Wiley: Computer Graphics Forum: Table of Contents

Dress Anyone: A Method and Dataset for 3D Garment Retargeting

Mon, 08 Jun 2026 03:35:11 -0700

Retargeting of real, non-parametric garments over any arbitrary target body while being agnostic to pose and shape.

Abstract

Retargeting 3D garment meshes over digital characters and avatars involves non-rigid deformation of garments to plausibly fit the target body in arbitrary poses. Existing learning-based methods for garment retargeting require the garments to be canonicalized and can only be retargeted onto parametric human body models. In this work, we present a novel framework for retargeting garments in arbitrary poses to any given human mesh. We adopt a robust Isomap-based representation to first estimate correspondences between garment and body mesh to achieve an initial coarse retargeting. We further adapt a fast and efficient neural optimization step, governed by Physics-based constraints to obtain realistic draping of the garment. We show generalization to biped cartoon characters and non-parametric human meshes in arbitrary poses. We perform extensive experiments on publicly available datasets and our proposed dataset of 3D clothing, demonstrating the effectiveness of our method.

CanvasDream: Sketch‐Guided Controllable Text‐to‐3D Generation

C. Zheng, D. Huang, Y. Liu, Y. Shi — Thu, 04 Jun 2026 06:32:42 -0700

We introduce CanvasDream, a sketch-guided text-to-3D generation approach capable of producing high-fidelity 3D assets with realistic textures and rich details that faithfully preserve both textual descriptions and sketch compositions. We design a novel two-stage 3D generation framework that disentangles geometry and appearance learning.

Abstract

Recent advancements in text-to-3D generation techniques have successfully produced high-fidelity 3D content. However, existing approaches lack fine-grained control over the results, making it challenging to produce controllable 3D content. Hand-drawn sketches are brief and intuitive, enabling interactive 3D control. However, their ambiguity complicates text-to-3D pipelines. We introduce CanvasDream, a sketch-guided text-to-3D generation approach capable of producing high-fidelity 3D assets with realistic textures and rich details that faithfully preserve both textual descriptions and sketch compositions. We design a novel two-stage 3D generation framework that disentangles geometry and appearance learning. For geometry, we leverage 3D Gaussian Splatting to generate object shape under the guidance of our proposed sketch-guided multi-view diffusion model. For appearance, we propose Physically-Based Rendering (PBR), enabling the resulting assets to be directly used in downstream applications. Extensive experiments demonstrate the effectiveness and superior performance of our method in 3D generation compared to state-of-the-art approaches. Moreover, user studies further highlight the high controllability of our approach in 3D generation, affirming its practical value.

HamCat: Ego‐Centric Relationship Exploration for Multidimensional Categorical Data

H. Balaka, H. Hauser, L. A. Garrison — Thu, 04 Jun 2026 01:49:27 -0700

Abstract

We introduce HamCat, a novel visualization method for exploring and analyzing multidimensional categorical survey data. Typical visualization approaches for multidimensional categorical data do not support simultaneous analysis of attributes and items, nor do they allow for in-depth similarity analysis of an entire dataset from the perspective of a specific reference point. HamCat, in contrast, aims to facilitate detailed analysis of multidimensional categorical data across both attributes and items. Our approach builds on the concept of a Hammingball combined with a force-directed layout to support ego-centric, user-steered analysis of inter-item and inter-attribute relationships in multidimensional categorical survey data. In addition, our method supports the inclusion and nuanced visualization of missingness. We illustrate the value of HamCat through two case studies. The first case focuses on a survey on wellbeing collected by the European Social Survey, while the second is an expert-driven study for a survey on sense of belonging in computer science higher education. These case studies show how HamCat complements existing analysis workflows to reveal relationships and item groupings across attributes that are not easily discoverable through conventional means. Supplementary materials for our method are available at https://osf.io/uz2jv/.

Uncertainty‐Aware Visual Analysis of Force Networks in 2D Granular Materials

Thu, 04 Jun 2026 01:48:30 -0700

Abstract

Uncertainty in experimental measurements makes it challenging to determine which features are intrinsic to the phenomenon and which are most stable and reliable. Granular materials, such as sand, form a complex system in which the forces between individual particles influence the material's macroscopic behavior. However, these forces are also subject to uncertainty, as repeated measurements can yield different results. In this paper, we investigate how to model and visually analyze the uncertain structure of forces in granular materials. We adopt different perspectives on uncertainty by considering it as variance, probability, or additional variable. For a nuanced analysis of granular material data, we propose combining visualizations that represent these perspectives into a single visual analytics approach. We integrate uncertainty-aware spatial visualizations that convey the probability of features, visualizations of derived measures and their variance over changes in the packing fraction, and overviews of varying probability thresholds. Finally, we evaluate our proposed approach in a case study conducted together with geotechnical engineers for the example of a 2D ensemble of photoelastic disks.

When the Chain Breaks: Interactive Diagnosis of LLM Chain‐of‐Thought Reasoning Errors

Thu, 04 Jun 2026 01:46:15 -0700

Abstract

Current Large Language Models (LLMs), especially Large Reasoning Models, can generate Chain-of-Thought (CoT) reasoning traces to illustrate how they produce final outputs, thereby facilitating trust calibration for users. However, these CoT reasoning traces are usually lengthy and tedious, and can contain various issues, such as logical and factual errors, which make it difficult for users to interpret the reasoning traces efficiently and accurately. To address these challenges, we develop an error detection pipeline that combines external fact-checking with symbolic formal logical validation to identify errors at the step level. Building on this pipeline, we propose ReasonDiag, an interactive visualization system for diagnosing CoT reasoning traces. ReasonDiag provides 1) an integrated arc diagram to show reasoning-step distributions and error-propagation patterns, and 2) a hierarchical node-link diagram to visualize high-level reasoning flows and premise dependencies. We evaluate Reason-Diag through a technical evaluation for the error detection pipeline, two case studies, and user interviews with 16 participants. The results indicate that ReasonDiag helps users effectively understand CoT reasoning traces, identify erroneous steps, and determine their root causes.

Designing Annotations in Visualization: Considerations from Visualization Practitioners and Educators

Md Dilshadur Rahman, Devin Lange, Ghulam Jilani Quadri, Paul Rosen — Thu, 04 Jun 2026 01:44:41 -0700

Abstract

Annotation is a central mechanism in visualization design that enables people to communicate key insights. Prior research has provided essential accounts of the visual forms annotations take, but less attention has been paid to the decisions behind them. This paper examines how annotations are designed in practice and how educators reflect on those practices. We conducted a two-phase qualitative study: interviews with ten practitioners from diverse backgrounds revealed the heuristics they draw on when creating annotations, and interviews with seven visualization educators offered complementary perspectives situated within broader concerns of clarity, guidance, and viewer agency. These studies provide a systematic account of annotation design knowledge in professional settings, highlighting the considerations, trade-offs, and contextual judgments that shape the use of annotations. By making this tacit expertise explicit, our work complements prior form-focused studies, strengthens understanding of annotation as a design activity, and points to opportunities for improved tool and guideline support.

DenseSMPLify: 3D Human Body Parametric Reconstruction Using Pixel Aligned Dense Normal Maps

Zhaobo Zhang, Heming Li, Yongwei Nie, Chuhua Xian, Guiqing Li — Thu, 04 Jun 2026 00:22:31 -0700

We present DenseSMPLify, a novel 3D human mesh recovery framework that addresses the fundamental depth ambiguity in monocular image-based reconstruction by leveraging dense pseudo-normal maps from Sapiens as rich geometric priors, significantly outperforming traditional sparse 2D joint-based approaches.

Abstract

Recent advances in 3D human mesh recovery from monocular images primarily rely on optimizing the SMPL model using sparse 2D cues such as 2D key points or silhouettes. However, these constraints often lead to incorrect 3D pose estimations due to inherent depth ambiguity. To address this fundamental limitation, we propose DenseSMPLify, a novel optimization framework that leverages dense pseudo-normal maps estimated by Sapiens as complementary geometric guidance. Our key insight is that normal maps encode richer 3D shape and depth cues than sparse 2D joints, significantly reducing ambiguity during optimization. A critical challenge lies in the misalignment between SMPL-derived normals and pseudo-normal maps. We address this via an alignment algorithm based on Continuous Surface Embedding (CSE), which establishes pixel-level correspondence between the two normal maps before error computation. Experiments demonstrate that DenseSMPLify outperforms SMPLify and other optimization baselines in both pose accuracy and shape realism. Notably, our method consistently improves upon initial poses, even when initialized with state-of-the-art regression models.

Individualized Pathfinding in Rugged Open Terrains Using Semantic Navigation Meshes

C. Creus, N. Pelechano, O. Argudo — Wed, 03 Jun 2026 00:00:00 -0700

We introduce semantic navigation meshes for complex outdoor terrains, partitioning space into regions with coherent semantic properties like slope or vegetation density. The resulting graphs enable efficient pathfinding with agent-specific, parameterized cost functions without precomputation, accelerating search while maintaining path quality close to optimal solutions.

Abstract

Traditional navigation meshes split the terrain into convex walkable regions, which work well for nearly flat environments with clearly defined obstacles. However, in complex outdoor terrains with variable slopes and heights, these approaches fail to capture areas with consistent characteristics. Moreover, human preferences vary per individual: some may favour smooth paths with gentle slopes, while others prefer shorter routes even if steeper. We present a semantic navigation mesh that generates cells with coherent, low-variance values for chosen semantics, such as slope or vegetation density. Cells are constructed via a region-growing partitioning algorithm, optionally refined for compactness, and can incorporate multiple semantic dimensions simultaneously. This approach preserves terrain complexity without relying on a binary walkable/non-walkable classification. From this partition, we propose different alternatives for generating a navigation graph and perform semantically-aware pathfinding for agent-specific cost functions, which prevents the use of precomputed hierarchical paths. Our evaluation demonstrates that semantic meshes accelerate pathfinding on complex open terrains while producing trajectories close to those obtained by a brute-force search over the entire domain. This provides a flexible, generic framework that balances graph size, subdomain homogeneity, search efficiency, and path quality.

UrbanClipAtlas: A Visual Analytics Framework for Event and Scene Retrieval in Urban Videos

Tue, 02 Jun 2026 01:47:52 -0700

Abstract

Extracting actionable insights from long-duration urban videos is often labor-intensive: analysts must manually sift through raw footage to pinpoint target events or uncover broader behavioral trends. In this work, we present UrbanClipAtlas , a visual analytics system for exploring long urban videos recorded at street intersections. UrbanClipAtlas combines retrieval-augmented generation (RAG), taxonomy-aware entity extraction, and video grounding to support event retrieval and interpretation. The system segments extended recordings into short clips, generates textual descriptions with a vision–language model, and indexes them for semantic retrieval. A knowledge graph maps entities and relations from LLM answers onto a domain-specific taxonomy and aligns them with detected objects and trajectories to support visual grounding and verification. UrbanClipAtlas supports scene retrieval through an augmented chat-based interface and improves scene interpretation by tightly aligning textual outputs with video evidence. This design strengthens the connection between textual reasoning and visual evidence, reducing the effort required to validate model outputs and refine hypotheses. We demonstrate the usefulness of UrbanClipAtlas on the StreetAware dataset through two case studies involving hazardous scenarios and crossing dynamics at street intersections. UrbanClipAtlas helps analysts reason about safety- and mobility-related patterns across large urban video collections.

Engagement vs. Understanding: Comparing Immersive Virtual Reality and Desktop Displays for Climate Data Visualization

A. Arunkumar, J. Liu, C. Bryan — Tue, 02 Jun 2026 01:47:16 -0700

Abstract

Immersive virtual reality (IVR) is increasingly used for scientific data visualization, with the expectation that greater immersion will enhance both engagement and understanding. However, prior research suggests a potential trade-off: IVR can heighten affective responses while impairing cognitive learning due to increased cognitive load. To examine this tension in abstract scientific visualization, we conducted a between-subjects study (N=84) comparing head-mounted display (HMD) VR with traditional desktop displays as participants explored three climate datasets on an interactive 3D globe. We measured cognitive and affective learning outcomes after each exploration trial and at a two-week follow-up. We additionally logged patterns of visualization exploration during trials, and measured user experience post-study and at two-week follow-up. Results reveal systematic divergences: desktop displays led to significantly higher recall accuracy and promoted more systematic exploration patterns, while HMD-based VR produced stronger short-term increases in climate concern and higher satisfaction ratings. Critically, affective changes largely reverted to baseline at follow-up regardless of modality. We further identify five distinct exploration strategies that emerge differentially across modalities and relate directly to learning outcomes. Overall, our findings highlight how modality choice should align with visualization goals, and offer actionable insights into designing effective scientific visualizations that balance cognitive and affective learning objectives. All data and materials are available at: https://osf.io/24w7s/.

Efficient Shape‐Improved Bi‐3 Subdivision Surfaces

Kestutis Karčiauskas, Jörg Peters — Fri, 29 May 2026 03:17:11 -0700

Efficient Shape-Improved Bi-cubic Subdivision shares the minimal 3n patches per ring of Catmull–Clark surfaces, but generates far better shape and uses fewer pieces to cover the same n-sided surface region for faster graphics processes and smaller files. Key technical innovations are a non-uniform refinement and the relaxing of formal smoothness.

Abstract

Efficient Shape-Improved Bi-cubic Subdivision (ESIS) is a new curvature-bounded bi-cubic subdivision scheme with the same minimal patch layout as Catmull–Clark surfaces, that is, 3n$3n$ patches per subdivision surface ring, but generating far better shape and using fewer pieces to cover the same n$n$-sided surface region. Fewer pieces means: faster downstream graphics processes and rendering, smaller files for data exchange, and fewer equations for computing on surfaces and simulation. The key technical innovations are: a non-uniform refinement and the relaxing of formal smoothness of subdivision surface rings while preserving good highlight line distributions.

Single Line Drawing Generation via Semantics‐Driven Optimization

Tanguy Magne, Alexandre Binninger, Ruben Wiersma, Olga Sorkine‐Hornung — Thu, 28 May 2026 05:52:47 -0700

We present a method for automatically generating single-line drawings in vector format, guided by a text prompt or an input image. Our approach leverages score distillation sampling to optimize the parameters of a uniform rational B-spline (URBS) curve, ensuring that the drawing consists of a single continuous stroke by design.

Abstract

Line drawings are a highly expressive art form that requires the artist to abstract and distill the essence of their subject. We present the first semantics-driven method for automatically generating single-line drawings in vector format, guided either by a text prompt describing the concept or an input image depicting it. Our approach leverages score distillation sampling to optimize the parameters of a uniform rational B-spline (URBS) curve, ensuring that the drawing consists of a single continuous stroke by design. This representation provides fine-grained control over the level of detail, while additional loss terms allow us to steer the final artistic style. We demonstrate that our method outperforms state-of-the-art text-to-image models and optimization pipelines for this task, producing results that are both more aesthetically pleasing and more faithful to the style of continuous line drawing artists. Furthermore, because our method generates a vectorized curve, it directly supports downstream fabrication processes such as embroidery, laser engraving and wire bending. Our code and results are available at https://github.com/tanguymagne/SLDgen.

Fast Nodal Hessian Computation for Peridynamic Fracture Simulation

Yuxiong Qin, Qingfu Zhang, Zhongkai Zhang — Thu, 28 May 2026 05:00:20 -0700

A fast, exact nodal Hessian computation for Non-Ordinary State-Based Peridynamics is introduced through analytical simplification and a warp-centric GPU strategy. The method accelerates preconditioned solvers and Vertex Block Descent, enabling interactive fracture simulation with physical accuracy.

Abstract

Simulating dynamic fracture with physical realism and computational efficiency remains a formidable challenge in computer graphics. Non-ordinary state-based peridynamics (NOSB-PD) offers a compelling framework for this task, but its non-local nature makes the computation of second-order derivatives (namely, the nodal Hessian blocks essential for implicit integration and advanced solvers) prohibitively expensive, preventing real-time application. In this paper, we present a comprehensive framework that transforms this computational bottleneck into an efficient, practical operation. Our approach is built on two pillars: first, a novel analytical reformulation that exploits hidden mathematical symmetries and sparsity in NOSB-PD, drastically reducing the complexity of Hessian block computation without approximation; second, a co-designed ‘Warp-Vertices-Parallel’ GPU architecture that effectively manages irregular workloads and eliminates memory conflicts inherent in peridynamics. We demonstrate the versatility of our efficient computational core by seamlessly accelerating two distinct classes of algorithms: it enables a block-diagonal preconditioner that outperforms traditional Jacobi preconditioning, and serves as the engine for a high-performance Vertex Block Descent solver. Extensive experiments show that our method achieves substantial performance gains, enabling stable, interactive simulation of complex elastic and fracture phenomena while maintaining full physical accuracy. Our work establishes fast nodal Hessian computation as a fundamental primitive, bridging the gap between the strong physical capabilities of NOSB-PD and its practical performance in graphics applications.

HydroGaussian: Physically‐Guided Hierarchical Gaussian Splatting for Underwater Scene Reconstruction

Xinyu Shi, Chenrui Wu, Zhenzhong Chu, Yao Huang — Fri, 22 May 2026 03:19:10 -0700

HydroGaussian: Physically-guided Hierarchical Gaussian Splatting for Underwater Scene Reconstruction This paper introduces HydroGaussian, a physically-guided three-dimensional (3D) Gaussian Splatting framework for underwater scene reconstruction. By explicitly computing ray-Gaussian intersections and employing a hierarchical depth regularization, the method effectively decouples surface radiance from scattering media. The approach significantly mitigates underwater visual artifacts, achieving state-of-the-art fidelity in both colour restoration and geometric reconstruction across diverse underwater datasets.

Abstract

Reconstructing underwater three-dimensional (3D) scenes presents a significant yet captivating challenge, with applications spanning from naval robotics to virtual reality. Although 3D Gaussian Splatting (3DGS) demonstrates great potential in underwater 3D scene reconstruction, its capability as an explicit renderer to express underwater lighting models remains insufficient. This paper introduces a hierarchical Gaussian splatting model constrained by a physical-guided underwater lighting model. The proposed method streamlines the 3DGS fitting process for backscattering media through the analysis of the relationship between the underwater rendering equation and the lighting model. The proposed model achieves optimal decoupling between surface colours and backscattering effects by explicitly computing ray-Gaussian intersections to constrain 3DGS positions near depth surfaces. To address the depth estimation challenges in infinite background regions, we introduce a hierarchical structure that effectively regularizes far-field depth values. Experiments on public underwater datasets demonstrate that HydroGaussian successfully mitigates challenging underwater light scattering effects, yielding substantial improvements in both visual fidelity and scene restoration quality compared to state-of-the-art methods. The implementation will be made publicly available at https://github.com/SXYuuuuuu/HydroGaussian.

Deep Residual Combiner: A Learned Fusion of Spatial, Temporal, and Multiscale Correlated Pixel Estimates

W. Zhou, E. Hughes, T. Hachisuka — Wed, 20 May 2026 03:19:37 -0700

Abstract

Correlation-based rendering techniques continue to advance, and efficiently exploiting correlations between pixel estimates has become increasingly important. The deep combiner framework [BHHM20] allows us to fuse independent and correlated pixel estimates but focuses solely on spatial correlations. We propose a generalization of the deep combiner framework, the deep residual combiner, that is designed to exploit correlations across spatial, temporal, and multiscale domains. The deep residual combiner enables robust cross-domain fusion, effectively reducing systematic artifacts and significantly enhancing temporal coherence, both of which are especially important in animation scenarios. We demonstrate the effectiveness of our proposed method through several practical applications, showcasing improvements in temporal stability, visual fidelity, and reduction of residual errors across diverse rendering scenarios compared to prior approaches.

Dynamic Wave Trains: A Procedural Approach to Spatially Varying Ocean Synthesis

Tue, 19 May 2026 06:09:24 -0700

In this paper, we present a high performance procedural model capable of synthesizing and rendering a virtually infinite ocean. Our method exhibits spatial variations as the orientations, speed and choppiness of the waves is parametrized for each frequencies and each point in space. The provided visual abstract exhibits a sunset rendering of a large 40 km * 40 km ocean. A refraction phenomenon is achieved by using physical equations as an input for the model. According to the water depth, waves slow down as they get closer to the shore, as a result, their choppiness and frequency rises, and waves tend to align to the shore. Additionally, the model exposes useful data such as the phase of individual waves, enabling phase-dependent visual effects such as foam or sub-surface scattering.

Abstract

Vast oceans are a critical component of many virtual environments. While procedural methods efficiently generate non-repetitive, large-scale content with low memory footprints, they usually lack the ability to model in real-time local interactions with obstacles such as diffraction, reflection and refraction, or spatial variations such as localized wind ripples. We present a procedural ocean synthesis method that addresses this by enabling local control over wave properties including frequency, orientation, velocity and amplitude. By controlling a sum of independent wave trains, our approach seamlessly integrates spatial variations and wavelength-dependent interactions with the environment. The model allows for high performance parallel evaluation at any scale, enabling the real-time rendering and filtering of height fields, normal maps and rough BRDF, depending on rendering requirements. Additionally, we leverage the phase information encoded in the model to simulate secondary effects such as foam at a low computational cost.

Physically Based Neural BRDF: A Framework for Physically Correct Material Reconstruction, Generation and Editing

Fri, 15 May 2026 03:52:03 -0700

We introduce the physically based neural BRDF, a novel continuous representation for material appearance based on neural fields. Our model accurately performs real-world material reconstruction, generation and editing, while uniquely enhancing the physical accuracy of neural BRDFs via three key aspects: energy conservation, Helmholtz reciprocity and chromaticity enforcement.

Abstract

We introduce the physically based neural bidirectional reflectance distribution function (PBNBRDF), a novel continuous representation for material appearance based on neural fields. Our model accurately performs real-world material reconstruction, generation and editing, while uniquely enforcing physical properties for realistic BRDFs: Helmholtz reciprocity via reparametrisation and energy conservation via efficient analytical integration. We conduct a systematic analysis demonstrating the benefits of adhering to these physical laws on the visual quality of reconstructed materials. Additionally, we enhance the colour accuracy of neural BRDFs by introducing chromaticity enforcement supervising the norms of RGB channels. Through both qualitative and quantitative experiments on multiple databases of measured real-world BRDFs, we show that adhering to these physical constraints enables neural fields to more faithfully and stably represent the original data and achieve higher rendering quality.

Affinification: A Fine Approximation of Deformations

A. Mercier‐Aubin, T. Schneider, P.G. Kry, S. Andrews — Wed, 13 May 2026 01:12:25 -0700

Abstract

We introduce affinification, a novel method for accelerating physics-based animation of elastic solids. During a time-dependent simulation, our method automatically partitions the space into affine and elastic regions depending on the deformation. As such, we capture localized deformations while significantly reducing computational costs with larger regions of model reduction. We design a new clustering method based on deformation rates to capture affinely deforming regions, and explore multiple heuristics for seeding, pattern generation, and the impact of physical parameters on coarsened regions. We compare our method with the ground truth, showing performance increasing with resolution and recorded simulations up to 17× faster compared to elastic simulations, while retaining similar levels of visual fidelity.

TreeON: Reconstructing 3D Tree Point Clouds from Orthophotos and Heightmaps

Wed, 13 May 2026 01:11:16 -0700

Abstract

We present TreeON, a novel neural-based framework for reconstructing detailed 3D tree point clouds from sparse top-down geodata, using only a single orthophoto and its corresponding Digital Surface Model (DSM). Our method introduces a new training supervision strategy that combines both geometric supervision and a differentiable shadow and silhouette loss to learn point cloud representations of trees without requiring species labels, procedural rules, detailed terrestrial reconstruction data, or ground laser scan data. To address the lack of ground truth data, we generate a synthetic dataset of point clouds from procedurally modeled trees and train our network on it. Quantitative and qualitative experiments demonstrate better reconstruction quality and coverage compared to existing methods, as well as strong generalization to real-world data, leading to visually appealing and structurally plausible tree point cloud representations that can be integrated into interactive digital 3D maps. The codebase, synthetic dataset, and pretrained model are publicly available at https://angelikigram.github.io/treeON/.

Real‐time by‐example texture synthesis and filtering using local statistics exchange

Nicolas Lutz, Guillaume Gilet — Wed, 13 May 2026 01:10:38 -0700

Abstract

Real-time by-example texture synthesis is used in interactive virtual worlds to generate the appearance of an unbounded surface from an exemplar texture with as few repetitions as possible. Currently, leading real-time methods rely on a tiling and blending scheme which is known to synthesize well texture patterns with little spatial organization (such as random scratches or noise) or patterns with periodic spatial organization (such as brick walls or regular tilings).

However, attempting to synthesize texture patterns composed of distinct, non-periodic regions with such methods remains a challenge and can lead to visual artifacts. In this paper, we propose a novel texture synthesis method to address this issue. The key of our technique relies on the observation that exchanging the appearance of periodically tiled regions is sufficient to hide repetition artifacts. We therefore present a synthesis scheme that relies on the real-time exchange of local statistics, including means, covariance matrices, or histograms.

Since our target texture is evaluated on the fly, naive filtering schemes relying on precomputation, such as direct MIP-mapping, do not provide accurate results. Therefore, we propose an adequate real-time filtering approximation and show that our method produces high-quality results with little artifacts and a GPU-friendly implementation.

Advances in Feed‐Forward 3D Reconstruction and View Synthesis: A Survey

Wed, 06 May 2026 05:47:11 -0700

This survey reviews feed-forward approaches for 3D reconstruction and view synthesis, which replace iterative optimization with fast, generalizable neural inference. We present a unified taxonomy across representations such as point clouds, 3D Gaussian Splatting, and NeRF, analyse key tasks and datasets, and highlight emerging applications in vision, graphics, and robotics.

Abstract

3D reconstruction and view synthesis are foundational problems in computer vision, graphics, and immersive technologies such as augmented reality (AR), virtual reality (VR), and digital twins. Traditional methods rely on computationally intensive iterative optimization in a complex chain, limiting their applicability in real-world scenarios. Recent advances in feed-forward approaches, driven by deep learning, have revolutionized this field by enabling fast and generalizable 3D reconstruction and view synthesis. This survey offers a comprehensive review of feed-forward techniques for 3D reconstruction and view synthesis, with a taxonomy according to the underlying representation architectures including point cloud, 3D Gaussian Splatting (3DGS), Neural Radiance Fields (NeRF), etc. We examine key tasks such as pose-free reconstruction, dynamic 3D reconstruction, and 3D-aware image and video synthesis, highlighting their applications in digital humans, SLAM, robotics, and beyond. In addition, we review commonly used datasets with detailed statistics, along with evaluation protocols for various downstream tasks. We conclude by discussing open research challenges and promising directions for future work, emphasizing the potential of feed-forward approaches to advance the state of the art in 3D vision. The project page is available at https://fnzhan.com/projects/Feed-Forward-3D Feed-Forward-3D.

A Theoretical Approach for Structuring and Analysing Knowledge Provenance for Visual Analytics

L. Christino, S. Rezaeipour, E. Milios, F. Paulovich — Mon, 04 May 2026 23:47:00 -0700

VAKG unfolds the interactions within the current knowledge model (A) into a temporal knowledge graph (B), which is structured as a 4-way graph containing two temporal (green) and two static (yellow) knowledge graphs. By using VAKG, one can structure and store the user's knowledge-gathering process and all related interactions for eventual analysis (C).

Abstract

The primary goal of visual analytics (VA) is to enable user-guided knowledge generation. Theoretical VA aims to explain how the different aspects of a VA tool yield new insights through user interactivity, which itself can be captured using tracking methods for reproduction or evaluation. However, the strategy of automatically capturing the user's thought processes, such as intent and insights, and associating them with user interaction events is largely ignored. Also, two forms of interactivity capture are typically ambiguous and intermixed: the temporal aspect, which indicates sequences of events, and the atemporal aspect, which explains the workflow as sequences of states within a state-space. In this article, we propose the visual analytics knowledge graph (VAKG), a conceptual framework that brings VA modelling theory to practice through a novel set-theory formalization of knowledge modelling. By extracting such a model from a VA tool, VAKG constructs a 4-way temporal knowledge graph that describes user behaviour and the associated knowledge-gain process. Such knowledge graphs can be populated manually or automatically during user analytical sessions, and then analysed using graph-based methods. VAKG is demonstrated by modelling and collecting Tableau and visual text-mining workflows, enabling the extraction of comparative user satisfaction, tool efficacy, and overall workflow shortcomings from the produced knowledge graph.

Pro‐DG: Procedural Diffusion Guidance for Architectural Facade Generation

A. Plocharski, J. Swidzinski, P. Musialski — Fri, 01 May 2026 06:49:57 -0700

We introduce a framework that integrates symbolic procedural facade grammars with diffusion models for full-image facade editing and propose a novel hierarchical matching algorithm that combines structural and visual metrics to compute transformation maps. We demonstrate significant improvements over existing baselines in preserving both structural accuracy and visual coherence.

Abstract

We use hierarchical procedural rules for the generation of control maps within the stable diffusion framework to produce photo-realistic architectural facade images. Starting from a single input image and its segmentation, we apply an inverse procedural module to identify the facade's hierarchical layout. Leveraging this hierarchy and structural features, we introduce a novel ControlNet pipeline that generates new facade imagery guided by procedural transformations. Our method enables various structural edits, including floor duplication and window rearrangement, by integrating hierarchical alignment directly into control maps. This precisely guides the diffusion-based generative process, ensuring local appearance fidelity alongside extensive structural modifications. Comprehensive evaluations, including comparisons with inpainting-based approaches and synthetic benchmarks, confirm our approach's superior capability in preserving architectural identity and achieving accurate, controllable edits. Quantitative results and user feedback validate our method's effectiveness.

Mixed Super‐Circles

Emile Hohnadel, Thibaut Métivet, Florence Bertails‐Descoubes — Fri, 01 May 2026 01:11:45 -0700

Abstract

We introduce mixed super-circles, a position-curvature formulation of the original dynamic 2D super-helix model. Compared to the latter, purely curvature-based model – the so-called chained formulation –, the mixed formulation that we propose here drastically reduces the algorithmic complexity of the solving scheme – from quadratic to quasi-linear – and simplifies the handling of positional constraints, including contacts. As such, it recovers the main advantages of classical position-based models, while at the same time preserving the high-order convergence of curvature-based models, hence offering an interesting trade-off. Furthermore, the smooth, piecewise circular arc representation of super-circles allows to avoid the spurious jumps in contact forces that are difficult to get rid of with position-based models. Our model is validated quantitatively against demanding mechanical tests involving contact, friction, snapping, and vibrations. Moreover, its versatility, robustness and efficiency are demonstrated through several interactive dynamic scenarios featuring multiple planar elastic rods subject to various types of boundary conditions and constraints. The corresponding source code, Circonflex, is freely delivered to the research community under the GNU GPL v3 licence.

NAADF: Globally Illuminated Voxel Worlds Accelerated with Nested Axis‐Aligned Distance Fields

A. Ulschmid, M. Ott, J. Macho, M. Wimmer, S. Ohrhallinger — Fri, 01 May 2026 01:08:48 -0700

Abstract

Achieving realistic rendering of 3D scenes in real time using path tracing is challenging due to the high sample count required, with ray tracing as the bottleneck. Focusing on voxels as a geometry representation offers significant opportunities for optimizations, especially for tracing the rays, but also for computing the samples. We propose a novel multilayered spatial structure augmented with in-cell axis-aligned distance fields (AADF) operating as caches. Our nested cell structure already accelerates ray tracing 3-5x compared to the state-of-the-art dense spatial structures, such as variants of directed acyclic graphs (DAG). Using the AADFs (constructed while rendering) inside the cells, we can double the ray throughput again (total 10x). As an application, exploiting nested AADFs (NAADFs) also allows us to double the speed of global illumination computations while significantly reducing artifacts from camera motion, such as flickering, blurring, ghosting, and aliasing, all of which are especially important in voxel worlds with sharp edges. We achieve this by adapting temporal antialiasing (TAA) to retain the last 32 frames rather than a single history buffer to create the final antialiased image, since the discretized voxel structure requires much less memory to store the quantized positions and normals of ray bounces. The sample accumulation for global illumination is optimized by compressing and separating lit/unlit samples, and we apply 8x8 window spatial resampling based on a reservoir-based spatiotemporal importance resampling (ReSTIR) method. Our proposed NAADFs support editing with quick updates to the acceleration in the background, overlays of non-aligned dynamic geometry, and can be easily extended to support transform-aware compression or to represent huge real-world scans. Retaining many past frames rather than just combining them opens up new opportunities to remove spatial and temporal artifacts in path tracing for global illumination.

OUGS: Active View Selection via Object‐aware Uncertainty Estimation in 3DGS

Haiyi Li, Qi Chen, Denis Kalkofen, Hsiang‐Ting Chen — Thu, 30 Apr 2026 08:25:58 -0700

Abstract

Recent advances in 3D Gaussian Splatting (3DGS) have achieved state-of-the-art results for novel view synthesis. However, efficiently capturing high-fidelity reconstructions of specific objects within complex scenes remains a significant challenge. A key limitation of existing active reconstruction methods is their reliance on scene-level uncertainty metrics, which are often biased by irrelevant background clutter and lead to inefficient view selection for object-centric tasks. We present OUGS, a novel framework that addresses this challenge with a more principled, physically-grounded uncertainty formulation for 3DGS. Our core innovation is to derive uncertainty directly from the explicit physical parameters of the 3D Gaussian primitives (e.g., position, scale, rotation). By propagating the covariance of these parameters through the rendering Jacobian, we establish a highly interpretable uncertainty model. This foundation allows us to then seamlessly integrate semantic segmentation masks to produce a targeted, object-aware uncertainty score that effectively disentangles the object from its environment. This allows for a more effective active view selection strategy that prioritizes views critical to improving object fidelity. Experimental evaluations on public datasets demonstrate that our approach significantly improves the efficiency of the 3DGS reconstruction process and achieves higher quality for targeted objects compared to existing state-of-the-art methods, while also serving as a robust uncertainty estimator for the global scene.

Step2Motion: Locomotion Reconstruction from Pressure Sensing Insoles

Thu, 30 Apr 2026 08:21:00 -0700

Abstract

Human motion is fundamentally driven by continuous physical interaction with the environment. Whether walking, running, or simply standing, the forces exchanged between our feet and the ground provide crucial insights for understanding and reconstructing human movement. Recent advances in wearable insole devices offer a compelling solution for capturing these forces in diverse, real-world scenarios. Sensor insoles pose no constraint on the users' motion (unlike mocap suits) and are unaffected by line-of-sight limitations (in contrast to optical systems). These qualities make sensor insoles an ideal choice for robust, unconstrained motion capture, particularly in outdoor environments. Surprisingly, leveraging these devices with recent motion reconstruction methods remains largely unexplored. Aiming to fill this gap, we present Step2Motion, the first approach to reconstruct human locomotion from multi-modal insole sensors. Our method utilizes pressure and inertial data—accelerations and angular rates—captured by the insoles to reconstruct human motion. We evaluate the effectiveness of our approach across a range of experiments to show its versatility for diverse locomotion styles, from simple ones like walking or jogging up to moving sideways, on tiptoes, slightly crouching, or dancing. The complete source code, trained model, data, and supplementary material used in this paper can be found at: https://vcai.mpi-inf.mpg.de/projects/Step2Motion/

Robo‐Saber: Generating and Simulating Virtual Reality Players

N. H. Kim, M. J. Liu, J. Lehtinen, P. Hämäläinen, J. F. O'Brien, X. B. Peng — Thu, 30 Apr 2026 08:19:41 -0700

Abstract

We present the first motion generation system for playtesting virtual reality (VR) games. Our player model generates VR headset and handheld controller movements from in-game object arrangements, guided by style reference gameplay examples. We train on the large BOXRR-23 dataset and apply our framework on the popular VR game Beat Saber. The resulting model Robo-Saber reproduces skilled performance and captures diverse player behaviors present in the training data. Robo-Saber demonstrates promise in synthesizing rich gameplay data for predictive applications and enabling whole-body physics-based VR playtesting.

DiskScissors: Cutting Arbitrary‐Topology Solids for Bijective Mapping

S. Hinderink, M. Campen — Thu, 30 Apr 2026 08:17:17 -0700

Abstract

An algorithm for cutting solid objects in a topology-controlled manner is presented. Concretely, given a loop on the object boundary, a disk-topology cut surface bounded by the loop is constructed in the interior. In contrast to various previous approaches, both disk topology and conformance to the prescribed loop are ensured by construction, while supporting not only contractible but also incontractible loops on the boundaries of manifold objects of higher genus and arbitrary non-trivial topology. We describe an implementation of this algorithm in the discrete setting, with triangle mesh cut surfaces embedded in tetrahedral mesh objects. Making use of this novel cutting algorithm, we describe a method for the reliable construction of bijective volumetric maps between solid objects, demonstrating the algorithm's utility. This mapping method overcomes restrictions of the state of the art to topological balls, extending coverage to objects of arbitrary genus, specifically so-called 1-handlebodies.

Differentiable Variable Fonts

Kinjal Parikh, Danny M. Kaufman, David I.W. Levin, Alec Jacobson — Thu, 30 Apr 2026 08:15:42 -0700

Abstract

Typography is an essential component of visual communication. Editing and animating text appearance for graphic designs, title sequences, commercials, and logos remain highly skilled tasks requiring detailed, hands-on efforts from pro artists. Automating these challenging manual workflows requires balancing the competing goals of maintaining a text's legibility and aesthetics, while enabling creative expression. Variable fonts, recent parametric extensions to traditional fonts, offer the promise of new ways to ease and automate typographic design and animation. Variable fonts provide custom-constructed parameters along which fonts can be smoothly varied. These parameterizations could then potentially serve as high-value continuous design spaces, opening the door to modern automated design-optimization tools. However, currently variable fonts are underutilized in creative applications, exactly because artists so far still need to manually tune font parameters. Our work provides intuitive and automated font design and animation workflows with differentiable variable fonts. To do so we distill the current variable font specification to a compact mathematical formulation that differentiably connects the highly non-linear, non-invertible mapping of variable font parameters to the underlying vector graphics representing the text. In turn, this enables us to construct a differentiable framework, with respect to variable font parameters, that allows us to perform gradient-based optimization of energies defined on vector graphics control points, and likewise, via differentiable SVG rasterization, on target rasterized images. We demonstrate the utility of this framework with a range of applications, including direct shape manipulation, overlap aware modeling, physics-based text animation, and automated font-design optimization. Our work now enables leveraging the carefully designed affordances of variable fonts with differentiability to use modern design-optimization technologies, and so opens new possibilities for easy, intuitive and expressive typographic design workflows.

Authoring Terrestrial Planets with Diffusion Models

Thu, 30 Apr 2026 08:15:10 -0700

Abstract

To support the design and subsequent generation of terrestrial planets for use in the creative media, we propose a solution that employs a generative model trained on satellite data from planetary bodies with a defined solid surface, such as the Earth and Mars. A user sketches coarse elevation, landcover, temperature, and precipitation directly onto a globe. Our model then infers high-resolution heightmap and surface appearance layers at planetary scales, with sufficient detail to enable animated flyovers within the exosphere at a distance of a few thousand kilometers from the planet surface. We address the issue of distortion in the mapping from atlas to globe using a quadsphere representation, and the consistency of large-scale geomorphological features by extracting a global river network from the sketch inputs and providing this as conditioning to the diffusion. As our results demonstrate, our generative model provides a balance between: authoring control through a multi-layer painting interface with a satellite image pre-visualization; computation times proportional to the surface area being generated; landscape diversity, displaying, without repetition artefacts, the full range of elevation and landcover features drawn from multiple source planets, and geomorphological plausibility through the provision of a consistent uninterrupted exorheic global river network, where the input sketches allow.

VQ‐Style: Disentangling Style and Content in Motion with Residual Quantized Representations

Thu, 30 Apr 2026 08:13:50 -0700

Abstract

Human motion data is inherently rich and complex, containing both semantic content and subtle stylistic features that are challenging to model. We propose a novel method for effective disentanglement of the style and content in human motion data to facilitate style transfer. Our approach is guided by the insight that content corresponds to coarse motion attributes while style captures the finer, expressive details. To model this hierarchy, we employ Residual Vector Quantized Variational Autoencoders (RVQ-VAEs) to learn a coarse-to-fine representation of motion. We further enhance the disentanglement by integrating codebook learning with contrastive learning and a novel information leakage loss to organize the content and the style across different codebooks. We harness this disentangled representation using our simple and effective inference-time technique Quantized Code Swapping, which enables motion style transfer without requiring any fine-tuning for unseen styles. Our framework demonstrates strong versatility across multiple inference applications, including style transfer, style removal, and motion blending.

Contouring Signed Distance Fields by Approximating Gradients

M. Kohlbrenner, M. Alexa — Thu, 30 Apr 2026 08:12:43 -0700

Abstract

Signed distance fields are often represented by discrete samples (e.g., on a grid). Recovering the contour implicitly represented by the distance samples requires an approximation algorithm. Several recent approaches have shown that exploiting the information carried in each distance sample by explicitly constructing a surface point gives better results than classical contouring algorithms. We explore the idea of generating surface points by simply approximating the gradient of the signed distance function from a tesselation of the sample locations. The distance value together with gradient yields a potential surface point. To avoid problems resulting from bad approximation, surface points are removed if they are too close to any of the distance samples. Using the regular triangulation as tesselation facilitates this filtering. The resulting approximation algorithm is conceptually simple, easy to implement, and significantly faster than existing alternatives, yielding reconstructions that are on par.

Edge‐preserving noise for diffusion models

Jente Vandersanden, Sascha Holl, Xingchang Huang, Gurprit Singh — Thu, 30 Apr 2026 08:12:41 -0700

Abstract

Classical diffusion models typically rely on isotropic Gaussian noise, treating all regions uniformly and overlooking structural information important for high-quality generation. We introduce an edge-preserving diffusion process that generalizes isotropic models via a hybrid noise scheme with an edge-aware scheduler that smoothly transitions from edge-preserving to isotropic noise. This enables the model to capture fine structural details while generally maintaining global performance. We evaluate the impact of structure-aware noise in both diffusion and flow-matching frameworks, and show that existing isotropic models can be efficiently fine-tuned with edge-preserving noise, making our framework practical for adapting pre-trained systems. Beyond unconditional generation, our method particularly shows improvements in structure-guided tasks such as stroke-to-image synthesis, improving robustness and perceptual quality, as evidenced by consistent improvements across FID, KID, and CLIP-score.

Advances in Neural 3D Mesh Texturing: A Survey

Sai Raj Kishore Perla, Hao Zhang, Ali Mahdavi‐Amiri — Thu, 30 Apr 2026 08:12:00 -0700

Abstract

Texturing 3D meshes plays a vital role in determining the visual realism of digital objects and scenes. Although recent generative 3D approaches based on Neural Radiance Fields and Gaussian Splatting can produce textured assets directly, polygonal meshes remain the core representation across modeling, animation, visual effects, and gaming pipelines. Neural 3D mesh texturing therefore continues to be an essential and active area of research. In this survey, we present a comprehensive review of recent advances in neural 3D mesh texturing, covering methods for texture synthesis, transfer, and completion. We first summarize key foundations in mesh geometry, texture mapping, differentiable rendering, and neural generative models, and then organize the literature into a unified taxonomy spanning early GAN-based methods to modern diffusion-based pipelines. We further analyze common architectures and supervision strategies, review datasets and evaluation protocols, and discuss emerging applications, practical/commercial systems, and open challenges. Together, these insights provide a structured perspective on the current landscape and help guide future developments in learning-based 3D mesh texturing.

Project Page: sairajk.github.io/neural-mesh-texturing

Hierarchical Optimization of the As‐Rigid‐As‐Possible Energy

Hendrik Meyer, Bernd Bickel, Marc Alexa — Thu, 30 Apr 2026 08:09:39 -0700

Abstract

The As-Rigid-As-Possible (ARAP) energy [SA07] has become a versatile ingredient in various geometry processing and machine learning methods. The classic method for its minimization is a block coordinate descent, alternating between local rotation estimation and a global linear solve, which converges slowly for large problem instances. We develop and evaluate a multi-level scheme targeted specifically at the optimization of the ARAP energy on large meshes. The main points of our approach are (1) a mesh hierarchy that provides the necessary control over topology while being fast, (2) methods for upsampling the rotations from coarser to finer levels of the hierarchy, and (3) using direct solvers for the linear system. The resulting optimization, remarkably, yields smaller energy while typically being faster on a large number of test cases. The hierarchical approach generalizes to related energies and compares favorably to acceleration schemes such as ADMM, which, in turn, also profit from the hierarchical approach.

SAGE: Structure‐Aware Generative Video Transitions between Diverse Clips

Mia Kan, Yilin Liu, Niloy J. Mitra — Thu, 30 Apr 2026 08:08:43 -0700

Abstract

Video transitions aim to synthesize intermediate frames between two clips, but naïve approaches such as linear blending introduce artifacts that limit professional use or break temporal coherence. Traditional techniques (cross-fades, morphing, frame interpolation) and recent generative inbetweening methods can produce high-quality plausible intermediates, but they struggle with bridging diverse clips involving large temporal gaps or significant semantic differences, leaving a gap for content-aware and visually coherent transitions. We address this challenge by drawing on artistic workflows, distilling strategies such as aligning silhouettes and interpolating salient features to preserve structure and perceptual continuity. Building on these strategies, we propose SAGE (Structure-Aware Generative vidEo transitions) as a simple yet effective zeroshot approach that combines structural guidance, provided via line maps and motion flow, with generative synthesis, enabling smooth, motion-consistent transitions without fine-tuning. Extensive experiments and comparison with current alternatives, namely [RKT*22, ZCL*24, ZZX*24, JHM*25, ZRW*25], demonstrate that SAGE outperforms both classical and the latest generative baselines on quantitative metrics and user studies for producing transitions between diverse clips. The simple method effectively bypasses the need to acquire suitable training data, which is particularly difficult in our creative setting involving diverse clips. Code is available via the project page at https://kan32501.github.io/sage.github.io/.

Embedding Optimization of Layouts via Distortion Minimization

A. Heuschling, I. Lim, L. Kobbelt — Thu, 30 Apr 2026 08:06:56 -0700

Abstract

Given an embedding of a layout in the surface of a target mesh, we consider the problem of optimizing the embedding geometrically. Layout embeddings partition the surface into multiple disk-like patches, making them particularly useful for parametrization and remeshing tasks, such as quad-remeshing, since these problems can then be solved on simpler subdomains. Existing methods can either not guarantee to maintain patch connectivity, limiting downstream applications, or are specialized for quad layout optimization, relying on principal curvature information. We propose a framework that balances per-patch distortion minimization with strict connectivity control through an explicit representation. By inserting additional nodes along layout arcs, they can be embedded as piecewise geodesic curves on the surface. This sampling of arcs provides additional flexibility where required, enabling joint optimization of both node positions and arc embeddings. Our representation naturally supports a multi-resolution workflow: optimization on coarse meshes can be prolongated to high-resolution inputs. We demonstrate its effectiveness in applications requiring connectivity-preserving, low-distortion surface layouts.

Code will be available at https://github.com/7-AlexH/layout-embedding-optimization.

Skeletal‐Driven Animation of Anatomical Humans via Neural Deformation Gradients

G. Nolte, F. Kemper, U. Schwanecke, M. Botsch — Thu, 30 Apr 2026 08:06:53 -0700

Abstract

Most real-time animation techniques for digital humans are limited to deforming the outer skin surface. Geometric skinning methods are highly efficient but struggle with artifacts such as collapsing joints or self-intersections when animating inner anatomy along with the outer skin. Volumetric physics-based simulations, on the other hand, naturally resolve these issues by coordinating bones, muscles, and skin, but are far too slow for interactive use.

We solve this problem by training a neural network to predict deformation gradients. Learning deformation gradients instead of vertex displacements makes our method naturally robust to artifacts such as element inversion or volume deviation. Our model, trained on high-quality finite element simulations, generalizes well across diverse body shapes and poses. This enables anatomically consistent and physically grounded animation of bones, muscles, and skin at interactive frame rates.

Survey on differential estimators for 3d point clouds

Thu, 30 Apr 2026 08:06:01 -0700

Abstract

Recent advancements in 3D scanning technologies, including LiDAR and photogrammetry, have enabled the precise digital replication of real-world objects. These methods are widely used in fields such as GIS, robotics, and cultural heritage. However, the point clouds generated by such scans are often noisy and unstructured, posing challenges for traditional geometry processing tasks. Accurately estimating differential properties like surface curvatures and normals is crucial for tasks such as shape matching and classification, but remains complex due to these inherent challenges. This paper reviews state-of-the-art methods for estimating differential properties from 3D point clouds, with a focus on approaches that offer strong mathematical foundations and theoretical guarantees. We also benchmark these methods using various datasets, evaluating their performance in terms of accuracy, robustness, and efficiency. Our contributions include the release of datasets, tools, and code to promote reproducibility and support future research in this area.

Terrain Synthesis and Authoring based on Iso‐Contours

B. Huftier, H. Schott, E. Galin, O. Argudo, A. Peytavie, E. Guérin — Thu, 30 Apr 2026 08:03:35 -0700

Abstract

Digital terrains are central to realistic landscape depiction, yet authoring tools must balance perceptual realism with intuitive artistic control. We propose a compact vector-based representation that models terrain as nested iso-contours, inspired by geomorphology and cartography. Our method departs from traditional grid-based elevation models by generating contours through an inward Open Eden Growth simulation, followed by marching-triangles reconstruction into a Triangulated Irregular Network. This contour framework supports direct editing such as warping, slope modulation, and smoothing, while allowing reconstruction of a standard elevation map for downstream processing, including erosion and amplification. The approach enables the creation of diverse, realistic terrains from minimal user input and offers simple yet powerful control for designers.

CANRIG: Cross‐Attention Neural Face Rigging with Variable Local Control

Thu, 30 Apr 2026 08:00:23 -0700

Abstract

Facial animation is one of the most labor-intensive aspects of animation and VFX, as traditional rigging consumes weeks of expert time and forces animators to spend countless hours manipulating hundreds of controls to achieve varied expressions. This technical complexity creates a barrier between artistic vision and execution, limiting creative exploration and iteration. In this paper, we introduce CANRig, a fully automated neural facial rigging approach that simplifies the process of creating and editing facial poses by benefiting from global correlations learned from data. Unlike existing neural face models that either sacrifice local control or demand extensive manual region setup, our method introduces continuous local control through a novel conditioning mechanism that operates on a variable region. By modeling deformation as cross-attention between control handles and mesh vertices—modulated by a user-defined region—we enable seamless transitions from precise local adjustments to broad global changes. We further expand our method with a shape-preserving workflow that enables iterative edits, guaranteeing that changes remain untouched even as controls are reconfigured. Our method delivers the best of both worlds: the automation and naturalness of neural methods with the granular control that professional animators demand, and we demonstrate its effectiveness across multiple applications in both animation and high-end visual effects pipelines.

High‐Gloss SVBRDF Capture Using Bounce Light

Tomáš Iser, Andrei‐Timotei Ardelean, Tim Weyrich — Thu, 30 Apr 2026 00:00:00 -0700

Abstract

Reflectance capture aims at the visual reproduction of an object under varying illumination. Past works differ substantially in their experimental overhead, from single- or few-image approaches, that employ significant (often learned) priors at the expense of biased reconstructions, to more accurate approaches that tend to be time-consuming, which to a good part is due to the need for carefully controlled illumination. Moreover, as we will show, the frequently employed point-light or directional lighting tends to clip highlights and under-sample the reflectance of glossy surfaces, leading to incorrect reconstructions under previously unseen illumination. Our work aims to strike a new balance, combining a low-overhead capture methodology with a fast (neural) model fit. A key feature of our approach is the use of handheld, indirect bounce light that enables a convenient capture methodology, limits the dynamic range of the reflectance (effectively avoiding highlight clipping) and ensures contiguous hemispherical incidence, even with few images, eliminating under-sampling of highly specular reflectance lobes. Moreover, our approach does not require training on pre-existing material datasets and thus is not restricted by the choice of dataset, and its inference scales linearly with the number of pixels, scaling exceptionally well to large image sizes. As a result, our method enables high-resolution capture of a spatially-varying reflectance distribution function (SVBRDF) from a small set of casually captured, indirectly lit photographs, making high-quality material acquisition practical even on consumer hardware. Overall, we believe that our method occupies a unique trade-off between acquisition effort, model assumptions and resulting quality, and it has the potential to transform areas that routinely use handheld point-light sources, such as the popular reflectance transformation imaging (RTI), leading to more faithful reproductions of artefacts and their surface characteristics.

Self‐supervised Learning of Fine‐to‐Coarse Cuboid Shape Abstraction

Sat, 25 Apr 2026 01:56:13 -0700

Abstract

The abstraction of 3D objects with simple geometric primitives like cuboids allows us to infer structural information from complex geometry. It is important for 3D shape understanding, structural analysis and geometric modeling.

We introduce a novel fine-to-coarse self-supervised learning approach to abstract collections of 3D shapes. Our architectural design allows us to reduce the number of primitives from hundreds (fine reconstruction) to only a few (coarse abstraction) during training. This allows our network to optimize the reconstruction error and adhere to a user-specified number of primitives per shape while simultaneously learning a consistent structure across the whole collection of data. We achieve this through our abstraction loss formulation which increasingly penalizes redundant primitives. Furthermore, we introduce a reconstruction loss formulation to account not only for surface approximation but also volume preservation. Combining both contributions allows us to represent 3D shapes more precisely with fewer cuboid primitives than previous work.

We evaluate our method on collections of man-made and humanoid shapes comparing with previous state-of-the-art learning methods on commonly used benchmarks. Our results confirm an improvement over previous cuboid-based shape abstraction techniques. Furthermore, we demonstrate our cuboid abstraction in downstream tasks like clustering, retrieval, and partial symmetry detection.

2D Piecewise Linear Scalar Fields with Invertible Integral Lines

T.L. Erxleben, M. Motejat, C. Rössl, H. Theisel — Sat, 25 Apr 2026 01:53:43 -0700

Abstract

Integral lines of the gradient flow are standard features in continuously differentiable scalar fields that enjoy some useful properties: They cover the domain densely, do not split, merge, or intersect, and are therefore invertible. For widely used discretizations of scalar fields, the corresponding polygonal approximations of integral lines do not enjoy these properties anymore. We analyze conditions for integral lines in 2D piecewise linear (PL) scalar fields to be invertible by identifying and classifying critical edges in the underlying triangulation. We show that under mild conditions, every 2D PL scalar field can be transformed into an arbitrarily close PL field with invertible integral lines. We present an algorithm that computes this transformation and apply it to a number of test data sets.

Multi‐Spectral Gaussian Splatting with Neural Color Representation

Sat, 25 Apr 2026 01:52:55 -0700

Abstract

3D Gaussian Splatting (3DGS) [KKLD23] has transformed novel-view synthesis from RGB images, yet remains restricted to the visible spectrum. Many applications, including agricultural monitoring, rely on multi-spectral imaging, where spectral camera alignment and scalability pose major challenges.

We present MS-Splatting—a multi-spectral 3DGS framework enabling unified multi-view consistent reconstruction and rendering across both visible and invisible spectra. Our key component is a neural color representation that encodes per-primitive features shared across spectral bands, decoded through a shallow multi-layer perceptron into spectrum-specific radiance. By leveraging inter-band correlations, this formulation enhances detail while reducing memory consumption compared to independent band modeling via per-channel modeling with spherical harmonics.

Our method enables accurate parallax-free novel-view vegetation index rendering for plant monitoring and enhances RGB novel view synthesis quality by exploiting details revealed through multi-spectral bands. Our evaluation demonstrates that MS-Splatting exceeds the current leading methods in both categories. In addition, we introduce a multi-spectral dataset from aerial captures covering outdoor environments, specifically designed for evaluating these applications. We will release our code and dataset to facilitate further research. The project page is located at: https://meyerls.github.io/ms_splatting

See4D: Pose‐Free 4D Generation via Auto‐Regressive Video Inpainting

Sat, 25 Apr 2026 01:51:21 -0700

Abstract

Immersive applications call for synthesizing spatiotemporal 4D content from casual videos without costly 3D supervision. Existing video-to-4D methods typically rely on manually annotated camera poses, which are labor-intensive and brittle for in-the-wild footage. Recent warp-then-inpaint approaches mitigate the need for pose labels by warping input frames along a novel camera trajectory and using an inpainting model to fill missing regions, thereby depicting the 4D scene from diverse viewpoints. However, this trajectory-to-trajectory formulation often entangles camera motion with scene dynamics and complicates both modeling and inference. We introduce See4D, a pose-free, trajectory-to-camera framework that replaces explicit trajectory prediction with rendering to a bank of fixed virtual cameras, thereby separating camera control from scene modeling. A view-conditional video inpainting model is trained to learn a robust geometry prior by denoising realistically synthesized warped images and to inpaint occluded or missing regions across virtual viewpoints, eliminating the need for explicit 3D annotations. Building on this inpainting core, we design a spatiotemporal autoregressive inference pipeline that traverses virtual-camera splines and extends videos with overlapping windows, enabling coherent generation at bounded per-step complexity. We validate See4D on cross-view video generation and sparse reconstruction benchmarks. Across quantitative metrics and qualitative assessments, our method achieves superior generalization and improved performance relative to pose- or trajectory-conditioned baselines, advancing practical 4D world modeling from casual videos.

Variable‐Rate Texture Compression: Real‐Time Rendering with JPEG

Elias Kristmann, Michael Wimmer, Markus Schütz — Sat, 25 Apr 2026 01:50:23 -0700

Abstract

Although variable-rate compressed image formats such as JPEG are widely used to efficiently encode images, they have not found their way into real-time rendering due to special requirements such as random access to individual texels. In this paper, we investigate the feasibility of variable-rate texture compression on modern GPUs using the JPEG format, and how it compares to the GPU-friendly fixed-rate compression approaches BC1 and ASTC. Using a deferred rendering pipeline, we are able to identify the subset of blocks that are needed for a given frame, decode these, and colorize the framebuffer's pixels. Despite the additional ∼0.17 bit per pixel that we require for our approach, JPEG maintains significantly better quality and compression rates compared to BC1, and depending on the type of image, outperforms or competes with ASTC. The JPEG rendering pipeline increases rendering duration by less than 0.3 ms on an NVIDIA RTX 4090, demonstrating that sophisticated variable-rate compression schemes are feasible on modern GPUs, even in VR.

Source code and data sets are available at: https://github.com/elias1518693/jpeg_textures

SketchCrafter: Sketch Extraction as a Training‐Free Generative Process

Thu, 23 Apr 2026 09:06:41 -0700

SketchCrafter reformulates sketch extraction as a generative process to address two key limitations: internal aggregation often fails to preserve brushstroke consistency with the style image, while external aggregation can misalign content details with the source image. This approach enables high-fidelity sketch extraction with customizable brushstroke styles.

Abstract

We introduce SketchCrafter, a training-free framework that formulates sketch extraction as a generative process rather than relying on aggregation. SketchCrafter initiates generation with an edge-based initialization, projecting edge features from the content image into the diffusion latent space to create a structured noise foundation. This foundation enables controlled integration of style and content, achieving high fidelity to both the source image and reference style. A dual-control mechanism ensures coherence at multiple levels: locally, content features fuse with edge structures to preserve fine details, while globally, a content encoder reinforces structural alignment. Style information is integrated in parallel, with key and value tokens capturing fine stylistic details, supported by a global style encoder for overall consistency. Experiments show that SketchCrafter excels in content alignment and style fidelity across benchmarks, proving it as an effective solution for training-free sketch extraction.

TABI: Tight and Balanced Interactive Atlas Packing

F. Gu, N. Vining, A. Sheffer — Tue, 21 Apr 2026 08:11:57 -0700

Abstract

Atlas packing is a key step in many computer graphics applications. Packing algorithms seek to arrange a set of charts within a fixed-size atlas with as little downscaling as possible. Many packing applications such as content creation tools, dynamic atlas generation for video games, and texture space shading require on-the-fly interactive atlas packing. Unfortunately, while many methods have been developed for generating tight high-quality packings, they are designed for offline settings and have running times two or more orders of magnitude greater than what is required for interactive performance. While real-time GPU packing methods exist, they significantly downscale packed charts compared to offline methods. We introduce a GPU packing method that targets interactive speeds, provides packing quality approaching that of offline methods, and supports flexible user control over the tradeoff between performance and quality. We observe that current real-time packing methods leave large gaps between charts and often produce asymmetric, or poorly balanced, packings. These artifacts dramatically degrade packing quality. Our Tight And Balanced method eliminates these artifacts while retaining Interactive performance. TABI generates tight packings by compacting empty space between irregularly shaped charts both horizontally and vertically, using two approximations of chart shape that support efficient parallel processing. We balance packing outputs by automatically adjusting atlas row widths and orientations to accommodate varying chart heights. We show that our method significantly reduces chart downscaling compared to existing interactive methods while remaining orders of magnitude faster than offline alternatives.

A Real‐Time Multi‐Scale Neural Representation for Complex Surface Reflectance

Heikki Timonen, Pauli Kemppinen, Jaakko Lehtinen — Tue, 21 Apr 2026 08:10:42 -0700

Abstract

Recent machine learning methods have significantly advanced the state of the art in the classic problem of representing surface appearance over angle, space, and scale. The models tend, however, to be relatively heavy compared to traditional fixed-function representations, making real-time application challenging. We present a neural shading architecture that allows the use of smaller and faster-to-evaluate neural networks than current state of the art, while faithfully representing complex spatial and angular variation. We target the angular complexity that arises both from prefiltering normal-mapped SVBRDFs, as well as complex, measured homogeneous BRDFs. A key architectural innovation is the introduction of a multiplicative interaction (“gating”) between learnable parameters that significantly increases our model's expressive power. Our straightforward, unop-timized shader implementation renders over 1000 full HD frames per second on a consumer GPU using our default parameters.

Statistical Denoising of Transient Rendering

Oscar Pueyo‐Ciutad, Alvaro Lopez, Diego Gutierrez — Tue, 21 Apr 2026 08:09:28 -0700

Abstract

Transient rendering simulates light in motion, measuring the time of flight from the light source to the camera. However, the stochastic nature of Monte Carlo is aggravated in transient rendering, since samples are now spread along the temporal domain. In our work, we propose to denoise transient Monte Carlo renders by exploiting the spatio-temporal correlation of transient light transport, extending a recent statistical denoising formulation. By relying on statistics, we achieve a near-optimal tradeoff between reduced variance and introduced bias. We efficiently collect per-time-bin statistics in the temporal domain while avoiding impractical memory requirements, and use these collected statistics to analyze the spatio-temporal correlation and discriminate which time bins should be combined. Our statistics-based transient denoiser does not hallucinate, guarantees convergence of the result, is efficient, does not require any training and naturally handles participating media. We believe that the generality of our method might pave the way for denoising time-resolved Monte Carlo simulations in other domains, such as non-line-of-sight imaging, acoustic rendering, or absorption microscopy.

Orthographer: Generating Orthographic‐Style Projections for Elongated Architectural Structures

Yu‐Hsuan Hsieh, Yu‐Ting Wu — Mon, 20 Apr 2026 06:27:26 -0700

This paper presents a framework for generating orthographic-style projections of elongated buildings, which are difficult to capture with standard imaging systems due to field-of-view limitations. Experimental results demonstrate that our approach preserves architectural accuracy more effectively than conventional panoramas, while also providing a more accessible and streamlined image acquisition process.

Abstract

Photographing architecturally expansive structures presents significant challenges due to the limited field of view (FoV) inherent to standard imaging systems. Traditional solutions, including the use of wide-angle optics or panoramic stitching techniques, frequently introduce significant geometric distortions that compromise the fidelity of architectural forms. To address this limitation, we introduce a novel framework for synthesizing orthographic-style projections of elongated buildings, with an emphasis on structural preservation. The proposed method leverages a neural semantic segmentation network to autonomously identify architectural components and foreground elements within a panoramic image. Geometric rectification is performed via a forward mapping algorithm, derived from extracted boundary contours and vanishing point estimations, and is complemented by an inpainting stage to recover missing regions. Empirical evaluations confirm that our pipeline maintains architectural accuracy more effectively than conventional techniques, while offering an accessible and streamlined image acquisition process.

Improving Facial Rig Semantics for Tracking and Retargeting

D. Omens, A. Thurman, J. Yu, R. Fedkiw — Wed, 15 Apr 2026 03:20:01 -0700

Abstract

In this paper, we consider retargeting a tracked facial performance to other people or virtual characters. We utilize the same rig framework for both tracking and animation to remove the difficulties associated with retargeting the semantics of one framework to another. Our carefully designed set of Simon-Says expressions and regularizers is used to calibrate each rig to the motion signatures of the relevant performer or target. Although a uniform set of Simon-Says expressions can likely be used for all person-to-person retargeting, we argue that person-to-virtual-character retargeting benefits from an expression set that captures the distinct motion signature of the virtual character rig. The Simon-Says calibrated rigs tend to produce the desired expressions when exercising animation controls. Unfortunately, these well-calibrated rigs still lead to undesirable controls when tracking a performance, even though they generally produce acceptable geometry reconstructions. Thus, we propose a fine-tuning approach that modifies the rig used by the tracker to promote the output of more semantically meaningful animation controls, facilitating high efficacy retargeting. To better address real-world scenarios, the fine-tuning relies on implicit differentiation so that the tracker can be treated as a potentially non-differentiable black box. Experiments demonstrate the benefits of our calibration methods on high-fidelity expressive performance retargeting for different capture conditions, trackers, and rig frameworks.

Establishing Shape Correspondences: A Survey

A. Heuschling, H. Meinhold, L. Kobbelt — Wed, 15 Apr 2026 03:16:08 -0700

Abstract

Shape correspondence between surfaces in 3D is a central problem in geometry processing, concerned with establishing meaningful relations between surfaces. While all correspondence problems share this goal, specific formulations can differ significantly: Downstream applications require certain properties that correspondences must satisfy, while the type of input data and computational constraints influence the choice of method. In this survey, we provide an overview of different correspondence problems, popular paradigms for generating and refining correspondences, and strategies for evaluating their quality. Further, we discuss topological aspects that are especially important for correspondences between surfaces with higher genus. By offering a structured overview and highlighting open challenges, we aim to support and guide future research in the field.

Real‐time Rendering with a Neural Irradiance Volume

Wed, 15 Apr 2026 03:10:25 -0700

Abstract

Rendering diffuse global illumination in real-time is often approximated by pre-computing and storing irradiance in a 3D grid of probes. As long as most of the scene remains static, probes approximate irradiance for all surfaces immersed in the irradiance volume, including novel dynamic objects. This approach, however, suffers from aliasing artifacts and high memory consumption. We propose Neural Irradiance Volume (NIV), a neural-based technique that allows accurate real-time rendering of diffuse global illumination via a compact pre-computed model, overcoming the limitations of traditional probe-based methods, such as the expensive memory footprint, aliasing artifacts, and scene-specific heuristics. The key insight is that neural compression creates an adaptive and amortized representation of irradiance, circumventing the cubic scaling of grid-based methods. Our superior memory-scaling improves quality by at least 10x at the same memory budget, and enables a straightforward representation of higher-dimensional irradiance fields, allowing rendering of time-varying or dynamic effects without requiring additional computation at runtime. Unlike other neural rendering techniques, our method works within strict realtime constraints, providing fast inference (around 1 ms per frame on consumer GPUs at full HD resolution), reduced memory usage (1–5 MB for medium-sized scenes), and only requires a G-buffer as input, without expensive ray tracing or denoising.

Generative Cutout Animation

Ivan Puhachov, Noam Aigerman, Thibault Groueix, Mikhail Bessmeltsev — Wed, 15 Apr 2026 03:07:51 -0700

Abstract

Cutout animation is one of the earliest forms of animation, and to this day remains a popular technique featured in numerous films including Monty Python and South Park series. Most computer animation systems, however, focus on different styles, including cel animation, making cutout animation somewhat underexplored. As creating cutouts is meticulous, we propose a novel generative cutout animation system. Taking a skeletal animation and a text prompt as input, we automatically generate a 2.5D cutout rig ready for production in films and games. Our system optimizes cutout images with an SDS (Score Distillation Sampling) loss with a LoRA (Low-Rank Adaptation) prior, in multiple target poses. Naïvely optimizing an SDS loss, however, would lead to inconsistent target pose images, and, as a result, blurry or transparent cutouts. To address this, we introduce a novel optimization with techniques targeting pose and noise consistency, resulting in coherent target images and sharp cutouts. We validate our system by demonstrating a gallery of results, comparing with previous works, ablations, and other analyses. Once generated, our cutout rigs can be used both for the given input animation and repurposed for other animations or edited as independent assets.

Latent Diffusion‐GAN: Adversarial Learning in the Autoencoded Latent Space

U‐Chae Jun, Jaeeun Ko, Jiwoo Kang — Wed, 15 Apr 2026 03:07:17 -0700

Abstract

Diffusion models are powerful generative frameworks for producing high-quality images by denoising latent variables from random noise. However, training with likelihood-based objectives, such as denoising score matching, can lead to locally oversmoothed high-frequency details, including fine textures and sharp edges, thereby limiting perceptual fidelity and structural detail. Adversarial training with GANs enhances sharpness but typically requires additional discriminator networks, increasing computational costs and destabilizing training. To this end, we propose Latent Diffusion Generative Adversarial Networks (LD-GAN), a novel framework that seamlessly integrates adversarial learning into diffusion models without modifying their original pipeline. LD-GAN leverages the pretrained variational autoencoder (VAE) in latent diffusion models as an energy-based discriminator, enabling adversarial training without extra parameters and preserving the structured latent priors learned from large datasets. We also introduce a structural consistency energy that aligns encoder and decoder feature representations, thereby enhancing perceptual quality and compatibility with the pretrained latent space. Extensive experiments demonstrate that LD-GAN significantly improves sample fidelity, perceptual sharpness, and diversity over state-of-the-art baseline methods across various generation tasks while ensuring efficient training dynamics.

Conversational Gesture Model (CGM): Extending Speaker‐Centric Audio‐Driven Motion Generation to Full Conversation Gestures

T. Koren, A. Rosenthal, D. Friedman, A. Shamir — Wed, 15 Apr 2026 03:07:16 -0700

Abstract

In this work we extend speaker-centric audio-driven gesture synthesis toward a unified conversational model that jointly captures both speaking and listening behaviors. Existing speaker-centric models effectively generate gestures aligned with speech but overlook the bidirectional dynamics that characterize natural dialogue. To address this limitation, we propose the Conversational Gesture Model (CGM), a cross-attention-based model capable of synthesizing gestures conditioned on interlocutor conversational cues such as gestures, tone, and textual semantics. By leveraging cross-attention mechanisms, the model fuses interlocutor audio and text features with character gesture encodings, enabling a single system to seamlessly alternate between speaking and listening roles of the same character. Hence, our model enables a single system to act as both speaker and listener, capturing the fluid role shifts and mutual influence inherent in conversation. Experiments demonstrate that this approach preserves the quality of speaker-driven gestures while significantly improving the realism, coherence, and responsiveness of full conversational interactions.

Non‐Rigid 3D Shape Correspondences: From Foundations to Open Challenges and Opportunities

Wed, 15 Apr 2026 03:06:34 -0700

Abstract

Estimating correspondences between deformed shape instances is a long-standing problem in computer graphics; numerous applications, from texture transfer to statistical modelling, rely on recovering an accurate correspondence map. Many methods have thus been proposed to tackle this challenging problem from varying perspectives, depending on the downstream application. This state-of-the-art report is geared towards researchers, practitioners, and students seeking to understand recent trends and advances in the field. We categorise developments into three paradigms: spectral methods based on functional maps, combinatorial formulations that impose discrete constraints, and deformation-based methods that directly recover a global alignment. Each school of thought offers different advantages and disadvantages, which we discuss throughout the report. Meanwhile, we highlight the latest developments in each area and suggest new potential research directions. Finally, we provide an overview of emerging challenges and opportunities in this growing field, including the recent use of vision foundation models for zero-shot correspondence and the particularly challenging task of matching partial shapes.

MF‐SDF: Neural Implicit Surface Reconstruction using Mixed Incident Illumination and Fourier Feature Optimization

Xueyang Zhou, Xukun Shen, Yong Hu — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

The utilization of neural implicit surface as a geometry representation has proven to be an effective multi-view surface reconstruction method. Despite the promising results achieved, reconstructing geometry from objects in real-world scenes remains challenging due to the interaction between surface materials and complex ambient light, as well as shadow effects caused by self-occlusion, making it a highly ill-posed problem. To address this challenge, we propose MF-SDF, a method that use a hybrid neural network and spherical gaussian representation to model environmental lighting, so that the model can express the situation of multiple light sources including directional light (such as outdoor sunlight) in real-world scenarios. Benefit from this, our method effectively reconstructs coherent surfaces and accurately locates the shadow location on the surface. Furthermore, we adopt a shadow aware multi-view photometric consistency loss, which mitigates the erroneous reconstruction results of previous methods on surfaces containing shadows, thereby improve the overall smoothness of the surface. Additionally, unlike previous approaches that directly optimize spatial features, we propose a Fourier feature optimization method that directly optimizes the tensorial feature in the frequency domain. By optimizing the high-frequency components, this approach further enhances the details of surface reconstruction. Finally, through experiments, we demonstrate that our method outperforms existing methods in terms of reconstruction accuracy on real captured data.

Region‐Aware Sparse Attention Network for Lane Detection

Yan Deng, Guoqiang Xiao — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Lane detection is a fundamental task in intelligent driving systems. However, the slender and sparse structure of lanes, combined with the dominance of irrelevant background regions in road scenes, makes accurate lane localization particularly challenging, especially under complex and adverse conditions. To address these issues, we propose a novel Region-Aware Sparse Attention Network (RSANet), which is designed to selectively enhance lane-relevant features while suppressing background interference. Specifically, we introduce the Region-guided Pooling Predictor (RPP) that generates lane region activation maps to guide the backbone network in focusing on informative areas. To improve the multi-scale feature fusion capability of the Feature Pyramid Network (FPN), we propose the Bilateral Pooling Attention Module (BPAM) that captures discriminative features by jointly modeling dependencies along both the channel and spatial dimensions. Furthermore, the Lane-guided Sparse Attention Mechanism (LSAM) efficiently aggregates global contextual information from the most relevant spatial regions to reinforce lane prior representations while significantly reducing redundant computation. Extensive experiments on benchmark datasets demonstrate that RSANet outperforms state-of-the-art methods in a variety of challenging scenarios. Notably, RSANet achieves an F1@50 score of 80.04% on the CULane dataset that shows notable improvements.

A Solver‐Aided Hierarchical Language for LLM‐Driven CAD Design

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Parametric CAD systems use domain-specific languages (DSLs) to represent geometry as programs, enabling both flexible modeling and structured editing. With the rise of large language models (LLMs), there is growing interest in generating such programs from natural language. This raises a key question: what kind of DSL best supports both CAD generation and editing, whether performed by a human or an AI? In this work, we introduce AIDL, a hierarchical, solver-aided DSL designed to align with the strengths of LLMs while remaining interpretable and editable by humans. AIDL enables high-level reasoning by breaking problems into abstract components and structural relationships, while offloading low-level geometric reasoning to a constraint solver. We evaluate AIDL in a 2D text-to-CAD setting using a zero-shot prompt-based interface and compare it to OpenSCAD, a widely used CAD DSL that appears in LLM training data. AIDL produces results that are visually competitive and significantly easier to edit. Our findings suggest that language design is a powerful complement to model training and prompt engineering for building collaborative AI–human tools in CAD. Code is available at https://github.com/deGravity/aidl.

Gaussian Splatting for Large‐Scale Aerial Scene Reconstruction From Ultra‐High‐Resolution Images

Qiulin Sun, Wei Lai, Yixian Li, Yanci Zhang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Using 3D Gaussian splatting to reconstruct large-scale aerial scenes from ultra-high-resolution images is still a challenge problem because of two memory bottlenecks - excessive Gaussian primitives and the tensor sizes for ultra-high-resolution images. In this paper, we propose a task partitioning algorithm that operates in both object and image space to generate a set of small-scale subtasks. Each subtask's memory footprints is strictly limited, enabling training on a single high-end consumer-grade GPU. More specifically, Gaussian primitives are clustered into blocks in object space, and the input images are partitioned into sub-images according to the projected footprints of these blocks. This dual-space partitioning significantly reduces training memory requirements. During subtask training, we propose a depth comparison method to generate a mask map for each sub-image. This mask map isolates pixels primarily contributed by the Gaussian primitives of the current subtask, excluding all other pixels from training. Experimental results demonstrate that our method successfully achieves large-scale aerial scene reconstruction using 9K resolution images on a single RTX 4090 GPU. The novel views synthesized by our method retain significantly more details than those from current state-of-the-art methods.

PARC: A Two‐Stage Multi‐Modal Framework for Point Cloud Completion

Yujiao Cai, Yuhao Su — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Point cloud completion is vital for accurate 3D reconstruction, yet real world scans frequently exhibit large structural gaps that compromise recovery. Meanwhile, in 2D vision, VAR (Visual Auto-Regression) has demonstrated that a coarse-to-fine “next-scale prediction” can significantly improve generation quality, inference speed, and generalization. Because this coarse-to-fine approach closely aligns with the progressive nature of filling missing geometry in point clouds, we were inspired to develop PARC (Patch-Aware Coarse-to-Fine Refinement Completion), a two-stage multimodal framework specifically designed for handling missing structures. In the pretraining stage, PARC leverages complete point clouds alongside a Patch-Aware Coarse-to-Fine Refinement (PAR) strategy and a Mixture-of-Experts (MoE) architecture to generate high-quality local fragments, thereby improving geometric structure understanding and feature representation quality. During finetuning, the model is adapted to partial scans, further enhancing its resilience to incomplete inputs. To address remaining uncertainties in areas with missing structure, we introduce a dual-branch architecture that incorporates image cues: point cloud and image features are extracted independently and then fused via the MoE with an alignment loss, allowing complementary modalities to guide reconstruction in occluded or missing regions. Experiments conducted on the ShapeNet-ViPC dataset show that PARC has achieved highly competitive performance. Code is available at https://github.com/caiyujiaocyj/PARC.

Geometric Integration for Neural Control Variates

D. Meister, T. Harada — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Control variates are a variance-reduction technique for Monte Carlo integration. The principle involves approximating the integrand by a function that can be analytically integrated, and integrating using the Monte Carlo method only the residual difference between the integrand and the approximation, to obtain an unbiased estimate. Neural networks are universal approx-imators that could potentially be used as a control variate. However, the challenge lies in the analytic integration, which is not possible in general. In this manuscript, we study one of the simplest neural network models, the multilayered perceptron (MLP) with continuous piecewise linear activation functions, and its possible analytic integration. We propose an integration method based on integration domain subdivision, employing techniques from computational geometry to solve this problem in 2D. We demonstrate that an MLP can be used as a control variate in combination with our integration method, showing applications in the light transport simulation.

IPFNet: Implicit Primitive Fitting for Robust Point Cloud Segmentation

Shengdi Zhou, Xiaoqiang Zan, Bin Zhou — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

The segmentation and fitting of geometric primitives from point clouds is a widely adopted approach for modelling the underlying geometric structure of objects in reverse engineering and numerous graphics applications. Existing methods either overlook the role of geometric information in assisting segmentation or incorporate reconstruction losses without leveraging modern neural implicit field representations, leading to limited robustness against noise and weak expressive power in reconstruction. We propose a point cloud segmentation and fitting framework based on neural implicit representations, fully leveraging neural implicit fields' expressive power and robustness. The key idea is the unification of geometric representation within a neural implicit field framework, enabling seamless integration of geometric loss for improved performance. In contrast to previous approaches that focus solely on clustering in the feature embedding space, our method enhances instance segmentation through semantic-aware point embeddings and simultaneously improves semantic predictions via instance-level feature fusion. Furthermore, we incorporate 3D-specific cues such as spatial dimensions and geometric connectivity, which are uniquely informative in the 3D domain. Extensive experiments and comparisons against previous methods demonstrate our robustness and superiority.

StyleMM: Stylized 3D Morphable Face Model via Text‐Driven Aligned Image Translation

Seungmi Lee, Kwan Yun, Junyong Noh — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

We introduce StyleMM, a novel framework that can construct a stylized 3D Morphable Model (3DMM) based on user-defined text descriptions specifying a target style. Building upon a pre-trained mesh deformation network and a texture generator for original 3DMM-based realistic human faces, our approach fine-tunes these models using stylized facial images generated via text-guided image-to-image (i2i) translation with a diffusion model, which serve as stylization targets for the rendered mesh. To prevent undesired changes in identity, facial alignment, or expressions during i2i translation, we introduce a stylization method that explicitly preserves the facial attributes of the source image. By maintaining these critical attributes during image stylization, the proposed approach ensures consistent 3D style transfer across the 3DMM parameter space through image-based training. Once trained, StyleMM enables feed-forward generation of stylized face meshes with explicit control over shape, expression, and texture parameters, producing meshes with consistent vertex connectivity and animatability. Quantitative and qualitative evaluations demonstrate that our approach outperforms state-of-the-art methods in terms of identity-level facial diversity and stylization capability. The code and videos are available at kwanyun.github.io/stylemm_page.

Categories and Subject Descriptors (according to ACM CCS): I.3.6 [Computer Graphics]: Methodology and Techniques—

Swept Volume Computation with Enhanced Geometric Detail Preservation

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Swept volume computation—the determination of regions occupied by moving objects—is essential in graphics, robotics, and manufacturing. Existing approaches either explicitly track surfaces, suffering from robustness issues under complex interactions, or employ implicit representations that trade off geometric fidelity and face optimization difficulties. We propose a novel inversion of motion perspective: rather than tracking object motion, we fix the object and trace spatial points backward in time, reducing complex trajectories to efficiently linearizable point motions. Based on this, we introduce a multi-field tetrahedral framework that maintains multiple distance fileds per element, preserving fine geometric details at trajectory intersections where single-field methods fail. Our method robustly computes swept volumes for diverse motions, including translations and screw motions, and enables practical applications in path planning and collision detection.

Uncertainty‐Aware Adjustment via Learnable Coefficients for Detailed 3D Reconstruction of Clothed Humans from Single Images

Yadan Yang, Yunze Li, Fangli Ying, Aniwat Phaphuangwittayakul, Riyad Dhuny — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Although single-image 3D human reconstruction has made significant progress in recent years, few of the current state-of-the-art methods can accurately restore the appearance and geometric details of loose clothing. To achieve high-quality reconstruction of a human body wearing loose clothing, we propose a learnable dynamic adjustment framework that integrates side-view features and the uncertainty of the parametric human body model to adaptively regulate its reliability based on the clothing type. Specifically, we first adopt the Vision Transformer model as an encoder to capture the image features of the input image, and then employ SMPL-X to decouple the side-view body features. Secondly, to reduce the limitations imposed by the regularization of the parametric model, particularly for loose garments, we introduce a learnable coefficient to reduce the reliance on SMPL-X. This strategy effectively accommodates the large deformations caused by loose clothing, thereby accurately expressing the posture and clothing in the image. To evaluate the effectiveness, we validate our method on the public CLOTH4D and Cape datasets, and the experimental results demonstrate better performance compared to existing approaches. The code is available at https://github.com/yyd0613/CoRe-Human.

ClothingTwin: Reconstructing Inner and Outer Layers of Clothing Using 3D Gaussian Splatting

Munkyung Jung, Dohae Lee, In‐Kwon Lee — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

We introduce ClothingTwin, a novel end-to-end framework for reconstructing 3D digital twins of clothing that capture both the outer and inner fabric —without the need for manual mannequin removal. Traditional 2D “ghost mannequin” photography techniques remove the mannequin and composite partial inner textures to create images in which the garment appears as if it were worn by a transparent model. However, extending such method to photorealistic 3D Gaussian Splatting (3DGS) is far more challenging. Achieving consistent inner-layer compositing across the large sets of images used for 3DGS optimization quickly becomes impractical if done manually. To address these issues, ClothingTwin introduces three key innovations. First, a specialized image acquisition protocol captures two sets of images for each garment: one worn normally on the mannequin (outer layer exposed) and one worn inside-out (inner layer exposed). This eliminates the need to painstakingly edit out mannequins in thousands of images and provides full coverage of all fabric surfaces. Second, we employ a mesh-guided 3DGS reconstruction for each layer and leverage Non-Rigid Iterative Closest Point (ICP) to align outer and inner point-clouds despite distinct geometries. Third, our enhanced rendering pipeline—featuring mesh-guided back-face culling, back-to-front alpha blending, and recalculated spherical harmonic angles—ensures photorealistic visualization of the combined outer and inner layers without inter-layer artifacts. Experimental evaluations on various garments show that ClothingTwin outperforms conventional 3DGS-based methods, and our ablation study validates the effectiveness of each proposed component.

Feature Disentanglement in GANs for Photorealistic Multi‐view Hair Transfer

Jiayi Xu, Zhengyang Wu, Chenming Zhang, Xiaogang Jin, Yaohua Ji — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Fast and highly realistic multi-view hair transfer plays a crucial role in evaluating the effectiveness of virtual hair try-on systems. However, GAN-based generation and editing methods face persistent challenges in feature disentanglement. Achieving pixel-level, attribute-specific modifications—such as changing hairstyle or hair color without affecting other facial features—remains a long-standing problem. To address this limitation, we propose a novel multi-view hair transfer framework that leverages a hair-only intermediate facial representation and a 3D-guided masking mechanism. Our approach disentangles tri-plane facial features into spatial geometric components and global style descriptors, enabling independent and precise control over hairstyle and hair color. By introducing a dedicated intermediate representation focused solely on hair and incorporating a two-stage feature fusion strategy guided by the generated 3D mask, our framework achieves fine-grained local editing across multiple viewpoints while preserving facial integrity and improving background consistency. Extensive experiments demonstrate that our method produces visually compelling and natural results in side-to-front view hair transfer tasks, offering a robust and flexible solution for high-fidelity hair reconstruction and manipulation.

View‐Independent Wire Art Modeling via Manifold Fitting

HuiGuang Huang, Dong‐Yi Wu, Yulin Wang, Yu Cao, Tong‐Yee Lee — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

This paper presents a novel fully automated method for generating view-independent abstract wire art from 3D models. The main challenge in creating line art is to strike a balance among abstraction, structural clarity, 3D perception, and consistent aesthetics from different viewpoints. Many existing approaches have been proposed, including extracting wire art from mesh, reconstructing it from pictures, etc. But they all suffer from the fact that the wires are usually very unorganized and cumbersome and usually can only guarantee the observation effect of specific viewpoints. To overcome these problems, we propose a paradigm shift: instead of predicting the line segments directly, we consider the generation of wire art as an optimization-driven manifold-fitting problem. Thus we can abstract/generalize the 3D model while retaining the key properties necessary for appealing line art, including structural topology and connectivity, and maintain the three-dimensionality of the line art with a multi-perspective view. Experimental results show that our view-independent method outperforms previous methods in terms of line simplicity, shape fidelity, and visual consistency.

SPG: Style‐Prompting Guidance for Style‐Specific Content Creation

Qian Liang, Zichong Chen, Yang Zhou, Hui Huang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Although recent text-to-image (T2I) diffusion models excel at aligning generated images with textual prompts, controlling the visual style of the output remains a challenging task. In this work, we propose Style-Prompting Guidance (SPG), a novel sampling strategy for style-specific image generation. SPG constructs a style noise vector and leverages its directional deviation from unconditional noise to guide the diffusion process toward the target style distribution. By integrating SPG with Classifier-Free Guidance (CFG), our method achieves both semantic fidelity and style consistency. SPG is simple, robust, and compatible with controllable frameworks like ControlNet and IPAdapter, making it practical and widely applicable. Extensive experiments demonstrate the effectiveness and generality of our approach compared to state-of-the-art methods. Code is available at https://github.com/Rumbling281441/SPG.

Introducing Unbiased Depth into 2D Gaussian Splatting for High‐accuracy Surface Reconstruction

Yixin Yang, Yang Zhou, Hui Huang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Recently, 2D Gaussian Splatting (2DGS) has demonstrated superior geometry reconstruction quality than the popular 3DGS by using 2D surfels to approximate thin surfaces. However, it falls short when dealing with glossy surfaces, resulting in visible holes in these areas. We find that the reflection discontinuity causes the issue. To fit the jump from diffuse to specular reflection at different viewing angles, depth bias is introduced in the optimized Gaussian primitives. To address that, we first replace the depth distortion loss in 2DGS with a novel depth convergence loss, which imposes a strong constraint on depth continuity. Then, we rectify the depth criterion in determining the actual surface, which fully accounts for all the intersecting Gaussians along the ray. Qualitative and quantitative evaluations across various datasets reveal that our method significantly improves reconstruction quality, with more complete and accurate surfaces than 2DGS. Code is available at https://github.com/XiaoXinyyx/Unbiased_Surfel.

TopoGen: Topology‐Aware 3D Generation with Persistence Points

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Topological properties play a crucial role in the analysis, reconstruction, and generation of 3D shapes. Yet, most existing research focuses primarily on geometric features, due to the lack of effective representations for topology. In this paper, we introduce TopoGen, a method that extracts both discrete and continuous topological descriptors–Betti numbers and persistence points–using persistent homology. These features provide robust characterizations of 3D shapes in terms of their topology. We incorporate them as conditional guidance in generative models for 3D shape synthesis, enabling topology-aware generation from diverse inputs such as sparse and partial point clouds, as well as sketches. Furthermore, by modifying persistence points, we can explicitly control and alter the topology of generated shapes. Experimental results demonstrate that TopoGen enhances both diversity and controllability in 3D generation by embedding global topological structure into the synthesis process.

Text‐Guided Diffusion with Spectral Convolution for 3D Human Pose Estimation

Liyuan Shi, Suping Wu, Sheng Yang, Weibin Qiu, Dong Qiang, Jiarui Zhao — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Although significant progress has been made in monocular video-based 3D human pose estimation, existing methods lack guidance from fine-grained high-level prior knowledge such as action semantics and camera viewpoints, leading to significant challenges for pose reconstruction accuracy under scenarios with severely missing visual features, i.e., complex occlusion situations. We identify that the 3D human pose estimation task fundamentally constitutes a canonical inverse problem, and propose a motion-semantics-based diffusion(MS-Diff) framework to address this issue by incorporating high-level motion semantics with spectral feature regularization to eliminate interference noise in complex scenes and improve estimation accuracy. Specifically, we design a Multimodal Diffusion Interaction (MDI) module that incorporates motion semantics including action categories and camera viewpoints into the diffusion process, establishing semantic-visual feature alignment through a cross-modal mechanism to resolve pose ambiguities and effectively handle occlusions. Additionally, we leverage a Spectral Convolutional Regularization (SCR) module that implements adaptive filtering in the frequency domain to selectively suppress noise components. Extensive experiments on large-scale public datasets Human3.6M and MPI-INF-3DHP demonstrate that our method achieves state-of-the-art performance.

FAHNet: Accurate and Robust Normal Estimation for Point Clouds via Frequency‐Aware Hierarchical Geometry

Chengwei Wang, Wenming Wu, Yue Fei, Gaofeng Zhang, Liping Zheng — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Point cloud normal estimation underpins many 3D vision and graphics applications. Precise normal estimation in regions of sharp curvature and high-frequency variation remains a major bottleneck; existing learning-based methods still struggle to isolate fine geometry details under noise and uneven sampling. We present FAHNet, a novel frequency-aware hierarchical network that precisely tackles those challenges. Our Frequency-Aware Hierarchical Geometry (FAHG) feature extraction module selectively amplifies and merges cross-scale cues, ensuring that both fine-grained local features and sharp structures are faithfully represented. Crucially, a dedicated Frequency-Aware geometry enhancement (FA) branch intensifies sensitivity to abrupt normal transitions and sharp features, preventing the common over-smoothing limitation. Extensive experiments on synthetic benchmarks (PCPNet, FamousShape) and real-world scans (SceneNN) demonstrate that FAHNet outperforms state-of-the-art approaches in normal estimation accuracy. Ablation studies further quantify the contribution of each component, and downstream surface reconstruction results validate the practical impact of our design.

Multimodal 3D Few‐Shot Classification via Gaussian Mixture Discriminant Analysis

Yiqi Wu, Huachao Wu, Ronglei Hu, Yilin Chen, Dejun Zhang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

While pre-trained 3D vision-language models are becoming increasingly available, there remains a lack of frameworks that can effectively harness their capabilities for few-shot classification. In this work, we propose PointGMDA, a training-free framework that combines Gaussian Mixture Models (GMMs) with Gaussian Discriminant Analysis (GDA) to perform robust classification using only a few labeled point cloud samples. Our method estimates GMM parameters per class from support data and computes mixture-weighted prototypes, which are then used in GDA with a shared covariance matrix to construct decision boundaries. This formulation allows us to model intra-class variability more expressively than traditional single-prototype approaches, while maintaining analytical tractability. To incorporate semantic priors, we integrate CLIP-style textual prompts and fuse predictions from geometric and textual modalities through a hybrid scoring strategy. We further introduce PointGMDA-T, a lightweight attention-guided refinement module that learns residuals for fast feature adaptation, improving robustness under distribution shift. Extensive experiments on ModelNet40 and ScanObjectNN demonstrate that PointGMDA outperforms strong baselines across a variety of few-shot settings, with consistent gains under both training-free and fine-tuned conditions. These results highlight the effectiveness and generality of our probabilistic modeling and multimodal adaptation framework. Our code is publicly available at https://github.com/djzgroup/PointGMDA.

LucidFusion: Reconstructing 3D Gaussians with Arbitrary Unposed Images

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Recent large reconstruction models have made notable progress in generating high-quality 3D objects from single images. However, current reconstruction methods often rely on explicit camera pose estimation or fixed viewpoints, restricting their flexibility and practical applicability. We reformulate 3D reconstruction as image-to-image translation and introduce the Relative Coordinate Map (RCM), which aligns multiple unposed images to a “main” view without pose estimation. While RCM simplifies the process, its lack of global 3D supervision can yield noisy outputs. To address this, we propose Relative Coordinate Gaussians (RCG) as an extension to RCM, which treats each pixel's coordinates as a Gaussian center and employs differentiable rasterization for consistent geometry and pose recovery. Our LucidFusion framework handles an arbitrary number of unposed inputs, producing robust 3D reconstructions within seconds and paving the way for more flexible, pose-free 3D pipelines.

Projective Displacement Mapping for Ray Traced Editable Surfaces

Rama Hoetzlein — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Displacement mapping is an important tool for modeling detailed geometric features. We explore the problem of authoring complex surfaces while ray tracing interactively. Current techniques for ray tracing displaced surfaces rely on acceleration structures that require dynamic rebuilding when edited. These techniques are typically used for massive static scenes or the compression of detailed source assets. Our interest lies in modeling and look development of artistic features with real-time ray tracing. We introduce projective displacement mapping as a direct sampling method combined with a hardware BVH. Quality and performance are improved over existing methods with smoothed displaced normals, thin feature sampling, tight prism bounds and ray bi-linear patch intersections.

FlatCAD: Fast Curvature Regularization of Neural SDFs for CAD Models

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Neural signed-distance fields (SDFs) are a versatile backbone for neural geometry representation, but enforcing CAD-style developability usually requires Gaussian-curvature penalties with full Hessian evaluation and second-order differentiation, which are costly in memory and time. We introduce an off-diagonal Weingarten loss that regularizes only the mixed shape operator term that represents the gap between principal curvatures and flattens the surface. We present two variants: a finite-difference version using six SDF evaluations plus one gradient, and an auto-diff version using a single Hessian-vector product. Both converge to the exact mixed term and preserve the intended geometric properties without assembling the full Hessian. On the ABC benchmarks the losses match or exceed Hessian-based baselines while cutting GPU memory and training time by roughly a factor of two. The method is drop-in and framework-agnostic, enabling scalable curvature-aware SDF learning for engineering-grade shape reconstruction. Our code is available at https://flatcad.github.io/.

Hybrid Sparse Transformer and Feature Alignment for Efficient Image Completion

L. Chen, H. Sun — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

In this paper, we propose an efficient single-stage hybrid architecture for image completion. Existing transformer-based image completion methods often struggle with accurate content restoration, largely due to their ineffective modeling of corrupted channel information and the attention noise introduced by softmax-based mechanisms, which results in blurry textures and distorted structures. Additionally, these methods frequently fail to maintain texture consistency, either relying on imprecise mask sampling or incurring substantial computational costs from complex similarity calculations. To address these limitations, we present two key contributions: a Hybrid Sparse Self-Attention (HSA) module and a Feature Alignment Module (FAM). The HSA module enhances structural recovery by decoupling spatial and channel attention with sparse activation, while the FAM enforces texture consistency by aligning encoder and decoder features via a mask-free, energy-gated mechanism without additional inference cost. Our method achieves state-of-the-art image completion results with the fastest inference speed among single-stage networks, as measured by PSNR, SSIM, FID, and LPIPS on CelebA-HQ, Places2, and Paris datasets.

FlowCapX: Physics‐Grounded Flow Capture with Long‐Term Consistency

N. Tao, L. Zhang, X. Ni, M. Chu, B. Chen — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

We present FlowCapX, a physics-enhanced framework for flow reconstruction from sparse video inputs, addressing the challenge of jointly optimizing complex physical constraints and sparse observational data over long time horizons. Existing methods often struggle to capture turbulent motion while maintaining physical consistency, limiting reconstruction quality and downstream tasks. Focusing on velocity inference, our approach introduces a hybrid framework that strategically separates representation and supervision across spatial scales. At the coarse level, we resolve sparse-view ambiguities via a novel optimization strategy that aligns long-term observation with physics-grounded velocity fields. By emphasizing vorticity-based physical constraints, our method enhances physical fidelity and improves optimization stability. At the fine level, we prioritize observational fidelity to preserve critical turbulent structures. Extensive experiments demonstrate state-of-the-art velocity reconstruction, enabling velocity-aware downstream tasks, e.g., accurate flow analysis, scene augmentation with tracer visualization and re-simulation. Our implementation is released at ://github.com/taoningxiao/FlowCapX.git.

Single‐Line Drawing Vectorization

Tanguy Magne, Olga Sorkine‐Hornung — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Vectorizing line drawings is a repetitive, yet necessary task that professional creatives must perform to obtain an easily editable and scalable digital representation of a raster sketch. State-of-the-art automatic methods in this domain can create series of curves that closely fit the appearance of the drawing. However, they often neglect the line parameterization. Thus, their vector representation cannot be edited naturally by following the drawing order. We present a novel method for single-line drawing vectorization that addresses this issue. Single-line drawings consist of a single stroke, where the line can intersect itself multiple times, making the drawing order non-trivial to recover. Our method fits a single parametric curve, represented as a Bézier spline, to approximate the stroke in the input raster image. To this end, we produce a graph representation of the input and employ geometric priors and a specially trained neural network to correctly capture and classify curve intersections and their traversal configuration. Our method is easily extended to drawings containing multiple strokes while preserving their integrity and order. We compare our vectorized results with the work of several artists, showing that our stroke order is similar to the one artists employ naturally. Our vectorization method achieves state-of-the-art results in terms of similarity with the original drawing and quality of the vectorization on a benchmark of single-line drawings. Our method's results can be refined interactively, making it easy to integrate into professional workflows. Our code and results are available at https://github.com/tanguymagne/SLD-Vectorization.

RT‐HDIST: Ray‐Tracing Core‐based Hausdorff Distance Computation

Young Woo Kim, Jaehong Lee, Duksu Kim — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

The Hausdorff distance is a fundamental metric with widespread applications across various fields. However, its computation remains computationally expensive, especially for large-scale datasets. This work targets exact point-to-point Hausdorff distance on point sets. In this work, we present RT-HDIST, the first Hausdorff distance algorithm accelerated by ray-tracing cores (RT-cores). By reformulating the Hausdorff distance problem as a series of nearest-neighbor searches and introducing a novel quantized voxel-index space, RT-HDIST achieves significant reductions in computational overhead while maintaining exact results. Extensive benchmarks demonstrate up to a two-order-of-magnitude speedup over prior state-of-the-art methods, underscoring RT-HDIST's potential for real-time and large-scale applications.

High‐Performance Elliptical Cone Tracing

U. Emre, A. Kanak, S. Steinberg — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

In this work, we discuss elliptical cone traversal in scenes that employ typical triangular meshes. We derive accurate and numerically-stable intersection tests for an elliptical conic frustum with an AABB, plane, edge and a triangle, and analyze the performance of elliptical cone tracing when using different acceleration data structures: SAH-based K-d trees, BVHs as well as a modern 8-wide BVH variant adapted for cone tracing, and compare with ray tracing. In addition, several cone traversal algorithms are analyzed, and we develop novel heuristics and optimizations that give better performance than previous traversal approaches. The results highlight the difference in performance characteristics between rays and cones, and serve to guide the design of acceleration data structures for applications that employ cone tracing.

GNF: Gaussian Neural Fields for Multidimensional Signal Representation and Reconstruction

Abelaziz Bouzidi, Hamid Laga, Hazem Wannous, Ferdous Sohel — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Neural fields have emerged as a powerful framework for representing continuous multidimensional signals such as images and videos, 3D and 4D objects and scenes, and radiance fields. While efficient, achieving high-quality representation requires the use of wide and deep neural networks. These, however, are slow to train and evaluate. Although several acceleration techniques have been proposed, they either trade memory for faster training and/or inference, rely on thousands of fitted primitives with considerable optimization time, or compromise the smooth, continuous nature of neural fields. In this paper, we introduce Gaussian Neural Fields (GNF), a novel compact neural decoder that maps learned feature grids into continuous non-linear signals, such as RGB images, Signed Distance Functions (SDFs), and radiance fields, using a single compact layer of Gaussian kernels defined in a high-dimensional feature space. Our key observation is that neurons in traditional MLPs perform simple computations, usually a dot product followed by an activation function, necessitating wide and deep MLPs or high-resolution feature grids to model complex functions. In this paper, we show that replacing MLP-based decoders with Gaussian kernels whose centers are learned features yields highly accurate representations of 2D (RGB), 3D (geometry), and 5D (radiance fields) signals with just a single layer of such kernels. This representation is highly parallelizable, operates on low-resolution grids, and trains in under 15 seconds for 3D geometry and under 11 minutes for view synthesis. GNF matches the accuracy of deep MLP-based decoders with far fewer parameters and significantly higher inference throughput. The source code is publicly available at https://grbfnet.github.io/.

Procedural Multiscale Geometry Modeling using Implicit Surfaces

Bojja Venu, Adam Bosak, Juan Raúl Padrón‐Griffe — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Materials exhibit geometric structures across mesoscopic to microscopic scales, influencing macroscale properties such as appearance, mechanical strength, and thermal behavior. Capturing and modeling these multiscale structures is challenging but essential for computer graphics, engineering, and materials science. We present a framework inspired by hypertexture methods, using implicit surfaces and sphere tracing to synthesize multiscale structures on the fly without precomputation. This framework models volumetric materials with particulate, fibrous, porous, and laminar structures, allowing control over size, shape, density, distribution, and orientation. We enhance structural diversity by superimposing implicit periodic functions while improving computational efficiency. The framework also supports spatially varying particulate media, particle agglomeration, and piling on convex and concave structures, such as rock formations (mesoscale), without explicit simulation. We demonstrate its potential in the appearance modeling of volumetric materials and investigate how spatially varying properties affect the perceived macroscale appearance. As a proof of concept, we show that microstructures created by our framework can be reconstructed from image and distance values defined by implicit surfaces, using both first-order and gradient-free optimization methods.

Self‐Supervised Humidity‐Controllable Garment Simulation via Capillary Bridge Modeling

M. Shi, X. Wang, J. Zhang, L. Gao, D. Zhu, H. Zhang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Simulating wet clothing remains a significant challenge due to the complex physical interactions between moist fabric and the human body, compounded by the lack of dedicated datasets for training data-driven models. Existing self-supervised approaches struggle to capture moisture-induced dynamics such as skin adhesion, anisotropic surface resistance, and non-linear wrinkling, leading to limited accuracy and efficiency. To address this, we present SHGS, a novel self-supervised framework for humidity-controllable clothing simulation grounded in the physical modeling of capillary bridges that form between fabric and skin. We abstract the forces induced by wetness into two physically motivated components: a normal adhesive force derived from Laplace pressure and a tangential shear-resistance force that opposes relative motion along the fabric surface. By formulating these forces as potential energy for conservative effects and as mechanical work for non-conservative effects, we construct a physics-consistent wetness loss. This enables self-supervised training without requiring labeled data of wet clothing. Our humidity-sensitive dynamics are driven by a multi-layer graph neural network, which facilitates a smooth and physically realistic transition between different moisture levels. This architecture decouples the garment's dynamics in wet and dry states through a local weight interpolation mechanism, adjusting the fabric's behavior in response to varying humidity conditions. Experiments demonstrate that SHGS outperforms existing methods in both visual fidelity and computational efficiency, marking a significant advancement in realistic wet-cloth simulation.

Computational Design of Body‐Supporting Assemblies

Yixuan He, Rulin Chen, Bailin Deng, Peng Song — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

A body-supporting assembly is an assembly of parts that physically supports a human body during activities like sitting, lying, or leaning. A body-supporting assembly has a complex global shape to support a specific human body posture, yet each component part has a relatively simple geometry to facilitate fabrication, storage, and maintenance. In this paper, we aim to model and design a personalized body-supporting assembly that fits a given human body posture, aiming to make the assembly comfortable to use. We choose to model a body-supporting assembly from scratch to offer high flexibility for fitting a given body posture, which however makes it challenging to determine the assembly's topology and geometry. To address this problem, we classify parts in the assembly into two categories according the functionality: supporting parts for fitting different portions of the body and connecting parts for connecting all the supporting parts to form a stable structure. We also propose a geometric representation of supporting parts such that they can have a variety of shapes controlled by a few parameters. Given a body posture as input, we present a computational approach for designing a body-supporting assembly that fits the posture, in which the supporting parts are initialized and optimized to minimize a discomfort measure and then the connecting parts are generated using a procedural approach. We demonstrate the effectiveness of our approach by designing body-supporting assemblies that accommodate to a variety of body postures and 3D printing two of them for physical validation.

DAATSim: Depth‐Aware Atmospheric Turbulence Simulation for Fast Image Rendering

Ripon Kumar Saha, Yufan Zhang, Jinwei Ye, Suren Jayasuriya — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Simulating the effects of atmospheric turbulence for imaging systems operating over long distances is a significant challenge for optical and computer graphics models. Physically-based ray tracing over kilometers of distance is difficult due to the need to define a spatio-temporal volume of varying refractive index. Even if such a volume can be defined, Monte Carlo rendering approximations for light refraction through the environment would not yield real-time solutions needed for video game engines or online dataset augmentation for machine learning. While existing simulators based on procedurally-generated noise or textures have been proposed in these settings, these simulators often neglect the significant impact of scene depth, leading to unrealistic degradations for scenes with substantial foreground-background separation. This paper introduces a novel, physically-based atmospheric turbulence simulator that explicitly models depth-dependent effects while rendering frames at interactive/near real-time (> 10 FPS) rates for image resolutions up to 1024 × 1024 (real-time 35 FPS at 256× 256 resolution with depth or 512×512 at 33 FPS without depth). Our hybrid approach combines spatially-varying wavefront aberrations using Zernike polynomials with pixel-wise depth modulation of both blur (via Point Spread Function interpolation) and geometric distortion or tilt. Our approach includes a novel fusion technique that integrates complementary strengths of leading monocular depth estimators to generate metrically accurate depth maps with enhanced edge fidelity. DAATSim is implemented efficiently on GPUs using Py-Torch incorporating optimizations like mixed-precision computation and caching to achieve efficient performance. We present quantitative and qualitative validation demonstrating the simulator's physical plausibility for generating turbulent video. DAAT-Sim is made publicly available and open-source to the community: https://github.com/Riponcs/DAATSim.

TensoIS: A Step Towards Feed‐Forward Tensorial Inverse Subsurface Scattering for Perlin Distributed Heterogeneous Media

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Estimating scattering parameters of heterogeneous media from images is a severely under-constrained and challenging problem. Most of the existing approaches model BSSRDF either through an analysis-by-synthesis approach, approximating complex path integrals, or using differentiable volume rendering techniques to account for heterogeneity. However, only a few studies have applied learning-based methods to estimate subsurface scattering parameters, but they assume homogeneous media. Interestingly, no specific distribution is known to us that can explicitly model the heterogeneous scattering parameters in the real world. Notably, procedural noise models such as Perlin and Fractal Perlin noise have been effective in representing intricate heterogeneities of natural, organic, and inorganic surfaces. Leveraging this, we first create HeteroSynth, a synthetic dataset comprising photorealistic images of heterogeneous media whose scattering parameters are modeled using Fractal Perlin noise. Furthermore, we propose Tensorial Inverse Scattering (TensoIS), a learning-based feed-forward framework to estimate these Perlin-distributed heterogeneous scattering parameters from sparse multi-view image observations. Instead of directly predicting the 3D scattering parameter volume, TensoIS uses learnable low-rank tensor components to represent the scattering volume. We evaluate TensoIS on unseen heterogeneous variations over shapes from the HeteroSynth test set, smoke and cloud geometries obtained from open-source realistic volumetric simulations, and some real-world samples to establish its effectiveness for inverse scattering. Overall, this study is an attempt to explore Perlin noise distribution, given the lack of any such well-defined distribution in literature, to potentially model real-world heterogeneous scattering in a feed-forward manner.

Project Page: https://yashbachwana.github.io/TensoIS/

Automatic Reconstruction of Woven Cloth from a Single Close‐up Image

C. Wu, A. Khattar, J. Zhu, S. Pettifer, L. Yan, Z. Montazeri — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Digital replication of woven fabrics presents significant challenges across a variety of sectors, from online retail to entertainment industries. To address this, we introduce an inverse rendering pipeline designed to estimate pattern, geometry, and appearance parameters of woven fabrics given a single close-up image as input. Our work is capable of simultaneously optimizing both discrete and continuous parameters without manual interventions. It outputs a wide array of parameters, encompassing discrete elements like weave patterns, ply and fiber number, using Simulated Annealing. It also recovers continuous parameters such as reflection and transmission components, aligning them with the target appearance through differentiable rendering. For irregularities caused by deformation and flyaways, we use 2D Gaussians to approximate them as a post-processing step. Our work does not pursue perfect matching of all fine details, it targets an automatic and end-to-end reconstruction pipeline that is robust to slight camera rotations and room light conditions within an acceptable time (15 minutes on CPU), unlike previous works which are either expensive, require manual intervention, assume given pattern, geometry or appearance, or strictly control camera and light conditions.

GS‐Share: Enabling High‐fidelity Map Sharing with Incremental Gaussian Splatting

Xinran Zhang, Hanqi Zhu, Yifan Duan, Yanyong Zhang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Constructing and sharing 3D maps is essential for many applications, including autonomous driving and augmented reality. Recently, 3D Gaussian splatting has emerged as a promising approach for accurate 3D reconstruction. However, a practical map-sharing system that features high-fidelity, continuous updates, and network efficiency remains elusive. To address these challenges, we introduce GS-Share, a photorealistic map-sharing system with a compact representation. The core of GS-Share includes anchor-based global map construction, virtual-image-based map enhancement, and incremental map update. We evaluate GS-Share against state-of-the-art methods, demonstrating that our system achieves higher fidelity, particularly for extrapolated views, with improvements of 11%, 22%, and 74% in PSNR, LPIPS, and Depth L1, respectively. Furthermore, GS-Share is significantly more compact, reducing map transmission overhead by 36%.

Joint Deblurring and 3D Reconstruction for Macrophotography

Yifan Zhao, Liangchen Li, Yuqi Zhou, Kai Wang, Yan Liang, Juyong Zhang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Macro lens has the advantages of high resolution and large magnification, and 3D modeling of small and detailed objects can provide richer information. However, defocus blur in macrophotography is a long-standing problem that heavily hinders the clear imaging of the captured objects and high-quality 3D reconstruction of them. Traditional image deblurring methods require a large number of images and annotations, and there is currently no multi-view 3D reconstruction method for macrophotography. In this work, we propose a joint deblurring and 3D reconstruction method for macrophotography. Starting from multi-view blurry images captured, we jointly optimize the clear 3D model of the object and the defocus blur kernel of each pixel. The entire framework adopts a differentiable rendering method to self-supervise the optimization of the 3D model and the defocus blur kernel. Extensive experiments show that from a small number of multi-view images, our proposed method can not only achieve high-quality image deblurring but also recover high-fidelity 3D appearance.

BoxFusion: Reconstruction‐Free Open‐Vocabulary 3D Object Detection via Real‐Time Multi‐View Box Fusion

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Open-vocabulary 3D object detection has gained significant interest due to its critical applications in autonomous driving and embodied AI. Existing detection methods, whether offline or online, typically rely on dense point cloud reconstruction, which imposes substantial computational overhead and memory constraints, hindering real-time deployment in downstream tasks. To address this, we propose a novel reconstruction-free online framework tailored for memory-efficient and real-time 3D detection. Specifically, given streaming posed RGB-D video input, we leverage Cubify Anything as a pre-trained visual foundation model (VFM) for single-view 3D object detection, coupled with CLIP to capture open-vocabulary semantics of detected objects. To fuse all detected bounding boxes across different views into a unified one, we employ an association module for correspondences of multi-views and an optimization module to fuse the 3D bounding boxes of the same instance. The association module utilizes 3D Non-Maximum Suppression (NMS) and a box correspondence matching module. The optimization module uses an IoU-guided efficient random optimization technique based on particle filtering to enforce multi-view consistency of the 3D bounding boxes while minimizing computational complexity. Extensive experiments on CA-1M and ScanNetV2 datasets demonstrate that our method achieves state-of-the-art performance among online methods. Benefiting from this novel reconstruction-free paradigm for 3D object detection, our method exhibits great generalization abilities in various scenarios, enabling real-time perception even in environments exceeding 1000 square meters.

G‐SplatGAN: Disentangled 3D Gaussian Generation for Complex Shapes via Multi‐Scale Patch Discriminators

Jiaqi Li, Haochuan Dang, Zhi Zhou, Junke Zhu, Zhangjin Huang — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Generating 3D objects with complex topologies from monocular images remains a challenge in computer graphics, due to the difficulty of modeling varying 3D shapes with disentangled, steerable geometry and visual attributes. While NeRF-based methods suffer from slow volumetric rendering and limited structural controllability. Recent advances in 3D Gaussian Splatting provide a more efficient alternative and its generative modeling with separate control over structure and appearance remains underexplored. In this paper, we propose G-SplatGAN, a novel 3D-aware generation framework that combines the rendering efficiency of 3D Gaussian Splatting with disentangled latent modeling. Starting from a shared Gaussian template, our method uses dual modulation branches to modulate geometry and appearance from independent latent codes, enabling precise shape manipulation and controllable generation. We adopt a progressive adversarial training scheme with multi-scale and patch-based discriminators to capture both global structure and local detail. Our model requires no 3D supervision and is trained on monocular images with known camera poses, reducing data reliance while supporting real image inversion through a geometry-aware encoder. Experiments show that G-SplatGAN achieves superior performance in rendering speed, controllability and image fidelity, offering a compelling solution for controllable 3D generation using Gaussian representations.

Accelerating Signed Distance Functions

Pierre Hubert‐Brierre, Eric Guérin, Adrien Peytavie, Eric Galin — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Processing and particularly visualizing implicit surfaces remains computationally intensive when dealing with complex objects built from construction trees. We introduce optimization nodes to reduce the computational cost of the field function evaluation for hierarchical construction trees, while preserving the Lipschitz or conservative properties of the function. Our goal is to propose acceleration nodes directly embedded in the construction tree, and avoid external, accompanying data-structures such as octrees. We present proxy and continuous level of detail nodes to reduce the overall evaluation cost, along with a normal warping technique that enhances surface details with negligible computational overhead. Our approach is compatible with existing algorithms that aim at reducing the number of function calls. We validate our methods by computing timings as well as the average cost for traversing the tree and evaluating the signed distance field at a given point in space. Our method speeds-up signed distance field evaluation by up to three orders or magnitude, and applies both to ray-surface intersection computation in Sphere Tracing applications, and to polygonization algorithms.

Using Saliency for Semantic Image Abstractions in Robotic Painting

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

We present an adaptive, semantics-based abstraction approach that balances aesthetic quality and structural coherence within the practical constraints of robotic painting. We apply panoptic segmentation with color-based over-segmentation to partition images into meaningful regions aligned with semantic objects, while providing flexible abstraction levels. Automatic parameter selection for region merging is enabled by semantic saliency maps, derived from Out-of-Distribution segmentation techniques in combination with machine learning methods for feature detection. This preserves the boundaries of salient objects while simplifying less prominent regions. A graph-based community detection step further refines the abstraction by grouping regions according to local connectivity and semantic coherence. The runtime of our method outperforms optimization-based image vectorization methods, enabling the efficient generation of multiple abstraction levels that can serve as hierarchical layers for robotic painting. We demonstrate the quality of our method by showing abstraction results, robotic paintings with the e-David robot, and a comparison to other abstraction methods.

EmoDiffGes: Emotion‐Aware Co‐Speech Holistic Gesture Generation with Progressive Synergistic Diffusion

Xinru Li, Jingzhong Lin, Bohao Zhang, Yuanyuan Qi, Changbo Wang, Gaoqi He — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Co-speech gesture generation, driven by emotional expression and synergistic bodily movements, is essential for applications such as virtual avatars and human-robot interaction. Existing co-speech gesture generation methods face two fundamental limitations: (1) producing inexpressive gestures due to ignoring the temporal evolution of emotion; (2) generating incoherent and unnatural motions as a result of either holistic body oversimplification or independent part modeling. To address the above limitations, we propose EmoDiffGes, a diffusion-based framework grounded in embodied emotion theory, unifying dynamic emotion conditioning and part-aware synergistic modeling. Specifically, a Dynamic Emotion-Alignment Module (DEAM) is first applied to extract dynamic emotional cues and inject emotion guidance into the generation process. Then, a Progressive Synergistic Gesture Generator (PSGG) iteratively refines region-specific latent codes while maintaining full-body coordination, leveraging a Body Region Prior for part-specific encoding and Progressive Inter-Region Synergistic Flow for global motion coherence. Extensive experiments validate the effectiveness of our methods, showcasing the potential for generating expressive, coordinated, and emotionally grounded human gestures.

LTC‐IR: Multiview Edge‐Aware Inverse Rendering with Linearly Transformed Cosines

Dabeen Park, Junsuh Park, Jooeun Son, Seungyoung Lee, Joo‐Ho Lee — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Decomposing environmental lighting and materials is challenging as they are tightly intertwined and integrated over the hemisphere. In order to precisely decouple them, the lighting representation must represent general image features such as object boundaries or texture contrast, called light edges, which are often neglected in the existing inverse rendering methods. In this paper, we propose an inverse rendering method that efficiently captures light edges. We introduce a triangle mesh-based light representation that can express light edges by aligning triangle edges with light edges. We exploit the linearly transformed cosines as BRDF approximations to efficiently compute environmental lighting with our light representation. Our edge-aware inverse rendering precisely decouples distributions of reflectance and lighting through differentiable rendering by jointly reconstructing light edges and estimating the BRDF parameters. Our experiments, including various material/scene settings and ablation studies, demonstrate the reconstruction performance and computational efficiency of our method.

PaMO: Parallel Mesh Optimization for Intersection‐Free Low‐Poly Modeling on the GPU

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Reducing the triangle count in complex 3D models is a basic geometry preprocessing step in graphics pipelines such as efficient rendering and interactive editing. However, most existing mesh simplification methods exhibit a few issues. Firstly, they often lead to self-intersections during decimation, a major issue for applications such as 3D printing and soft-body simulation. Second, to perform simplification on a mesh in the wild, one would first need to perform re-meshing, which often suffers from surface shifts and losses of sharp features. Finally, existing re-meshing and simplification methods can take minutes when processing large-scale meshes, limiting their applications in practice. To address the challenges, we introduce a novel GPU-based mesh optimization approach containing three key components: (1) a parallel re-meshing algorithm to turn meshes in the wild into watertight, manifold, and intersection-free ones, and reduce the prevalence of poorly shaped triangles; (2) a robust parallel simplification algorithm with intersection-free guarantees; (3) an optimization-based safe projection algorithm to realign the simplified mesh with the input, eliminating the surface shift introduced by re-meshing and recovering the original sharp features. The algorithm demonstrates remarkable efficiency, simplifying a 2-million-face mesh to 20k triangles in 3 seconds on RTX4090. We evaluated the approach on the Thingi10K dataset and showcased its exceptional performance in geometry preservation and speed. https://seonghunn.github.io/pamo/

Preconditioned Deformation Grids

Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Dynamic surface reconstruction of objects from point cloud sequences is a challenging field in computer graphics. Existing approaches either require multiple regularization terms or extensive training data which, however, lead to compromises in reconstruction accuracy as well as over-smoothing or poor generalization to unseen objects and motions. To address these limitations, we introduce Preconditioned Deformation Grids, a novel technique for estimating coherent deformation fields directly from unstructured point cloud sequences without requiring or forming explicit correspondences. Key to our approach is the use of multi-resolution voxel grids that capture the overall motion at varying spatial scales, enabling a more flexible deformation representation. In conjunction with incorporating grid-based Sobolev preconditioning into gradient-based optimization, we show that applying a Chamfer loss between the input point clouds as well as to an evolving template mesh is sufficient to obtain accurate deformations. To ensure temporal consistency along the object surface, we include a weak isometry loss on mesh edges which complements the main objective without constraining deformation fidelity. Extensive evaluations demonstrate that our method achieves superior results, particularly for long sequences, compared to state-of-the-art techniques.

WaterGS: Physically‐Based Imaging in Gaussian Splatting for Underwater Scene Reconstruction

S. Q. Wang, W. B. Wu, M. Shi, Z. X. Li, Q. Wang, D. M. Zhu — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Reconstructing underwater object geometry from multi-view images is a long-standing challenge in computer graphics, primarily due to image degradation caused by underwater scattering, blur, and color shift. These degradations severely impair feature extraction and multi-view consistency. Existing methods typically rely on pre-trained image enhancement models as a preprocessing step, but often struggle with robustness under varying water conditions. To overcome these limitations, we propose WaterGS, a novel framework for underwater surface reconstruction that jointly recovers accurate 3D geometry and restores true object colors. The core of our approach lies in introducing a Physically-Based imaging model into the rendering process of 2D Gaussian Splatting. This enables accurate separation of true object colors from water-induced distortions, thereby facilitating more robust photometric alignment and denser geometric reconstruction across views. Building upon this improved photometric consistency, we further introduce a Gaussian bundle adjustment scheme guided by our physical model to jointly optimize camera poses and geometry, enhancing reconstruction accuracy. Extensive experiments on synthetic and real-world datasets show that WaterGS achieves robust, high-fidelity reconstruction directly from raw underwater images, outperforming prior approaches in both geometric accuracy and visual consistency.

Gaussians on their Way: Wasserstein‐Constrained 4D Gaussian Splatting with State‐Space Modeling

J. Deng, P. Shi, Y. Luo, Q. Li — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Dynamic scene rendering has taken a leap forward with the rise of 4D Gaussian Splatting, but there is still one elusive challenge: how to make 3D Gaussians move through time as naturally as they would in the real world, all while keeping the motion smooth and consistent. In this paper, we present an approach that blends state-space modeling with Wasserstein geometry, enabling a more fluid and coherent representation of dynamic scenes. We introduce a State Consistency Filter that merges prior predictions with the current observations, enabling Gaussians to maintain coherent trajectories over time. We also employ Wasserstein Consistency Constraint to ensure smooth, consistent updates of Gaussian parameters, reducing motion artifacts. Lastly, we leverage Wasserstein geometry to capture both translational motion and shape deformations, creating a more geometrically consistent model for dynamic scenes. Our approach models the evolution of Gaussians along geodesics on the manifold of Gaussian distributions, achieving smoother, more realistic motion and stronger temporal coherence. Experimental results show consistent improvements in rendering quality and efficiency.

(see https://www.acm.org/publications/class-2012)

Real‐Time Per‐Garment Virtual Try‐On with Temporal Consistency for Loose‐Fitting Garments

Zaiqiang Wu, I‐Chao Shen, Takeo Igarashi — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Per-garment virtual try-on methods collect garment-specific datasets and train networks tailored to each garment to achieve superior results. However, these approaches often struggle with loose-fitting garments due to two key limitations: (1) They rely on human body semantic maps to align garments with the body, but these maps become unreliable when body contours are obscured by loose-fitting garments, resulting in degraded outcomes; (2) They train garment synthesis networks on a per-frame basis without utilizing temporal information, leading to noticeable jittering artifacts. To address the first limitation, we propose a two-stage approach for robust semantic map estimation. First, we extract a garment-invariant representation from the raw input image. This representation is then passed through an auxiliary network to estimate the semantic map. This enhances the robustness of semantic map estimation under loose-fitting garments during garment-specific dataset generation. To address the second limitation, we introduce a recurrent garment synthesis framework that incorporates temporal dependencies to improve frame-to-frame coherence while maintaining real-time performance. We conducted qualitative and quantitative evaluations to demonstrate that our method outperforms existing approaches in both image quality and temporal coherence. Ablation studies further validate the effectiveness of the garment-invariant representation and the recurrent synthesis framework.

LayoutRectifier: An Optimization‐based Post‐processing for Graphic Design Layout Generation

I‐Chao Shen, Ariel Shamir, Takeo Igarashi — Thu, 16 Oct 2025 05:17:37 -0700

Abstract

Recent deep learning methods can generate diverse graphic design layouts efficiently. However, these methods often create layouts with flaws, such as misalignment, unwanted overlaps, and unsatisfied containment. To tackle this issue, we propose an optimization-based method called LayoutRectifier, which gracefully rectifies auto-generated graphic design layouts to reduce these flaws while minimizing deviation from the generated layout. The core of our method is a two-stage optimization. First, we utilize grid systems, which professional designers commonly use to organize elements, to mitigate misalignments through discrete search. Second, we introduce a novel box containment function designed to adjust the positions and sizes of the layout elements, preventing unwanted overlapping and promoting desired containment. We evaluate our method on content-agnostic and content-aware layout generation tasks and achieve better-quality layouts that are more suitable for downstream graphic design tasks. Our method complements learning-based layout generation methods and does not require additional training.

Issue Information

Thu, 16 Oct 2025 05:17:37 -0700

Computer Graphics Forum, Volume 44, Issue 7, Page i-xxi, October 2025.