory of CAD & CG, Zhejiang Universit y; LIGHTSPEED); Qiang Dai (LIGHTSPEED); and Weiwei Xu (State Key Lab CAD&C G, Zhejiang University, ZJU-Tencent Game and Intelligent Graphics Innovati on Technology Joint Lab)\n---------------------\nStyle Customization of Te xt-to-Vector Generation with Image Diffusion Priors\n\nWe propose a novel text-to-vector pipeline with style customization that disentangles content and style in SVG generation. Our method represents the first feed-forward text-to-vector diffusion model capable of generating SVGs in custom style s.\n\n\nPeiying Zhang (City University of Hong Kong), Nanxuan Zhao (Adobe Research), and Jing Liao (City University of Hong Kong)\n----------------- ----\nElastic Locomotion with Mixed Second-order Differentiation\n\nOur fr amework enables realistic and interesting elastic body locomotion by deter mining optimal muscle activations to achieve desired movements. It combine s interior-point method for contact modeling with a novel mixed second-ord er differentiation algorithm that merges analytic and numerical approach.. .\n\n\nSiyuan Shen, Tianjia Shao, and Kun Zhou (Zhejiang University); Chen fanfu Jiang (UCLA); Sheldon Andrews (École de Technologie Supérieure (ÉTS) ); Victor Zordan (Roblox); and Yin Yang (University of Utah)\n------------ ---------\nDAMO: A Deep Solver for Arbitrary Marker Configuration in Optic al Motion Capture\n\nThis paper introduces DAMO, a Deep solver for Arbitra ry Marker configuration in Optical motion capture. DAMO directly infers th e relationship between each raw marker point and 3D model joint, without u sing predefined marker labels and configuration information.\n\n\nKyeongMi n Kim and SeungWon Seo (Korea University), DongHeun Han (KyungHee Universi ty), HyeongYeop Kang (Korea University), and KyeongMin Kim\n-------------- -------\nRevisiting Tradition and Beyond: A Customized Bilateral Filtering Framework for Point Cloud Denoising\n\nTo combine deep learning's general ization with traditional methods' interpretability, we propose CustomBF—a hybrid framework that customizes bilateral filter components per point. By addressing key limitations of the classic bilateral filter, CustomBF achi eves robust, interpretable, and effect...\n\n\nPeng Li, Zeyong Wei, Honghu a Chen, Xuefeng Yan, and Mingqiang Wei (Nanjing University of Aeronautics and Astronautics)\n---------------------\nOctGPT: Octree-based Multiscale Autoregressive Models for 3D Shape Generation\n\nOctGPT is a novel multisc ale autoregressive model for 3D shape generation. It introduces hierarchic al serialized octree representation, octree-based transformer with 3D RoPE and token-parallel generation schemes. OctGPT significantly accelerates c onvergence, achieves performance rivaling or surpassi...\n\n\nSi-Tong Wei, Rui-Huan Wang, Chuan-Zhi Zhou, Baoquan Chen, and Peng-Shuai Wang (Peking University)\n---------------------\nCorrect your balance heuristic: Optimi zing balance-style multiple importance sampling weights\n\nMultiple import ance sampling (MIS) is vital to most rendering algorithms. MIS computes a weighted sum of samples from different techniques to handle diverse scene types and lighting effects.\nWe propose a practical weight correction sche me that yields better equal-time performance on bidirectional al...\n\n\nQ ingqin Hua and Pascal Grittmann (Saarland University) and Philipp Slusalle k (Saarland University, DFKI)\n---------------------\nModeling and Renderi ng Glow Discharge\n\nThis work presents a physically-based model for simul ating and rendering glow discharge, a luminous plasma effect seen in neon lights and gas discharge lamps. The model captures particle interactions a nd emission dynamics, integrates into volume rendering systems, and enable s realistic, interactive ...\n\n\nVenkataram Edavamadathil Sivaram, Ravi R amamoorthi, and Tzu-Mao Li (University of California San Diego)\n--------- ------------\nDiffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control\n\nDiffusion as Shader (DaS) is a unified approac h for controlled video generation that uses 3D tracking videos to enable v ersatile editing, including animating mesh-to-video, camera control, motio n transfer, and object manipulation, while improving temporal consistency. \n\n\nZekai Gu (Hong Kong University of Science and Technology), Rui Yan ( Zhejiang University), Jiahao Lu and Peng Li (Hong Kong University of Scien ce and Technology), Zhiyang Dou (University of Hong Kong), Chenyang Si (Na nyang Technological University), Zhen Dong (Wuhan University), Qifeng Liu (Hong Kong University of Science and Technology), Cheng Lin (University of Hong Kong), Ziwei Liu (Nanyang Technological University), Wenping Wang (T exas A&M University), and Yuan Liu (Hong Kong University of Science and Te chnology)\n---------------------\nVirCHEW Reality: On-Face Kinesthetic Fee dback for Enhancing Food-Intake Experience in Virtual Reality\n\nThis pape r presents VirCHEW Reality, a face-worn haptic device for virtual food int ake in VR. It uses pneumatic actuation to simulate food textures, enhancin g the chewing experience. User studies demonstrated its effectiveness in p roviding distinct kinesthetic feedback and improving virtual eating e...\n \n\nQingqin Liu, Ziqi Fang, and Jiayi Wu (School of Creative Media, City U niversity of Hong Kong); Shaoyu Cai (National University of Singapore); Ji anhui Yan (School of Creative Media, City University of Hong Kong); Tiande Mo and Shuk Ching CHAN (Hong Kong Productivity Council); and Kening Zhu ( City University of Hong Kong)\n---------------------\nTalking to the Midni ght Broadcast: Reviving 1990s City Memories with AI\n\nThis project explor es how AI can preserve and reinterpret cultural memory, raising profound q uestions about the role of technology in connecting past and future. By tr ansforming transient, everyday digital interactions into meaningful archiv es, it invites reflection on how today’s voices might...\n\n\nTongge Yu (T singhua University, Massachusetts Institute of Technology (MIT)) and Fan X iang (Tsinghua University)\n---------------------\nPartEdit: Fine-Grained Image Editing using Pre-Trained Diffusion Models\n\nWe present PartEdit, a novel diffusion-based system enabling precise, text-based edits of object parts without retraining or manual masks. Optimizing part-aware tokens ge nerates localized non-binary attention maps to guide seamless edits. Our n ovel blending strategy delivers high-quality visual resu...\n\n\nAleksanda r Cvejic, Abdelrahman Eldesokey, and Peter Wonka (King Abdullah University of Science and Technology (KAUST))\n---------------------\nPractical Styl ized Nonlinear Monte Carlo Rendering\n\nWe present a practical method for rendering scenes with complex, recursive nonlinear stylization applied to physically based rendering. Our approach introduces nonlinear path filteri ng(NL-PF) and nonlinear neural radiance caching(NL-NRC), which reduce the exponential sampling cost of stylized render...\n\n\nXiaochun Tong and Tos hiya Hachisuka (University of Waterloo)\n---------------------\npOps: Phot o-Inspired Diffusion Operators\n\npOps is a framework for learning semanti c manipulations in CLIP’s image embedding space. Built on a Diffusion Prio r model, it enables concept manipulation by training operators directly on image embeddings. This approach enhances semantic control and integrates easily with diffusion models for...\n\n\nElad Richardson (Tel Aviv Univers ity); Yuval Alaluf (Tel Aviv University, Snap); Ali Mahdavi-Amiri (Simon F raser University); and Daniel Cohen-Or (Tel Aviv University)\n------------ ---------\nDAM-VSR: Disentanglement of Appearance and Motion for Video Sup er-Resolution\n\nIn this work, we propose DAM-VSR, an appearance and motio n disentanglement framework for video super-resolution. Appearance enhance ment is achieved through reference image super-resolution, while motion co ntrol is achieved through video ControlNet. Additionally, we propose a mot ion-aligned bidirecti...\n\n\nZhe Kong (Sun Yat-sen University, Meituan); Le Li (Tianjin University); Yong Zhang and Feng Gao (Meituan); Shaoshu Yan g (School of Artificial Intelligence, University of Chinese Academy of Sci ences); Tao Wang (Nanjing University); Kaihao Zhang (Harbin Institute of T echnology); Zhuoliang Kang and Xiaoming Wei (Meituan); Guanying Chen (Sun Yat-sen University); and Wenhan Luo (The Hong Kong University of Science a nd Technology)\n---------------------\nNeurCross: A Neural Approach to Com puting Cross Fields for Quad Mesh Generation\n\nWe propose NeurCross, a se lf-supervised framework for quadrilateral mesh generation that jointly opt imizes principal curvature direction field and cross field by employing an optimizable neural SDF to approximate the input surface. NeurCross outper forms state-of-the-art methods in terms of singular ...\n\n\nQiujie Dong ( Shandong University, The University of Hong Kong); Huibiao Wen (Shandong U niversity); Rui Xu (The University of Hong Kong); Shuangmin Chen (Qingdao University of Science and Technology); Jiaran Zhou (Ocean University of Ch ina); Shiqing Xin and Changhe Tu (Shandong University); Taku Komura (The U niversity of Hong Kong); and Wenping Wang (Texas A&M University)\n-------- -------------\nControllable Tracking-Based Video Frame Interpolation\n\nWe present a tracking-based video frame interpolation method, optionally gui ded by user inputs. It utilizes sparse point tracks, first estimated using existing point tracking methods and then optionally refined by the user. Without any user input, it already achieves state-of-the-art results, with f...\n\n\nKarlis Martins Briedis (DisneyResearch|Studios, ETH Zürich); Ab delaziz Djelouah and Raphaël Ortiz (DisneyResearch|Studios); Markus Gross (DisneyResearch|Studios, ETH Zürich); and Christopher Schroers (DisneyRese arch|Studios)\n---------------------\nWhen Gaussian Meets Surfel: Ultra-fa st High-fidelity Radiance Field Rendering\n\nWe introduce Gaussian-enhance d Surfels (GESs), a bi-scale representation combining opaque surfels and G aussians for high-fidelity radiance field rendering. GES is entirely sorti ng free, enabling high-fidelity view-consistent rendering with ultra fast speeds.\n\n\nKeyang Ye, Tianjia Shao, and Kun Zhou (Zhejiang University)\n ---------------------\nDynamic Mesh Processing on the GPU\n\nIntroducing t he first GPU-based system for dynamic triangle mesh processing, delivering order-of-magnitude speedups over CPU solutions across diverse application s. Our system uses patch-based data structure, speculative conflict handli ng, and a novel programming model, enabling robust, high-performa...\n\n\n Ahmed H. Mahmoud (Computer Science and Artificial Intelligence Laboratory (CSAIL), Massachusetts Institute of Technology (MIT)) and Serban D. Porumb escu and John D. Owens (University of California, Davis)\n---------------- -----\nForceGrip: Reference-Free Curriculum Learning for Realistic Grip Fo rce Control in VR Hand Manipulation\n\nForceGrip is a reference-free reinf orcement learning-based agent for realistic VR hand manipulation. It uses a progressive curriculum (finger positioning, intention adaptation, dynami c stabilization) and physics simulation to convert VR controller inputs in to faithful grip forces. In user studies, p...\n\n\nDongHeun Han (Kyung He e University), Byungmin Kim (Korea University), RoUn Lee (Kyung Hee Univer sity), KyeongMin Kim (Korea University), Hyoseok Hwang (Kyung Hee Universi ty), and HyeongYeop Kang (Korea University)\n---------------------\nAccele rated Gamut Discovery via Massive Parallelization\n\nThis paper proposes a scalable framework using Bayesian Neural Networks and a novel 2mD acquisi tion function to efficiently discover gamut boundaries in performance spac e. Combining NSGA-II's diversity and Bayesian Optimization's efficiency, t he method enables large-batch, parallel optimization, out...\n\n\nNavid An sari, Hans-Peter Seidel, and Vahid Babaei (Max Planck Institute for Inform atics)\n---------------------\nMotion-example-controlled Co-speech Gesture Generation Leveraging Large Language Models\n\nWe present a framework to utilize Large Language Models (LLMs) for co-speech gesture generation with motion examples as direct conditions. It enables multi-modal controls ove r co-speech gesture generation, such as motion clips, a single pose, human video, or even text prompts.\n\n\nBohong Chen (State Key Laboratory of CA D & CG, Zhejiang University) and Yumeng Li, Youyi Zheng, Yao-Xiang Ding, a nd Kun Zhou (State Key Lab of CAD and CG, Zhejiang University)\n---------- -----------\nRNA: Relightable Neural Assets\n\nWe propose a neural represe ntation for 3D assets with complex shading. We precompute shading and scat tering on ground-truth geometry, enabling high-fidelity rendering with ful l relightability, eliminating complex shading models and multiple scatteri ng paths, offering significant speed-ups and seamle...\n\n\nKrishna Mullia , Fujun Luan, Xin Sun, and Miloš Hašan (Adobe Research) and Krishna Mullia \n---------------------\nHOIGaze: Gaze Estimation During Hand-Object Inter actions in Extended Reality Exploiting Eye-Hand-Head Coordination\n\nWe pr esent HOIGaze – a novel approach for gaze estimation during hand-object in teractions in extended reality. HOIGaze features: 1) a novel hierarchical framework that first recognises attended hand and then estimates gaze; 2) a new gaze estimator combining CNN, GCN, and cross-modal Transforme...\n\n \nZhiming Hu (University of Stuttgart, The Hong Kong University of Science and Technology (Guangzhou)); Daniel Haeufle (University of Tuebingen, The Center for Bionic Intelligence Tuebingen Stuttgart); Syn Schmitt (Univers ity of Stuttgart, The Center for Bionic Intelligence Tuebingen Stuttgart); and Andreas Bulling (University of Stuttgart)\n---------------------\nPow er-Linear Polar Directional Fields\n\nWe present a method for designing sm ooth directional fields on triangle meshes with precise control over singu larities. Our approach uses a power-linear polar representation, allowing singularities of any index to be placed anywhere on the mesh. The resultin g fields are smooth, robust to mesh qualit...\n\n\nJiabao Brad Wang and Am ir Vaxman (University of Edinburgh)\n---------------------\nWishGI: Lightw eight Static Global Illumination Baking via Spherical Harmonics Fitting\n\ nOur work is a lightweight static global illumination baking solution that achieves competitive lighting effects while using only approximately 5% o f the memory required by mainstream industry techniques. By adopting a ver tex-probe structure, we ensure excellent runtime performance, making it su itabl...\n\n\nJunke Zhu (University of Science and Technology of China, Te ncent Technology); Zehan Wu (Tencent Technology); Qixing Zhang (University of Science and Technology of China); Cheng Liao (Tencent Technology); and Zhangjin Huang (University of Science and Technology of China)\n--------- ------------\nA Divide-and-Conquer Approach for Global Orientation of Non- Watertight Scene-Level Point Clouds with 0-1 Integer Optimization\n\nWe pr opose a divide-and-conquer approach for orienting large-scale, non-waterti ght point clouds. The scene is first segmented into blocks, and normal ori entations are estimated independently within each block. These local orien tations are then globally unified through a graph-based formulation, solv. ..\n\n\nZhuodong Li, Fei Hou, and Wencheng Wang (Institute of Software, Ch inese Academy of Sciences; University of Chinese Academy of Sciences); Xuq uan Lu (The University of Western Australia); and Ying He (Nanyang Technol ogical University)\n---------------------\nVariational Green and Biharmoni c Coordinates for 2D Polynomial Cages\n\nWe present analytical formulas fo r evaluating Green and biharmonic 2D coordinates and their gradients and H essians, for 2D cages made of polynomial arcs.\nWe present results of 2D i mage deformations by direct interaction with the cage and through variatio nal solvers.\nWe demonstrate the flexibility\n\n\nElie Michel, Alec Jacobs on, Siddhartha Chaudhuri, and Jean-Marc Thiery (Adobe Research)\n--------- ------------\nUnbiased Differential Visibility Using Fixed-Step Walk-on-Sp herical-Caps And Closest Silhouettes\n\nWarped-area reparameterization is a powerful technique to compute differential visibility. The key is constr ucting a velocity field that is continuous in the domain interior and agre es with defined velocities on boundaries. We present a robust and efficien t unbiased estimator for differential visibi...\n\n\nLifan Wu, Nathan Morr ical, Sai Praveen Bangaru, Rohan Sawhney, Shuang Zhao, Chris Wyman, Ravi R amamoorthi, and Aaron Lefohn (NVIDIA)\n---------------------\nIP-Composer: Semantic Composition of Visual Concepts\n\nIP-Composer is a novel, traini ng-free method for compositional image generation from multiple reference images. Extending IP-Adapter, it uses natural language to identify concept -specific subspaces in CLIP, projects input images into these subspaces to extract targeted concepts, and fuses them into ...\n\n\nSara Dorfman and Dana Cohen-Bar (Tel Aviv University), Rinon Gal (NVIDIA), and Daniel Cohen -Or (Tel Aviv University)\n---------------------\nMyTimeMachine: Personali zed Facial Age Transformation\n\nWe personalize a pre-trained global aging prior using 50 personal selfies, allowing age regression (de-aging) and a ge progression (aging) with high fidelity and identity preservation.\n\n\n Luchao Qi (University of North Carolina at Chapel Hill (UNC)), Jiaye Wu (U niversity of Maryland College Park), Bang Gong (University of North Caroli na Chapel Hill), Annie Wang (University of North Carolina at Chapel Hill ( UNC)), David Jacobs (University of Maryland College Park), and Roni Sengup ta (University of North Carolina at Chapel Hill (UNC))\n------------------ ---\nMonte Carlo PDE simulation in participating media\n\nWe solve partial differential equations in domains involving complex microparticle geometr y that is impractical, or intractable, to model explicitly. Drawing inspir ation from volume rendering, we treat the domain as a participating medium with stochastic microparticle geometry and develop a volumetr...\n\n\nBai ley Miller (Carnegie Mellon University), Rohan Sawhney (NVIDIA), and Keena n Crane and Ioannis Gkioulekas (Carnegie Mellon University)\n------------- --------\nImproving Global Motion Estimation in Sparse IMU-based Motion Ca pture with Physics\n\nWe propose a physics-driven approach to IMU-based mo tion capture, improving global motion estimation with 3D contact modeling and gravity awareness. Our method estimates world-aligned 3D motion, conta ct points, contact forces, joint torques, and proxy surface interactions u sing only six IMUs in real...\n\n\nXinyu Yi, Shaohua Pan, and Feng Xu (Tsi nghua University)\n---------------------\nRELATE3D: REfocusing Latent Adap ter for Targeted local Enhancement and Editing in 3D Generation\n\nThe ali gnment of text,images,and 3D is very challenging,yet it is crucial and ben eficial for many tasks.We explore and reveal the characteristics of the na tive 3D latent space for 3D generation,make it decomposable and low-rank,t hereby enabling efficient learning for multimodal local alignment,achie... \n\n\nXiao-Lei Li (Tsinghua University, Tencent Video AI Center); Hao-Xian g Chen (Tsinghua University); Yanni Zhang (Tencent Video AI Center); Kai M a (Tencent PCG); Alan Zhao (Tencent Video AI Center); Tai-Jiang Mu (Tsingh ua University); Haoxiang Guo (Skywork AI, Kunlun Inc.); and Ran Zhang (Ten cent Video AI Center)\n---------------------\nLightning-fast Boundary Elem ent Method\n\nWe introduce an inverse-LU preconditioner to solve for the t ypical asymmetric and dense matrices generated by boundary element methods (BEM). The computational efficiency and low memory requirements of our ap proach conspire to scale up to millions of degrees of freedom, with orders of magnitude spee...\n\n\nJiong Chen (INRIA Saclay); Florian Schäfer (Geo rgia Institute of Technology); and Mathieu DESBRUN (INRIA Saclay, Ecole Po lytechnique)\n---------------------\nDeFillet: Detection and Removal of Fi llet Regions in Polygonal CAD Models\n\nDeFillet, the reverse of CAD fille ting, is vital for CAE and redesign but challenging with polygon CAD model s. Our algorithm uses Voronoi vertices as rolling-ball center candidates t o efficiently identify fillets. Sharp features are then reconstructed via quadratic optimization, validated on diverse...\n\n\nJing-En Jiang (School of Computer Science and Technology, Shandong University); Hanxiao Wang (I nstitute of Automation, Chinese Academy of Sciences, Beijing, China Th e School of Artificial Intelligence, University of Chinese Academy of Scie nces); Mingyang Zhao (Academy of Mathematics and Systems Science, Chinese Academy of Sciences, the University of Chinese Academy of Sciences); Dong- Ming Yan (Institute of Automation, Chinese Academy of Sciences); Shuangmin Chen (School of Information and Technology, Qingdao University of Science and Technology); Shiqing Xin (School of Computer Science and Technology, Shandong University); Changhe Tu (School of Computer Science and Technolog y, Shandong University); and Wenping Wang (Computer Science & Engineering , Texas A&M University)\n---------------------\nGaussian Compression for Precomputed Indirect Illumination\n\nWe propose a Gaussian fitting compres sion method for light field probes, reducing storage and memory demands in large scenes. Using low-bit adaptive Gaussians and GPU-accelerated decomp ression, our technique replaces traditional PCA-based approaches, achievin g 1:50 compression ratios. Real-time casc...\n\n\nZhi Zhou (Tencent, Unive rsity of Science and Technology of China); Chao Li, Zhenyuan Zhang, Mingco ng Tang, Zibin Li, and Shuhang Luan (Tencent); and Zhangjin Huang (Univers ity of Science and Technology of China)\n---------------------\nNeST: Neur al Stress Tensor Tomography by leveraging 3D Photoelasticity\n\nNeST enabl es non-destructive 3D stress analysis of transparent objects using the pol arization of light. Traditional 2D methods require destructively slicing t he object. Instead, we reconstruct the entire 3D stress field by jointly h andling phase unwrapping and tensor tomography using neural implicit...\n\ n\nAkshat Dave (Massachusetts Institute of Technology Media Lab), Tianyi Z hang (Rice University), Aaron Young and Ramesh Raskar (Massachusetts Insti tute of Technology Media Lab), Wolfgang Heidrich (King Abdullah University of Science and Technology), Ashok Veeraraghavan (Rice University), and Ak shat Dave\n---------------------\nIn Search of Empty Spheres: 3D Apolloniu s Diagrams on GPU\n\nWe introduce a novel construction algorithm of 3D Apo llonius diagrams designed for GPUs. Our method features a fast execution w hile allowing a comprehensive computation. This is made possible thanks to a light data structure, a cell update procedure and a spacial exploration strategy all designed to...\n\n\nCyprien Plateau--Holleville (Université de Limoges, XLIM); Benjamin Stamm (Universität Stuttgart, Institute of App lied Analysis and Numerical Simulation); Vincent Nivoliers (Université Cla ude Bernard Lyon 1, LIRIS); and Maxime Maria and Stéphane Mérillou (Univer sité de Limoges, XLIM)\n---------------------\nTetWeave: Isosurface Extrac tion using On-The-Fly Delaunay Tetrahedral Grids for Gradient-Based Mesh O ptimization\n\nTetWeave is a novel isosurface representation that jointly optimizes a tetrahedral grid and directional distances for gradient-based mesh processing like multi-view 3D reconstruction. It dynamically builds a daptive grids via Delaunay triangulation, ensuring watertight, manifold me shes. By resampling...\n\n\nAlexandre Binninger and Ruben Wiersma (ETH Zur ich), Philipp Herholz (Independent Contributor), and Olga Sorkine-Hornung (ETH Zurich)\n---------------------\nField Smoothness-Controlled Partition for Quadrangulation\n\nOur approach proposes a novel partition method for reliable feature-aligned quadrangulation. The core insight is that singul arity-distant smooth streamlines are more suitable as patch boundaries. Th e key implementation confines patch boundaries to high field smoothness re gions.\nValidated on large-sc...\n\n\nZhongxuan Liang, Wei Du, and Xiao-Mi ng Fu (University of Science and Technology of China)\n------------------- --\nDesigning 3D Anisotropic Frame Fields with Odeco Tensors\n\nOur method proposes a novel computational design framework for designing anisotropic tensor fields. It enables flexible control over scalings without requirin g users to specify orientations explicitly. We apply these anisotropic ten sor fields to various applications, such as anisotropic meshing, str...\n\ n\nHaikuan Zhu and Hongbo Li (Wayne State University); Hsueh-Ti Derek Liu (Roblox, University of British Columbia); Wenping Wang (Texas A&M Universi ty); and Jing Hua and Zichun Zhong (Wayne State University)\n------------- --------\nHistogram Stratification for Spatio-Temporal Reservoir Sampling\ n\nThis paper introduces stratification into resampled importance sampling (RIS) technique for real-time photorealistic rendering. It organizes samp le candidates into local histograms and then employs Quasi Monte Carlo and antithetic patterns for efficient sampling. This low-overhead approach si gnifica...\n\n\nCorentin Salaun and Martin Balint (Max Planck Institute fo r Informatics), Laurent Belcour and Eric Heitz (Intel), and Gurprit Singh and Karol Myszkowski (Max Planck Institute for Informatics)\n------------- --------\nPolicy-Space Diffusion for Physics-Based Character Animation\n\n We present a new perspective on physics-based character animation. Assumin g policies for similar motions should have similar weights, we introduce r egularization during RL training to preserve weight similarity. By modelin g the weights’ manifold with a diffusion model, we generate a continuum .. .\n\n\nMichele Rocca, Sune Darkner, and Kenny Erleben (University of Copen hagen); Sheldon Andrews (École de Technologie Supérieure (ÉTS)); and Miche le Rocca\n---------------------\nSingle View Garment Reconstruction Using Diffusion Mapping Via Pattern Coordinates\n\nWe introduce a novel method f or accurate 3D garment reconstruction from single-view images, bridging 2D and 3D representations. Our mapping model creates connections among image pixels, UV coordinates, and 3D geometry, resulting in realistic garments with intricate details and enabling downstream ap...\n\n\nRen Li (EPFL), C ong Cao (MBZUAI), Corentin Dumery and Yingxuan You (EPFL), Hao Li (MBZUAI) , and Pascal Fua (EPFL)\n---------------------\nSynchronized tracing of pr imitive-based implicit volumes\n\nThe paper presents a tile-based renderin g pipeline for modeling with implicit volumes, using blobtrees and smooth CSG operators. It requires no preprocessing when updating primitives and e nsures efficient ray processing with sphere tracing. The method uses a low -resolution A-buffer and bottom-up tre...\n\n\nCédric Zanni (Université de Lorraine CNRS, Inria, LORIA) and Cédric Zanni\n---------------------\nEch oes of the Coliseum: Towards 3D Live streaming of Sports Events\n\nWe pres ent a revolutionary method for experiencing live sports in stunning 3D, re defining the way games are seen, through immersive, interactive replays. A longside, we release a large-scale synthetic dataset built to benchmark re alism, motion, and human interaction in dynamic scenes, to fuel the nex... \n\n\nJunkai Huang, Saswat Subhajyoti Mallick, Alejandro Amat, Marc Ruiz O lle, Albert Mosella-Montoro, Bernhard Kerbl, Francisco Vicente Carrasco, a nd Fernando De la Torre (Carnegie Mellon University)\n-------------------- -\nAugmented Vertex Block Descent\n\nWe extend the Vertex Block Descent me thod for fast and unconditionally stable physics-based simulation using an Augmented Lagrangian formulation to enable simulating hard constraints wi th infinite stiffness and systems with high stiffness ratios. This allows simulating complex contact scenarios invo...\n\n\nChris Giles (Roblox) and Elie Diaz and Cem Yuksel (University of Utah)\n---------------------\nStr uctRe: Rewriting for Structured Shape Modeling\n\nThe paper presents Struc tRe, a structure rewriting system for 3D shape modeling. It uses an iterat ive process to rewrite objects, either upwards to more concise structures or downwards to more detailed ones, generating hierarchies. This localized rewriting approach enables probabilistic modeling of ...\n\n\nJiepeng Wan g (The University of Hong Kong, Microsoft Research Asia); Hao Pan (Microso ft Research Asia, Tsinghua University); Yang Liu and Xin Tong (Microsoft R esearch Asia); Taku Komura (The University of Hong Kong); Wenping Wang (Te xas A&M University); and Jiepeng Wang\n---------------------\nFacial Appea rance Capture at Home with Patch-Level Reflectance Prior\n\nGiven a single co-located smartphone video captured in a dim room as the input, our meth od can reconstruct high-quality facial assets within the distribution mode led by a diffusion prior trained on Light Stage scans, which can be export ed to common graphics engines like Blender for photo-realistic r...\n\n\nY uxuan Han and Junfeng Lyu (Tsinghua University); Kuan Sheng (ShanghaiTech University; Deemos Technology Co., Ltd.); Minghao Que (Tsinghua University ); Qixuan Zhang (ShanghaiTech University; Deemos Technology Co., Ltd.); La n Xu (ShanghaiTech University); and Feng Xu (Tsinghua University)\n------- --------------\nHoloChrome: Polychromatic Illumination for Speckle Reducti on in Holographic Near-Eye Displays\n\nHoloChrome introduces a novel holog raphic display method by multiplexing multiple wavelengths and two spatial light modulators to enhance image quality. By moving beyond standard thre e-color primary systems, it significantly reduces speckle noise while pres erving natural depth cues while achieving m...\n\n\nFlorian Schiffers (Ama zon Prime Video, Northwestern University); Grace Kuo, Nathan Matsuda, Doug las Lanman, and Oliver Cossairt (Meta Reality Labs); and Florian Schiffers and Oliver Cossairt\n---------------------\nMoVer: Motion Verification fo r Motion Graphics Animations\n\nLarge vision-language models often fail to capture spatio-temporal details in text-to-animation tasks. We introduce MoVer, a verification system using first-order logic to check properties l ike timing and positioning in motion graphics animations. Integrated into an LLM pipeline, MoVer enables itera...\n\n\nJiaju Ma and Maneesh Agrawala (Stanford University)\n---------------------\nA Fully-statistical Wave Sc attering Model for Heterogeneous Surfaces\n\nThis work presents a statisti cal wave-scattering model for surfaces with nanoscale mixtures in geometry and material. It predicts average appearance (BRDF) and draws realistic s peckles directly from surface statistics, without explicit definitions. Th e proposed model demonstrates various application...\n\n\nZhengze Liu and Yuchi Huo (State Key Lab of CAD & CG, Zhejiang University); Yifan Peng (Un iversity of Hong Kong); and Rui Wang (State Key Lab of CAD & CG, Zhejiang University)\n---------------------\nInstant Self-Intersection Repair for 3 D Meshes\n\nWe present a novel framework that instantly (< 1 sec) repairs self-intersections in static surface meshes, which commonly occur during t he 3D modeling process.\n\n\nWonjong Jang, Yucheol Jung, Gyeongmin Lee, an d Seungyong Lee (POSTECH)\n---------------------\nA Fluorescent Material M odel for Non-Spectral Editing & Rendering\n\nWe introduce a material model for diffuse fluorescence that is compatible with RGB and spectral renderi ng. This models builds on an analytical integrable Gaussian-based model of the spectral reradiation that is efficient enough to permits real-time re ndering and editing of such appearance.\n\n\nLaurent Belcour and Alban Fic het (Intel Labs) and Pascal Barla (Inria - LaBRI)\n---------------------\n SpotLessSplats: Ignoring Distractors in 3D Gaussian Splatting\n\n3D Gaussi an Splatting (3DGS) enables fast 3D reconstruction and rendering but strug gles with real-world captures due to transient elements and lighting chang es. We introduce SpotLessSplats, which leverages semantic features from fo undation models and robust optimization to remove transient effects, ...\n \n\nSara Sabour (Google Inc - Deepmind, University of Toronto); Lily Goli (University of Toronto); George Kopanas (Runway); Mark Matthews and Dmitry Lagun (Google Inc - Deepmind); Leonidas Guibas (Google Inc - Deepmind, St anford University); Alec Jacobson (University of Toronto); David Fleet (Go ogle Inc - Deepmind, University of Toronto); Andrea Tagliasacchi (Google I nc - Deepmind, Simon Fraser University); and Sara Sabour and Lily Goli\n-- -------------------\nGuiding-Based Importance Sampling for Walk on Stars\n \nWalk on stars (WoSt) has shown its power in being applied to Monte Carlo methods for solving PDEs but the sampling techniques in WoSt are not sati sfactory, leading to high variance. Inspired by Monte Carlo rendering, we propose a guiding-based importance sampling method to reduce the variance of WoS...\n\n\nTianyu Huang and Jingwang Ling (School of Software and BNRi st, Tsinghua University); Shuang Zhao (University of California Irvine); a nd Feng Xu (School of Software and BNRist, Tsinghua University)\n--------- ------------\nLayerPano3D: Layered 3D Panorama for Hyper-Immersive Scene G eneration\n\nLayerPano3D is a novel framework that generates hyper-immersi ve 3D panoramic scenes from a single text prompt. By decomposing panoramas into multiple layers and optimizing them as 3D Gaussians, it enables full 360°×180° exploration with consistent visual quality, unlocking new possi bilities for virt...\n\n\nShuai Yang (Shanghai Jiao Tong University, Shang hai Artificial Intelligence Laboratory); Jing Tan (The Chinese University of Hong Kong); Mengchen Zhang (Zhejiang University, Shanghai Artificial In telligence Laboratory); Tong Wu (The Chinese University of Hong Kong); Gor don Wetzstein (Stanford University); Ziwei Liu (Nanyang Technological Univ ersity); and Dahua Lin (The Chinese University of Hong Kong)\n------------ ---------\nJames-Stein Gradient Combiner for Inverse Monte Carlo Rendering \n\nThis paper introduces a gradient combiner that blends unbiased and bia sed gradients in parameter space using the James-Stein estimator to infer scene parameters (BSDFs and volumes) from images. This approach enhances o ptimization accuracy compared to relying solely on either unbiased or bias ed gradi...\n\n\nJeongmin Gu and Bochang Moon (Gwangju Institute of Scienc e and Technology)\n---------------------\nViSA: Physics-based Virtual Stun t Actors for Ballistic Stunts\n\nWe introduce ViSA (Virtual Stunt Actors), an interactive animation system using deep reinforcement learning to gene rate realistic ballistic stunt actions. It efficiently produces dynamic sc enes commonly seen in films and TV dramas, such as traffic accidents and s tairway falls. A novel action space d...\n\n\nMinseok Kim and Wonjeong Seo (Seoul National University), Sung-Hee Lee (Korea Advanced Institute of Sc ience and Technology (KAIST)), and Jungdam Won (Seoul National University) \n---------------------\nGarmentImage: Raster Encoding of Garment Sewing P atterns with Diverse Topologies\n\nGarment sewing patterns often rely on v ector formats, which struggle with discontinuities and unseen topologies. GarmentImage instead encodes geometry, topology, and placement into multi- channel grids, enabling smooth transitions and better generalization. Usin g simple CNNs, it works well in pattern...\n\n\nYuki Tatsukawa (The Univer sity of Tokyo); Anran Qi (INRIA, Université Côte d'Azur; The University of Tokyo); and I-Chao Shen and Takeo Igarashi (The University of Tokyo)\n\nI nterest Area: Arts & Design, Gaming & Interactive, New Technologies, Produ ction & Animation, Research & Education\n\nRecording: Livestreamed, Not Li vestreamed, Recorded, Not Recorded\n\nKeyword: Animation, Art, Artificial Intelligence/Machine Learning, Audio, Augmented Reality, Capture/Scanning, Computer Vision, Digital Twins, Display, Dynamics, Education, Ethics and Society, Fabrication, Games, Generative AI, Geometry, Image Processing, Mo deling, Performance, Physical AI, Real-Time, Robotics, Scientific Visualiz ation, Simulation, Spatial Computing, Virtual Reality\n\nRegistration Cate gory: Full Conference, Virtual Access, Experience, Sunday END:VEVENT END:VCALENDAR