method,arxiv_id,venue,year,category,core_innovation,code_url
3DGEER,,ICLR 2026,2026,Foundation,Exact ray-Gaussian integration replacing splatting approximation for fisheye/equirectangular cameras with closed-form 3D mean integral,https://github.com/boschresearch/3dgeer
3DSGS,2605.18334,arXiv 2026,2026,Foundation,General 3D Skew Gaussian primitives with re-derived CUDA rasterization for asymmetric shape modeling,
Argus,,arXiv,2026,Foundation,Image-derived LiDAR-level pose constraints for 3DGS without LiDAR; consumer-grade 3DGS applications (如视),
Cookbook of 3D Vision,2606.04291,arXiv 2026,2026,Foundation,"Data-centric taxonomy of 3D vision: point clouds, meshes, voxels, 3D Gaussians; dataset design",
DP-GES,2605.25345,arXiv,2026,Foundation,Depth peeling for high-fidelity Gaussian-enhanced surfel rendering; resolves transparency ordering,
E3DGS,2607.15536,arXiv 2026,2026,Foundation,Unified geometric-photometric equivariance for 3DGS via Color-as-Geometry embedding; builds SE(3)-equivariant architectures on Gaussian primitives by treating color as geometry,
Fast 3D Foundation Model Initialized GS,,arXiv,2026,Foundation,3D foundation model replacing SfM for pose/point cloud initialization; joint pose-Gaussian optimization; ~3 min/scene training,
MetaPoint,2606.05031,arXiv,2026,Foundation,Precise spatial control in agentic visual generation; represents continuous 2D coordinate as special token,
PRISM3D,,arXiv,2026,Foundation,First framework bootstrapping 3DGS from extreme motion blur; MCMC probabilistic densification + continuous Bezier trajectory physics modeling,
PureLight,2606.04319,arXiv,2026,Foundation,Learning complex luminaires with light tracing; forward light tracing for efficient difficult light paths,
SalientGS,2607.11285,arXiv,2026,Foundation,Unified SfM-to-3DGS pipeline with importance-guided MCMC Gaussian allocation; reallocates capacity from well-fit to underfit regions; 15-min end-to-end SOTA,
SAM 3D,,CVPR 2026 Honorable Mention,2026,Foundation,Foundation model for single-image 3D reconstruction; generalizable 3D prior enables diverse downstream tasks,
SNS,2605.15010,arXiv,2026,Foundation,3D Gaussian Splatting (3DGS) has emerged as a leading representation for real-time novel view synthesis and has been widely adopted in various downstream applications. The core strength of 3DGS lie...,
Softmax-GS,2604.27437,CVPR 2026 Findings,2026,Foundation,Learnable softmax-based competition replacing α-compositing,
SpatialQ,2607.26595,arXiv 2026,2026,Foundation,"3D Gaussian Splatting (3DGS) has emerged as an effective representation for novel view synthesis and 3D scene reconstruction, creating an increasing demand for reliable quality assessment",
Topo-GS,2605.17011,arXiv 2026,2026,Foundation,Topological GS for high-dimensional data embedding; ARAP prior + tangent space alignment via Procrustes; topology-aware loss,
UniTriSplat,2606.29794,arXiv 2026,2026,Foundation,"To address this limitation, we propose UniTriSplat, a unified 3DGS framework for universal cameras that reformulates Gaussian splatting on the unit sphere via HEALPix discretization",
GaussRender,2502.05040,CVPR 2025,2025,Foundation,Unified rendering pipeline for GS supporting multi-modal output (RGB/D/N/S),
UniGS,2510.12174,CVPR 2025,2025,Foundation,Unified Gaussian Splatting: single model supporting multiple rendering modes (RGB/depth/semantic),
6DGS,2410.04974,ECCV 2024,2024,Foundation,6-DoF Gaussian Splatting: explicit orientation-aware primitive with full 6D pose parameterization,https://github.com/r4dl/6dgs
DisC-GS,2405.15196,NeurIPS 2024,2024,Foundation,Progressive low-pass + discontinuity boundary detection preventing splat artifacts at edges,
GaussianImage,2403.08551,ECCV 2024,2024,Foundation,2D Gaussian image codec at 1000+ FPS,
Mip-Splatting,2311.16493,CVPR 2024 (Best Student Paper),2024,Foundation,Anti-aliased 3DGS with 3D smoothing + 2D Mip filter,https://github.com/autonomousvision/mip-splatting
Normal-GS,2410.20593,NeurIPS 2024,2024,Foundation,Normal-involved rendering: normal constraint + differentiable normal guiding splat distribution,
ODGS,2410.20686,NeurIPS 2024,2024,Foundation,Spherical projection + panoramic camera GS rasterization adaptation for 360-degree images,https://github.com/esw0116/ODGS
3DGS,2308.04079,SIGGRAPH 2023,2023,Foundation,Anisotropic 3D Gaussians with tile-based differentiable rasterization,https://repo-sam.informatik.uni-halle.de/jkortner/gaussian-splatting/
AdaGScale,2604.18980,DAC 2026,2026,Optimization,"Reducing the number of Gaussian-tile pairs is one of the most promising approaches to improve 3D Gaussian Splatting (3D-GS) rendering speed on GPUs. However, the importance difference existing amon...",
AdpSplit,2605.06876,arXiv,2026,Optimization,"Adaptive density control in 3D Gaussian Splatting (3DGS) repeatedly grows the Gaussian population through fixed-cardinality random splitting to discover useful scene structure. However, in vanilla ...",
ArtifactWorld,2604.12251,arXiv,2026,Optimization,Video generation models for 3DGS artifact restoration at scale,
AV1-3DGS,2605.14629,arXiv 2026,2026,Optimization,AV1 motion vectors for dense SfM matching with 63% training time reduction,
CAdam,2605.20872,SIGGRAPH 2026,2026,Optimization,Context-Adaptive Moment Estimation for 3DGS densification in generative distillation; first moment interference principle; 85-97% Gaussian reduction,
Compact Neural Appearance,2609.05255,arXiv 2026,2026,Optimization,"Compact neural appearance models for efficient 3DGS; shared MLP decodes latent codes, 192->28 bytes per primitive vs SH3",
ConFi-GS,2605.24964,arXiv,2026,Optimization,Confidence-guided high-frequency injection + reliability-aware densification for 3DGS super-resolution,
Denoising-GS,2605.14880,arXiv,2026,Optimization,"Recent advances in 3D Gaussian Splatting (3DGS) have achieved remarkable success in high-fidelity Novel View Synthesis (NVS), yet the optimization process inevitably introduces noisy Gaussian primi...",
DP-Splat,2607.10912,arXiv,2026,Optimization,Bayesian nonparametric complexity control via Dirichlet-process prior; 5.9-7.6x fewer components at matched quality; exact monotonicity guarantee + truncation-error bound,
DualSplat,2604.21631,CVPR 2026,2026,Optimization,Failure-to-Prior framework from reconstruction failures,
E2EGS,2603.14684,CVPR 2026,2026,Optimization,Event-to-edge pose-free 3D reconstruction,
ELoG-GS,2604.12592,arXiv,2026,Optimization,Dual-branch luminance-guided extreme low-light GS enhancement,
EnerGS,2604.26238,Under review,2026,Optimization,Energy-based soft geometric guidance from LiDAR/depth priors,
Energy-GS,,CVPR,2026,Optimization,RGB-only pose-free 3DGS joint optimization; no depth/geometry priors required; energy-based formulation,
Faster-GS,,CVPR 2026,2026,Optimization,Systematic benchmark separating engineering from algorithmic acceleration,
FastGS,2511.04283,CVPR,2026,Optimization,Multi-view consistency-based densification/pruning replacing Gaussian budget; 3.32x training acceleration,
FreeFix,2601.20857,arXiv,2026,Optimization,Fine-tuning-free diffusion guidance for extrapolated 3DGS,
GEMM-GS,2604.02120,DAC 2026,2026,Optimization,"GEMM-compatible blending for Tensor Core utilization, 1.42x speedup",https://github.com/shieldforever/GEMM-GS
HarmoGS,2605.13073,arXiv 2026,2026,Optimization,Conflict-aware gradient harmonization for in-the-wild 3DGS: semantic consistency-guided masking + dual-view gradient rotation,
Hybrid-Capture Two-View Training,2605.00052,arXiv,2026,Optimization,Two-view-per-step as dominant training lever,
Laplacian Frequency Hierarchies,2609.03334,Pacific Graphics 2026,2026,Optimization,"Laplacian frequency hierarchies for efficient 3DGS training; coarse-to-fine frequency-staged, 1.73x speedup",
Learn2Splat,2605.15760,arXiv 2026,2026,Optimization,Meta-learned optimizer for 3DGS via checkpoint buffer + optimizer rollout; extends optimization horizon without degradation; zero-shot generalization (Geiger group),
Luminance-GS,2602.18322,TPAMI 2026,2026,Optimization,View-adaptive color/lightness correction for robust NVS,
MERID-GS,2604.24053,arXiv,2026,Optimization,Retinex-based illumination/reflectance decoupling for low-light,https://github.com/YhuoyuH/MERID-GS
NRGS,2604.22439,arXiv,2026,Optimization,Neural regularization for semantic 3DGS,
PDF-GS,2604.12580,arXiv,2026,Optimization,Progressive distractor filtering for robust 3DGS,
PocketGS,2601.17354,arXiv,2026,Optimization,On-device 3DGS training for mobile with geometry-faithful priors + anisotropic seeding,
Proxy-GS,,CVPR 2026,2026,Optimization,Lightweight proxy model for 2.5x speedup with no accuracy loss,
Prune Wisely,2602.24136,arXiv 2026,2026,Optimization,Adaptive pruning + Difference-of-Gaussian primitives for compact 3DGS; up to 90% Gaussian reduction,
QuadBox,2605.04844,arXiv 2026,2026,Optimization,Geometry-aware AABB boxes for 1.85x rendering acceleration,https://github.com/Powertony102/QuadBox
Rendering-Aware Bayesian 3DGS,,arXiv,2026,Optimization,Normal-Inverse-Wishart posterior tracking for Gaussian geometry; native predictive uncertainty for interval calibration + active view selection; ~17x coverage error reduction,
ReorgGS,2605.08739,arXiv 2026,2026,Optimization,Equivalent distribution reorganization fixing parameterization degeneration; kNN anisotropic covariance re-estimation,
RouteBridge,2609.09606,arXiv 2026,2026,Optimization,Reliability-routed bidirectional distillation between NeRF and 3DGS; complementary inductive biases via adaptive routing,
Structure-Aware Densification,2604.28016,SIGGRAPH 2026,2026,Optimization,Frequency-aware anisotropic split replacing heuristic clone/split,
TangoGS,2609.31248,arXiv 2026,2026,Optimization,"Compact Gaussian Splatting across scene scales; capture-derived learning allowance from total pixels after discounting re-observing views + training-quality guided density control; matches best baseline LeGS PSNR with 48% fewer Gaussians on 13 benchmarks, SOTA on 8 large captures",
TruncGradGS,2609.03534,Pacific Graphics 2026,2026,Optimization,Improved 3DGS via truncated gradient updates; addresses gradient vanishing for robust primitive learning,
TWINGS,2605.22069,CVPR,2026,Optimization,Thin Plate Splines warp-aligned initialization for sparse-view 3DGS,
VkSplat,2605.00219,Eurographics 2026 Short Papers,2026,Optimization,"Vulkan-based 3DGS training, 3.3x faster, 33% less VRAM",https://github.com/harry7557558/vksplat
YOGO,2604.21400,arXiv,2026,Optimization,Budget-aware equilibrium for ultra-dense scene control,https://jjrcn.github.io/yogo-project-home/
GaussHDR,2503.10143,CVPR 2025,2025,Optimization,HDR-robust GS with exposure-aware Gaussian decomposition,
GS-Blur,2410.23658,CVPR 2025,2025,Optimization,Motion blur-aware GS training with blur kernel estimation for sharp reconstruction,
MarineSTD-GS,2604.23551,ACM MM 2025,2025,Optimization,Spatiotemporal underwater degradation modeling,
3DGS-Enhancer,2410.16266,NeurIPS 2024,2024,Optimization,2D diffusion priors guiding iterative 3DGS refinement for view-consistent enhancement,https://github.com/xiliu8006/3DGS-Enhancer
DC-Gaussian,2405.17705,NeurIPS 2024,2024,Optimization,Reflection separation + degradation-aware training for reflective dashcam 3DGS,https://github.com/linhanwang/DC-Gaussian
Ev-GS,2407.11343,CVPR 2024,2024,Optimization,Event camera-integrated 3DGS for high-speed and HDR scene reconstruction,
GaussianPretrain,2411.12452,CVPR 2024,2024,Optimization,Self-supervised pre-training for Gaussian initialization from multi-view features,
GaussianSR,2406.10111,CVPR 2024,2024,Optimization,2D super-resolution diffusion prior guiding 3DGS iterative refinement for high-quality rendering,
GS-PT,2409.04963,ECCV 2024,2024,Optimization,Gaussian Splatting pre-training: self-supervised representation learning for Gaussian initialization,
LE3D,2406.06216,NeurIPS 2024,2024,Optimization,Low-light to HDR linear GS encoding + denoising for Lighting Every Darkness,https://github.com/Srameo/LE3D
SplatFields,2409.11211,ECCV 2024,2024,Optimization,Implicit neural field regularization on splat features for sparse-view 3D/4D reconstruction,
WildGaussians,2407.08447,NeurIPS 2024,2024,Optimization,Joint pose + 3DGS optimization from internet photos,https://github.com/jkulhanek/wild-gaussians/
2D GauSS-MI,2609.21516,arXiv 2026,2026,Surface & Rendering,Efficient active scene reconstruction on 2DGS with online mapping and a probabilistic reliability model for view selection.,
2D-SuGaR,2605.00569,Eurographics 2026,2026,Surface & Rendering,2DGS enhanced with monocular depth/normal priors,
3DSS,2605.05876,arXiv,2026,Surface & Rendering,"We present 3D Surface Splatting (3DSS), the first differentiable surface splatting renderer for physically-based inverse rendering from multi-view images. Our central insight is that the surface se...",
AmbiSuR,2605.12494,ICML 2026,2026,Surface & Rendering,Photometric ambiguity disambiguation for accurate GS surface reconstruction,
AnyGS2Mesh,2609.03304,arXiv 2026,2026,Surface & Rendering,Feed-forward mesh reconstruction from 3DGS with arbitrary-resolution views; Gaussian-guided transformer + TSDF fusion,
BEAST3D,2606.02937,arXiv 2026,2026,Surface & Rendering,"We address these limitations with BEAST3D, a self-supervised pretraining framework that learns 3D visual representations from unlabeled, calibrated multi-view video",
CLEAR-NeRF,2605.28125,arXiv,2026,Surface & Rendering,Collinearity and local-region enhanced accurate 3D reconstruction in unbounded scenes via geometric regularization,
CoMVS-GS,2608.18413,arXiv 2026,2026,Surface & Rendering,General surface reconstruction framework that combines Multi-View Stereo with Gaussian splatting,
CubicSplat,2608.20803,ECCV 2026,2026,Surface & Rendering,Differentiable vector graphics via error-bounded forward relaxation; uniform polyline proxy replacing Bezier nearest-point solver; O(S^-2) bounded geometric error with static computation graph; 2dB PSNR improvement,
Deformable Triangle Splatting,2607.22446,ECCV 2026,2026,Surface & Rendering,Flexible primitives for real-time radiance field rendering; K control points per edge with learnable scalar displacement for non-convex shape representation; barycentric coordinate rasterization with winding number test,
DinoComplete,2605.26949,arXiv,2026,Surface & Rendering,3D shape completion with DINOv2-distilled semantic priors + state space model,
DySurface,2605.10360,arXiv,2026,Surface & Rendering,"While novel view synthesis (NVS) for dynamic scenes has seen significant progress, reconstructing temporally consistent geometric surfaces remains a challenge. Neural Radiance Fields (NeRF) and 3D ...",
Elastic Triangle Splatting,2608.29106,arXiv 2026,2026,Surface & Rendering,"While neural rendering methods such as 3D Gaussian Splatting achieve remarkable visual fidelity, traditional polygonal meshes remain the backbone of established graphics pipelines",
G2ARD-GS,2608.05704,arXiv 2026,2026,Surface & Rendering,"Geometry-guided distillation method that converts a dense Gaussian prior instantiated either as a training-free point-cloud lift or a trained GS model into a compact, reusable representation",
G2SR,2607.14470,arXiv 2026,2026,Surface & Rendering,"G2SR, which exploits a well-posed core of the task: given cross-view 2D splat correspondences, 3D splats follow analytically from multi-view geometry",
GaussFusion,2607.05906,arXiv 2026,2026,Surface & Rendering,Multimodal pre-training framework for 3D Gaussian representations,
Gaussian Sculpting,2608.10602,arXiv 2026,2026,Surface & Rendering,"To address these issues, we propose Gaussian Sculpting, a fully differentiable end-to-end framework for high-quality surface reconstruction",
Gaussian-JEPA,2608.15651,arXiv 2026,2026,Surface & Rendering,Predicts representations of held-out Gaussian token blocks from visible context,
Geometry Gaussians,2606.05124,arXiv,2026,Surface & Rendering,Adds geometry opacity parameter per splat to decouple appearance and geometry in 3DGS,
GLINT,2603.26181,arXiv,2026,Surface & Rendering,Scene-scale transparency via decomposed Gaussian radiance transport for glass,
GS-CPE,2608.10938,arXiv 2026,2026,Surface & Rendering,"This study introduces GS-CPE (Gaussian Splatting based Camera Pose Estimation), a coarse-to-fine framework for 6-DoF camera pose estimation that unifies geometry-based coarse pose estimation with robust 3D Gaussian Splatting",
Heat Kernel Textures,,ECCV 2026 Best Paper,2026,Surface & Rendering,Geodesic Gaussians that do not splat; anisotropic heat kernel on non-Euclidean manifolds replacing standard Euclidean Gaussians for surface texture representation,
HiCo-GS,2608.14136,arXiv 2026,2026,Surface & Rendering,High-fidelity reconstruction framework with two complementary modules,
HoloTetSphere,2607.08398,ECCV 2026,2026,Surface & Rendering,Unified TetSphere mesh reconstruction for physical simulation; Gaussian sphere coupling + edge-based element pruning; bypasses error-prone tetrahedralization,
Incremental 3D Gaussian Triangulation,,arXiv,2026,Surface & Rendering,Online incremental mesh extraction from dense Gaussian triangulation; plane-based pulling constraint + dynamic historical region freezing,
LiteTex-GS,2609.23380,arXiv 2026,2026,Surface & Rendering,Lightweight texturing for Gaussian Splatting that attaches texture maps to primitives to reduce memory and optimization cost of high-frequency details.,
Manifold-GS,2608.00214,arXiv 2026,2026,Surface & Rendering,Certified hybrid asset layer for Gaussian scenes,
Mesh2GS,2606.21898,arXiv 2026,2026,Surface & Rendering,"Instead of relying on heuristic strategies that bind 3D Gaussians to the mesh, we propose a novel white-box 3DGS construction framework, termed Mesh2GS",
MeshWeaver,2606.04688,CVPR,2026,Surface & Rendering,Autoregressive next-vertex prediction guided by multi-level sparse-voxel encoder; 18% compression ratio,
Neural Gabor Splatting,2604.15941,CVPR 2026,2026,Surface & Rendering,Recent years have witnessed the rapid emergence of 3D Gaussian splatting (3DGS) as a powerful approach for 3D reconstruction and novel view synthesis. Its explicit representation with Gaussian prim...,
Neural Harmonic Textures,2604.01204,ECCV 2026,2026,Surface & Rendering,"High-quality primitive-based neural reconstruction; per-primitive learnable feature scaffolds with periodic activation; harmonic decomposition + deferred decoding; compatible with 3DGUT, Triangle Splatting, 2DGS",
ObjectSplat,2608.30423,arXiv,2026,Surface & Rendering,Object-level mesh splatting for enhanced mesh fidelity and interactivity; bridges 3DGS and mesh-based representations at object granularity,
P2M++,2605.00429,arXiv,2026,Surface & Rendering,Enhanced solver for point-to-mesh distance queries,
PAGaS,2604.22129,arXiv,2026,Surface & Rendering,Pixel-aligned 1DoF Gaussians for depth refinement,https://github.com/UZ-SLAMLab/pagas
PCAsplat,2610.11011,arXiv 2026,2026,Surface & Rendering,"Geometry-aware regularization framework for Gaussian splatting based on differentiable local PCA that updates occluded Gaussians, regularizes eigenvalues toward isotropic tangent-plane coverage, and aligns Gaussian normals with neighborhood normals to reduce floaters.",
PDE-Constrained 3DGS,,CVPR,2026,Surface & Rendering,PDE physical constraints for 3DGS artifact removal; geometry occupancy supervision for 3D consistency,
PointSplat,2604.09903,arXiv,2026,Surface & Rendering,Geometry-driven pruning + Transformer refinement for efficient 3DGS,
QuerySplat,2608.01186,arXiv 2026,2026,Surface & Rendering,"To overcome these deficiencies, we propose textbfQuerySplat, a feed-forward 3DGS framework driven by geometric priors and explicit appearance decoupling",
RGS (Reflection),2609.19421,arXiv 2026,2026,Surface & Rendering,"Physically-based deferred rendering for reflection-aware Gaussian Splatting, modeling specular regions on reflective objects.",
RT-Splatting,2605.18263,CVPR 2026 Highlight,2026,Surface & Rendering,Joint reflection-transmission modeling; geometric occupancy disentangled from optical opacity per Gaussian,
SAND,,SIGGRAPH 2026 TOG,2026,Surface & Rendering,Spatially adaptive network depth for efficient neural implicit surface sampling,
SatSurfGS,2605.07181,arXiv 2026,2026,Surface & Rendering,Generalizable 2DGS for sparse-view satellite surface reconstruction; confidence-aware multi-view,
SparseOIT,2605.13855,arXiv 2026,2026,Surface & Rendering,Order-independent transparency via active set method for glass/refractive scenes,
SplashSplat,2609.20818,arXiv 2026,2026,Surface & Rendering,Built on a single principle: impose physical structure only where the observations can constrain it,
StructureGS,2607.26889,arXiv 2026,2026,Surface & Rendering,"To address this limitation, we introduce StructureGS, a reconstruction framework for articulated objects that integrates structure-aware guidance into 3D Gaussian Splatting",
TetraSDF,2511.16273,ECCV 2026,2026,Surface & Rendering,Analytic isosurface extraction with multi-resolution tetrahedral grid; ReLU MLP composed with tetrahedral position encoding preserves CPWA structure for exact zero-level-set extraction; analytic input preconditioner,
TopoSurfel,2608.20687,arXiv 2026,2026,Surface & Rendering,"To address this limitation, we propose TopoSurfel, a novel framework that closes the loop between Gaussian surfels and continuous meshes",
TransmissiveGS,2605.10705,arXiv,2026,Surface & Rendering,Residual-guided disentangled Gaussian Splatting for transmissive scene reconstruction; separates reflection and transmission components via residual-guided decomposition,
TRiGS,2604.00538,CVPR 2024,2026,Surface & Rendering,Tri-plane augmented Gaussian Splatting: tri-plane features + Gaussian geometry for hybrid representation,
View-Dependent Splatting,2605.25426,SIGGRAPH,2026,Surface & Rendering,We present a differentiable framework to automatically learn view-dependent 2D kernels in a splatting-based pipeline to improve reconstruction quality and representation efficiency for novel 3D vie...,
VoxelGS,2605.26616,arXiv,2026,Surface & Rendering,Scaffold-anchored Gaussians tethered to voxelized SDF for fast monocular surface reconstruction,
GS-Manifold,,CVPR 2025,2025,Surface & Rendering,Manifold-constrained Gaussians for surface reconstruction with topological guarantees,
LeanGaussian,,CVPR 2025,2025,Surface & Rendering,Extreme compression from single RGB image for efficient large-scale rendering,
GOF,2312.13299,ECCV 2024,2024,Surface & Rendering,Gaussian Opacity Field: opacity-weighted TSDF fusion for high-fidelity surface extraction from GS,https://github.com/Janotor/GOF
GS2Mesh,2404.01810,CVPR 2024,2024,Surface & Rendering,Surface-regularized GS → mesh extraction with multi-view depth consistency constraints,
GSDF,2403.16964,NeurIPS 2024,2024,Surface & Rendering,"Dual representation: GS guides SDF geometry, SDF provides normal regularization for GS",https://github.com/city-super/GSDF
GSurf,2411.15723,CVPR 2024,2024,Surface & Rendering,Gaussian surface reconstruction with SDF-GS hybrid representation for watertight meshes,
GVKF,2411.01853,NeurIPS 2024,2024,Surface & Rendering,Gaussian Voxel Kernel Functions for highly efficient surface reconstruction via TSDF fusion,
NegGS,2405.18163,arXiv,2024,Surface & Rendering,Negative color values for ring/crescent/non-convex structures,
NeuSG,2312.00846,CVPR 2024,2024,Surface & Rendering,Neural surface-guided GS: SDF-guided Gaussian anchoring for consistent surface reconstruction,
PGSR,2406.06521,TVCG 2024,2024,Surface & Rendering,Planar-based regularizer for high-fidelity surface reconstruction,
SAGS,2510.27318,ECCV 2024,2024,Surface & Rendering,Shape-aware GS: shape priors guiding Gaussian distribution for anatomically faithful reconstruction,
SuGaR,2311.12775,CVPR 2024,2024,Surface & Rendering,Surface-aligned Gaussians for mesh extraction via TSDF + Marching Cubes,https://github.com/Anttwo/SuGaR
SuperGS,2410.02571,CVPR 2024,2024,Surface & Rendering,Super-resolution guided GS: using 2D SR priors to enhance 3DGS rendering quality,
SVGS,2411.18966,arXiv,2024,Surface & Rendering,"Gaussian Splatting demonstrates impressive results in multi-view reconstruction based on Gaussian explicit representations. However, the current Gaussian primitives only have a single view-dependen...",https://github.com/Xrvitd/SVGS
VCR-GauS,2406.05774,NeurIPS 2024,2024,Surface & Rendering,View-consistent depth-normal regularization for GS surface reconstruction,https://github.com/HLinChen/VCR-GauS
AtlasLC,2607.26525,arXiv 2026,2026,Compression & Streaming,"Source-free, training-free compression pipeline for object-centric 3DGS that operates directly on released Gaussian assets, without original images, camera poses, or per-asset optimization",
CAGS,2605.09279,SIGGRAPH 2026,2026,Compression & Streaming,"Volumetric video (VV) streaming enables real-time, immersive access to remote 3D environments, powering telepresence, ecological monitoring, and robotic teleoperation. These applications turn VV st...",https://github.com/yindaheng98/ColorAdaptiveGaussianSplatting
CC-4DGS,2609.02184,IEEE TVCG 2026,2026,Compression & Streaming,"Computational deformation and point-cloud compression for storage-efficient dynamic 4DGS; CDF replaces hash tables (1-3 MB), CCA compresses SH attributes 3-5x, total 20-30 MB",https://github.com/KyungdaePark/CC-4DGS
Clustered Codebook VQ,,arXiv,2026,Compression & Streaming,CGVQ for 2D Gaussian image compression; cluster-guided grouping before quantization; 20% bpp reduction with preserved visual quality,
CodecSplat,2605.25563,arXiv,2026,Compression & Streaming,Ultra-compact entropy-coded latent features for feed-forward 3DGS; 20-108 KiB per scene,
CoSAG,2607.10237,arXiv,2026,Compression & Streaming,Compact Semantic Anchor Gaussians via training-free rate-distortion coding; closed-form transmittance-weighted lift + spatially predictive entropy coder; 37-76x compression over LangSplatV2 at higher accuracy; sub-MB storage,
CRP-GS,2609.23005,arXiv 2026,2026,Compression & Streaming,Rate-distortion optimized 3DGS compression using cross-representation priors for anchor-level entropy modeling.,
CVT-GS,2609.08730,arXiv 2026,2026,Compression & Streaming,Learning to simplify 3DGS with centroidal Voronoi tessellation; reduces Gaussian primitive count,
Deformable 2D Gaussian Splatting,2609.14129,arXiv 2026,2026,Compression & Streaming,Deformable 2DGS for efficient 4K UHD video compression; outperforms neural video compression and implicit representations,
DLGStream,2606.28840,arXiv 2026,2026,Compression & Streaming,"Novel language-embedded FVV representation that streams time-varying language features alongside Gaussian attributes to support 4D environment interaction, scene editing, and spatial intelligence",
DSD-GS,2605.30863,arXiv 2026,2026,Compression & Streaming,Dynamic-Static Decomposition via feed-forward GS encoder + optical flow; static regions skip redundant computation; 10 min training; no COLMAP required,
DSGS,2605.17002,arXiv 2026,2026,Compression & Streaming,Decoder-Side Gaussian Splatting: feed-forward 3DGS from compressed bitstreams for immersive video delivery,
ECoNGS,2607.18466,IEEE VIS 2026,2026,Compression & Streaming,Neural GS for volume visualization; lightweight NNs predict splats from anchors; 6.1x model reduction / 5.9x training speedup,
Gaussians on a Diet,2604.20046,arXiv,2026,Compression & Streaming,"Memory-bounded training, 80% lower peak memory",
GenSplatCodec,2607.24403,arXiv 2026,2026,Compression & Streaming,"Detail-aware feed-forward Gaussian coding scheme within a dual-stream formulation, where the resulting compact Gaussian structural stream is complemented by a lightweight reference appearance stream",
GETA-3DGS,2605.02086,arXiv,2026,Compression & Streaming,End-to-end automatic joint pruning + quantization,
GS-NFS,2606.05650,arXiv 2026,2026,Compression & Streaming,"Accelerates dynamic 3DGS compression and decompression on a GPU, to the point where it can encode and decode at full frame rate",
GS-SCNet,2604.25330,arXiv,2026,Compression & Streaming,Generalizable 3DGS + semantic coding for immersive video,
GS2CI,2608.13502,arXiv 2026,2026,Compression & Streaming,Novel framework that reconstructs high-quality 3D scenes from a single SCI measurement by leveraging 3D Gaussian Splatting (3DGS) and the powerful priors of large-scale vision foundation models (VFMs),
GSCV,2610.07795,arXiv 2026,2026,Compression & Streaming,Gaussian Splatting sequence compression using standard video codecs: an Inter-PLAS variant produces correlated I/P-frame images and a high-bit-depth pipeline improves compressibility and quality over MPEG/point-cloud baselines.,https://github.com/Qi-Yangsjtu/GSCV
JSGS,2608.08659,arXiv 2026,2026,Compression & Streaming,"To address this problem, we propose JPEG State-Guided Supervision for 3D Gaussian Splatting from Mixed-Quality Views (JSGS)",
KISS-GS,2608.26948,arXiv 2026,2026,Compression & Streaming,"To make the gains more transparent, we propose KISS-GS, a modular compression pipeline named after the principle of keeping things simple, designed to decouple compression entirely from training",
LinearMask-GS,2609.10095,arXiv 2026,2026,Compression & Streaming,Stable-mask importance pruning for compact 3DGS; learned-mask pruning with stability guarantees,
MesonGS++,2604.26799,Under review,2026,Compression & Streaming,Post-training codec with 34x compression + ILP hyperparameter search,https://github.com/mmlab-sigs/mesongs_plus
MGS,2603.19234,arXiv,2026,Compression & Streaming,"The ability to render scenes at adjustable fidelity from a single model, known as level of detail (LoD), is crucial for practical deployment of 3D Gaussian Splatting (3DGS). Existing discrete LoD m...",
MMGS,2605.19304,arXiv 2026,2026,Compression & Streaming,10x compressed 3DGS via multi-view ranking + Optimal Transport aggregation; 10% primitives + 10x speedup,
MoQSplat,2609.18624,arXiv 2026,2026,Compression & Streaming,Maps 3DGS content onto the Media over QUIC (MoQ) transport hierarchy,
NanoGS,2603.16103,arXiv,2026,Compression & Streaming,Training-free simplification via local pairwise merging,https://github.com/saliteta/NanoGS
Observation-Gram,2609.28997,arXiv 2026,2026,Compression & Streaming,"Per-Gaussian observation Gram matrix as distortion metric for SH compression; closed-form degree reduction, Lagrangian degree allocation, matrix-weighted Lloyd VQ; +0.49 dB PSNR over Compressed3D, 15% smaller than GSICO at equal quality on Mip-NeRF 360",
OT-UVGS,2604.19127,Eurographics 2026 Short Papers,2026,Compression & Streaming,Optimal-transport UV mapping as capacity allocation,
Provable Pruning via Coresets,,arXiv,2026,Compression & Streaming,First provable coreset construction theorem for 3DGS pruning; sensitivity-based sampling with multiplicative approximation guarantees; SOTA under aggressive compression + minimal finetuning,
QIRF,2607.18067,arXiv 2026,2026,Compression & Streaming,"QIRF, a quantum-inspired non-orthogonal function-space compression method for 3D Gaussian Splatting",
QuARC-GS,2608.18285,arXiv 2026,2026,Compression & Streaming,"Quantization-aware anchor deformation, which suppresses insignificant motion updates while preserving meaningful deformations, maintaining reconstruction quality under low-storage streaming",
Spark 2.0,,Open Source,2026,Compression & Streaming,"Open-source Three.js/WebGL2 renderer for huge 3DGS worlds on the web; streaming Level-of-Detail splat tree, .RAD chunk format, shared GPU splat page table rendering 100M+ splats at steady frame rate on any device",https://github.com/sparkjsdev/spark
SpeedyGS,2607.12656,arXiv 2026,2026,Compression & Streaming,"To systematically address this challenge, we propose SpeedyGS, a Content-Aware 3DGS Compressor that separately optimizes the structural formation and statistical coding",
SplatStream,2607.25971,arXiv 2026,2026,Compression & Streaming,"This paper presents SplatStream, a fine granular scalable Gaussian splatting framework for dynamic 3D scene delivery",
StreamLoD-GS,2601.18475,arXiv 2026,2026,Compression & Streaming,Level-of-Detail structured 3DGS for streaming free-viewpoint video; progressive LoD delivery,
StreamSplat,2608.01659,arXiv 2026,2026,Compression & Streaming,Streaming feed-forward 3DGS framework that incrementally maintains a persistent geometry-grounded scene state and decodes it into renderable 3D Gaussians after each input chunk,
TSOG,2607.28049,arXiv 2026,2026,Compression & Streaming,"Temporally and Spatially Ordered Gaussians (TSOG), a format for efficient representation of 4D Gaussian Splatting (4DGS) content",
VEDAL,2606.02346,arXiv 2026,2026,Compression & Streaming,Variational free energy minimization for 3DGS pruning; prediction-error gating triggers asynchronous pruning + variational uncertainty head models pruning as latent variable with learnable prior; 5.2x compression,—
ZipSplat,2606.05102,arXiv,2026,Compression & Streaming,Token-based feed-forward 3DGS decoupling Gaussian count from pixel grid; ~6x fewer Gaussians,
CompGS,2311.18159,CVPR 2025,2025,Compression & Streaming,Compact GS with learned importance-aware quantization + progressive decoding,
HybridGS,2505.01938,ICML 2025,2025,Compression & Streaming,Hybrid GS compression combining explicit pruning + implicit neural coding,
Compact-3DGS,,arXiv,2024,Compression & Streaming,"Vector quantization + pruning, ~10-15x compression",
ContextGS,2405.20721,NeurIPS 2024,2024,Compression & Streaming,Anchor-level context model for entropy coding replacing uniform quantization in 3DGS,https://github.com/wyf0912/ContextGS
EAGLES,2312.04564,ECCV 2024,2024,Compression & Streaming,Quantized embeddings + coarse-to-fine training + pruning for 10-20x memory compression maintaining quality,https://github.com/Exyro/EAGLES
Embedded-3DGS,,arXiv,2024,Compression & Streaming,"Neural architecture search, ~10x compression",
HAC,2403.14530,ECCV 2024,2024,Compression & Streaming,"Hash-grid context modeling, ~100x compression",https://github.com/yihangchen-ee/HAC
LightGaussian,2311.17245,NeurIPS 2024,2024,Compression & Streaming,Global+local pruning + SVD distillation for 15x compression at 200+ FPS,https://github.com/VITA-Group/LightGaussian
LightGS,,arXiv,2024,Compression & Streaming,"Distillation-based, ~15-20x compression",
MobileGS,,arXiv,2024,Compression & Streaming,Extreme 50-100x compression for mobile deployment,
QUEEN,2412.04469,NeurIPS 2024,2024,Compression & Streaming,Quantized efficient encoding for streaming free-viewpoint video with dynamic Gaussians,
RDO-Gaussian,2406.01597,ECCV 2024,2024,Compression & Streaming,End-to-end rate-distortion optimization: dynamic pruning + ECVQ quantization for 40x+ compression with continuous rate control,
ABCD,2608.27735,arXiv 2026,2026,Acceleration,"ABCD (Alpha-Composited Block Coordinate Descent), an out-of-core training framework for alpha-composited radiance fields, instantiated here for 3D Gaussian Splatting",
ACE-GS,2606.21244,arXiv 2026,2026,Acceleration,"Progressive optimization framework tailored for accurate, compressed, and efficient scene representation",
ACEsplat,2606.22091,arXiv 2026,2026,Acceleration,"Fast per-scene optimization framework that reconstructs 3D Gaussian representations from RGB images and camera poses only, without requiring external 3D priors (e.g., precomputed SfM models or supervised depth maps)",
AGILE-GS,2609.34176,arXiv 2026,2026,Acceleration,Anchor-guided next-best-view selection for active 3DGS; SE(3) virtual anchor pose optimized by Riemannian gradient ascent on expected information gain; greedy ridge-leverage shortlisting; selection latency reduced by 1-2 orders of magnitude,
Atlas,2609.02352,arXiv 2026,2026,Acceleration,"Algorithm-hardware co-design for on-device city-scale 3DGS in VR; hierarchical memory offloading, 18.5x speedup",
Axis-Shared Rasterization Accelerator,,arXiv,2026,Acceleration,3DGS hardware accelerator with axis-shared rasterization + order-independent transmittance; optimized for efficient on-chip rendering (SJTU),
DDF-GS,2606.00817,arXiv 2026,2026,Acceleration,Directed Distance Fields for constant-time ray queries on 3DGS scenes; 52MB neural field distilled from trained 3DGS; 26-72x faster than SDF sphere tracing; supports GI secondary rays,https://github.com/smlab-niser/ddf-gs
DeGS,2608.02099,arXiv 2026,2026,Acceleration,"To address this issue, we propose DeGS, a scalable architecture for efficient 3DGS inference",
Flux-GS,,arXiv,2026,Acceleration,Monte Carlo Specular Energy Aggregator for mobile 3DGS: samples 3rd-order radiance residuals into compact latent space + Attribute-Conditioned SH Enhancement + Multi-view Alpha-based Densification/Pruning; real-time mobile rendering with significant parameter reduction (UTS/Baidu/U Adelaide),
H2-Ray,2610.13498,arXiv 2026,2026,Acceleration,"Hybrid and hierarchical ray tracing architecture for efficient 3DGS rendering with load-balanced spatial BVH, ray-cluster path reuse, and a unified intersection engine for mixed Gaussian/triangle primitives; 3.77x geomean speedup over RaySplats on RTX A5000.",
HiGS,2606.00352,arXiv 2026,2026,Acceleration,Hierarchically Tiled Gaussian Splatting: decouples coarse macro-tile partitioning from fine render-tile rasterization; up to 15.8x faster than original 3DGS while preserving exact front-to-back alpha compositing,
KC-3DGS,2606.03120,arXiv 2026,2026,Acceleration,Augments 3DGS training with wavelet-domain supervision based on natural image statistics,
Local-GS,2606.16566,arXiv 2026,2026,Acceleration,"To address this, we present Local-GS, a warp-coherent rendering paradigm that, organizes Gaussian primitives with respect to SIMT (Single Instruction, Multiple Threads) execution boundaries rather than scene geometry",
MACRO,,arXiv,2026,Acceleration,Training-free multi-plane attention for closeup render optimization; depth-plane decomposition + scale-matching reference; first standardized closeup NVS benchmark,
REFINE,2606.09074,arXiv 2026,2026,Acceleration,Highly accelerated 3DGS pruning framework centered on a novel rendering-free primitive importance metric,
RoofGS,2608.15785,arXiv 2026,2026,Acceleration,"Guided by this analysis, we develop RoofGS, a rendering framework that applies bottleneck-specific optimizations rather than generic kernel acceleration",
SPARE-GS,2607.16624,arXiv 2026,2026,Acceleration,"Based on this formulation, we propose SPARE-GS, a general plug-and-play framework that dynamically aligns the distribution of 3D Gaussian primitives with regional representational demand",
SplatQ,2610.13503,arXiv 2026,2026,Acceleration,"Combinatorial pruning algorithm that coordinates regional budget allocation and joint Gaussian selection under a fixed budget via a quadratic unconstrained binary optimization (QUBO) objective capturing joint deletion effects, solved by simulated annealing or a coherent Ising machine.",
StereoGS,,2026,2026,Acceleration,Energy-efficient 3DGS stereoscopic rendering processor; dual-eye shared compute and memory bandwidth for VR/AR,
TemporalGS,,arXiv,2026,Acceleration,Training-free plug-and-play 3DGS rendering acceleration via temporal priors; dynamic culling + selective rendering; up to 1.48x speedup,
TensorGS,2605.17855,arXiv 2026,2026,Acceleration,Tensor Core acceleration for 3DGS: tensorizes rasterization into FP16 matrix operations with cross-tile grouping; 1.65x end-to-end speedup,
TileGS,2609.03613,arXiv 2026,2026,Acceleration,"Tile-local depth binning for 3DGS rasterization; 1.44x raster-kernel speedup on RTX 4090, matches gsplat output",
TileSkipper,2610.09343,arXiv 2026,2026,Acceleration,"Region-adaptive tile pruning that selects a static per-Gaussian contribution cutoff for a frozen 3DGS checkpoint via calibration renders; one byte per Gaussian with no parameter updates, giving dataset-macro 1.088x-1.238x speedups at ~-0.02 dB.",
TurboGS,2606.15924,arXiv 2026,2026,Acceleration,Error-guided training framework that accelerates 3DGS by concentrating optimization on perceptually informative pixels,
VoroTracing,2608.17682,arXiv,2026,Acceleration,Differentiable ray tracing at 623 FPS (RTX 5090); Voronoi-based traversal with 3.2x throughput over fastest RT methods and 2.8x over 3DGS rasterization,
Speedy-Splat,2412.00578,CVPR 2025,2025,Acceleration,"Precise tile intersection (SnugBox + AccuTile) and efficient Hessian-based pruning (Soft + Hard Pruning); 6.71x render speedup, 10.6x model compression, 1.47x training speedup across Mip-NeRF 360, Tanks & Temples, Deep Blending",https://speedysplat.github.io/
BlitzGS,2605.13794,arXiv 2026,2026,Large-Scale,"We present BlitzGS, a distributed 3DGS framework that reduces active Gaussian workload for fast city-scale reconstruction. BlitzGS manages this workload at three coupled levels. At the system level...",
CaT-GS,2607.17842,CVPR 2026,2026,Large-Scale,Renderspeed: speculative multi-frame preprocessing + inter-frame caching; 10x faster than vanilla 3DGS,
ChronoFuseGS,2609.31339,Pacific Graphics 2026,2026,Large-Scale,Multi-temporal Gaussian fusion with per-splat persistence encoding; merges separately trained 3DGS models from distinct timesteps into single combined model with incremental extension; change-aware visualization at sub-object granularity; evaluated on 7-month flood management dataset,
City-Level 3D Surface,,arXiv,2026,Large-Scale,Viewpoint orientation partitioning for 3DGS scene division; similar-orientation views joint depth estimation; multi-GPU parallel city-scale surface reconstruction,
City-Mesh3R,2605.30310,arXiv,2026,Large-Scale,End-to-end image-to-mesh reconstruction for city-scale 3D mapping,
EvoGS,2606.07179,arXiv,2026,Large-Scale,First continuous-layering 3DGS representation for dynamic large-scale scenes,
GS4City,2604.11401,arXiv,2026,Large-Scale,Hierarchical semantic GS via city-model priors,
HLC-GS,2609.16772,arXiv 2026,2026,Large-Scale,Risk-map-guided height-layer consistency 3DGS for DSM reconstruction from optical satellite imagery; reduces MAE 1.46->1.18 m,
i3dgs,,SIGGRAPH 2026,2026,Large-Scale,Immediate 3DGS for large-scale unordered image collections; incremental reconstruction with on-the-fly pruning,https://github.com/graphdeco-inria/i3dgs
InceptionGS,2609.02747,arXiv 2026,2026,Large-Scale,Generative bootstrapping for large-scale 3DGS under unstructured view sampling; balances reconstruction and generation,
SkyAnchor,2609.13903,arXiv 2026,2026,Large-Scale,Updating metric-scale aerial 3DGS scenes from unposed ground-view sequences; metric SfM anchor + ground-view update,
STARS-GS,2609.03447,arXiv 2026,2026,Large-Scale,Structure-aware regularized 3DGS for large-scale aerial surface reconstruction; F1-score 0.640->0.698 (+9.1%),
TideGS,2605.20150,arXiv 2026,2026,Large-Scale,Out-of-core training for 1B+ Gaussians via SSD-CPU-GPU hierarchy on single 24GB GPU,
VDGS,2609.23049,arXiv 2026,2026,Large-Scale,Visibility-driven large-scale aerial 3DGS reconstruction framework reducing memory and uneven viewpoint coverage from UAV capture.,
CityGaussian,2404.01133,ECCV 2024,2024,Large-Scale,Hierarchical LOD for city-scale real-time rendering,
DOGS,2405.13943,NeurIPS 2024,2024,Large-Scale,Distributed GS with communication-efficient Gaussian consensus for large-scale reconstruction,https://github.com/AIBluefisher/DOGS
Octree-GS,,arXiv,2024,Large-Scale,Octree spatial partitioning + LOD management,
SCube,2410.20030,NeurIPS 2024,2024,Large-Scale,VoxSplats: voxelized splat with hierarchical LOD for large-scale streaming reconstruction,https://github.com/nv-tlabs/SCube
Street Gaussians,2401.01339,ECCV 2024,2024,Large-Scale,Static/dynamic decomposition for urban street scenes,https://github.com/hbb1/Street-Gaussians
VastGaussian,2402.17427,CVPR 2024,2024,Large-Scale,VastGaussian: high-quality large-scene 3DGS reconstruction and real-time rendering via progressive partitioning and decoupled appearance modeling,
Scaffold-GS,2312.00109,ICCV 2023,2023,Large-Scale,Anchor-based structure for efficient large-scale representation,
AdaptSplat,2605.10239,arXiv,2026,Feed-Forward,Adapting vision foundation models for feed-forward 3DGS; leverages pretrained ViT/DINO features for cross-scene generalizable Gaussian prediction,
AESplat,2609.36693,arXiv 2026,2026,Feed-Forward,"Pose-free feed-forward 3DGS with decoupled SH appearance modeling; zeroth-order SH derived directly from input images without training, higher-order SH predicted by shallow MLP with 3D-aware inductive biases; +0.8 dB PSNR over NAS3R, +1.1 dB over DepthSplat on RealEstate10K",https://aesplat.github.io/
AIR,2605.20820,arXiv 2026,2026,Feed-Forward,"AIR, a self-supervised feed-forward framework that amortizes iterative Gaussian fitting into a single network pass, eliminating per-image test-time optimization",
AnchorSplat,,arXiv,2026,Feed-Forward,3D-native end-to-end depth network via Point Anchor Mechanism for GS detail enhancement; 10^5x faster than optimization methods; zero-shot generalization to generated models,
AnyCity,2605.19949,arXiv 2026,2026,Feed-Forward,Observation-grounded generative reconstruction for sparse aerial urban scenes; gated diffusion residual,
ARGS,2604.00494,arXiv,2026,Feed-Forward,Auto-regressive O(log n) multi-scale 3D generation,
ArtSplat,2605.24304,arXiv,2026,Feed-Forward,First feed-forward articulated 3DGS; per-pixel joint map + Cross-State Attention; 400x faster,
AsySplat,2607.10995,arXiv,2026,Feed-Forward,Asymmetric architecture decoupling geometry and appearance modeling; coarse geometry branch + fine appearance branch; ~800x speedup over optimization-based methods,
ATSplat,2607.20417,arXiv 2026,2026,Feed-Forward,"Adaptive 3D Tokens for feed-forward 3DGS; 12 images to 1s reconstruction, 1136 FPS, 5.7x fewer Gaussians",
AVSplat,2609.05925,arXiv 2026,2026,Feed-Forward,Dense-view feed-forward 3DGS with assist-view preconditioning; positive view scaling stable in dense-view regime,
BA-GS,,CVPR 2026,2026,Feed-Forward,Bayesian Adaptive Gaussian Splatting for SfM-free 3D reconstruction,
CompactSplat,2610.13177,arXiv 2026,2026,Feed-Forward,"Feedforward framework enabling content-aware allocation of 3D Gaussians via texture-guided spatial partitioning and hierarchical sampling; concentrates primitives in complex regions while tiling flat areas, with dynamic sparsity control at inference.",
Cross-View Splatter,2605.19656,arXiv 2026,2026,Feed-Forward,Ground+satellite cross-view feature fusion for outdoor 3DGS reconstruction with unified GPS coordinate,
DeblurNVS,2606.01315,arXiv 2026,2026,Feed-Forward,First feed-forward NVS from sparse motion-blurred images without per-scene optimization; recovers intermediate geometry for multi-view reasoning,https://github.com/PKU-YuanGroup/DeblurNVS
DeltaSplat,2610.09853,arXiv 2026,2026,Feed-Forward,"Lightweight Gaussian refinement module for pose-free feed-forward 3DGS that iteratively renders current Gaussians and predicts per-Gaussian updates from residuals, conditioned on per-pixel Plucker rays and rendered depth; adds ~2.2% parameters and +1.75 dB on DL3DV.",
DensiTok,2610.07958,arXiv 2026,2026,Feed-Forward,"Plug-in module for pretrained feed-forward 3DGS models that densifies internal geometry tokens in a compact latent space, making a frozen backbone behave as though it observed more views; single flow-matching step completes unobserved-viewpoint latents.",
DerainSplat,2608.02191,arXiv 2026,2026,Feed-Forward,Weather net that predicts the weather factors from rainy context and yields two support maps,
DéjàView,2605.30215,arXiv,2026,Feed-Forward,Looped transformer block applied recurrently for K steps; exposes K as inference-time compute knob; matches larger baselines with fraction of parameters on 5 reconstruction benchmarks,
EvTrajGS,2608.08585,arXiv 2026,2026,Feed-Forward,"To address this trade-off, this paper presents EvTrajGS, an accurate and efficient 3D Gaussian Splatting framework for unposed event streams",
FastPano3D,2606.30352,arXiv 2026,2026,Feed-Forward,"To address the challenge of rapidly reconstructing detailed 3D indoor scenes from minimal input, we introduce FastPano3D, an end-to-end framework that directly generates renderable 3D Gaussian representations from a single",
FF-ProCams,2607.17803,arXiv 2026,2026,Feed-Forward,Feed-forward 3DGS inverse rendering for projector-camera; Mamba2-Transformer; 8 views beat 297-view optimization,
FlexSplat,2608.07937,arXiv 2026,2026,Feed-Forward,"Feed-forward framework for novel view synthesis (NVS) from uncalibrated, object-centric multi-view image collections",
ForeSplat,2605.22020,arXiv,2026,Feed-Forward,MetaGrad meta-gradient training makes feed-forward 3DGS output optimizer-friendly initializations,
Free Geometry,2604.14048,arXiv,2026,Feed-Forward,Self-evolving feed-forward models via LoRA without 3D GT,https://github.com/hiteacherIamhumble/Free-Geometry
GADA,2607.00595,ICML 2026,2026,Feed-Forward,Geometry-aware deformable aggregation for image-based 3DGS; deformable offsets + implicit confidence weighting; 2.13x faster FPS,
GaussianLens,2509.25603,ECCV 2026,2026,Feed-Forward,Localized high-resolution reconstruction via on-demand Gaussian densification; feed-forward densification with pixel-guided mechanism; scales to 1024x1024 input,
GenWildSplat,2604.28193,CVPR 2026,2026,Feed-Forward,Generalizable sparse-view 3D reconstruction from unconstrained images,
GeoNVS,2603.14965,ECCV 2026,2026,Feed-Forward,Geometry-grounded video diffusion for novel view synthesis; GS-Adapter lifts diffusion features into 3D Gaussian representations for geometry-constrained rendering,
GlobalSplat,2604.15284,arXiv,2026,Feed-Forward,Global latent tokens → 16K Gaussians in 78ms single forward pass,https://r-itk.github.io/globalsplat/
GrapeSplat,2609.23182,arXiv 2026,2026,Feed-Forward,Pose-free feed-forward 3DGS that amalgamates multi-view cues into a voxel-aligned representation and decodes Gaussians without per-scene optimization.,
GRF-Recon,,ECCV 2026,2026,Feed-Forward,Global ray-field optimization for long-sequence feed-forward reconstruction; coarse-to-fine trajectory alignment + LoRA distillation; hybrid-weight sparse ray-field optimization for drift suppression,
GUSH3R,,arXiv,2026,Feed-Forward,Single-forward-pass reconstruction of dynamic humans + static scenes as 3DGS primitives from monocular video; geometry-consistent online NVS,
HyperGS,2607.11500,arXiv,2026,Feed-Forward,Feedforward optimization-free Gaussian video representation; factorized spatiotemporal Transformer + learnable query-based Transformer; 10^4-10^5x speedup; +2.9-3.1 dB over prior video encoders,
IBRSteG,2606.30024,arXiv 2026,2026,Feed-Forward,Generalizable framework for 3DGS steganography that enables undetectable concealment of secret scenes within a steganographic scene,
IDESplat,2601.03824,CVPR 2026,2026,Feed-Forward,Iterative depth probability with epipolar attention,https://github.com/CVL-UESTC/IDESplat
InstanceSplat,2608.07144,arXiv 2026,2026,Feed-Forward,Unified feed-forward 3DGS framework for generalizable 3D reconstruction and instance-aware scene understanding from pose-free multi-view images,
Instant NuRec,2607.14203,arXiv 2026,2026,Feed-Forward,Feed-forward neural reconstruction model that turns a short multi-view driving log into a fully simulatable 3D Gaussian Splatting (3DGS) world in a single forward pass,
InvSplat,2607.02301,arXiv 2026,2026,Feed-Forward,Inverse feed-forward scene splatting; predicts structured 3D Gaussians with intrinsic material attributes (albedo metallic roughness) for relighting,
LangFlash,2605.23287,CVPR Findings,2026,Feed-Forward,Sparse semantic encoding for feed-forward 3D language GS from unposed images,
MAC-Splat,2607.10792,ECCV 2026,2026,Feed-Forward,Multi-Attribute Consistency for sparse-view 3DGS; MASt3R backbone + DINOv3 correspondences; +4.5 dB over Splatt3R,
NoDrift3R,,arXiv,2026,Feed-Forward,Raymap-Guided Coupling for drift-robust unposed feed-forward 3DGS; explicit geometry-appearance coupling module for long-sequence reconstruction,
NoPo4D,2605.22190,arXiv,2026,Feed-Forward,Feed-forward dynamic Gaussians from unposed multi-view videos; velocity decomposition + bidirectional motion,
OF3GS,2606.03254,arXiv 2026,2026,Feed-Forward,Two mechanisms for causal geometric stability: a Decoupled Intrinsic Recovery Head that mitigates cumulative camera-intrinsic bias and scene-scale jitter,
ParticleSplat,2609.19463,arXiv 2026,2026,Feed-Forward,Self-supervised object-centric learning method that decomposes scenes into a set of latent ''particles'' representing semantic entities through feedforward 3D Gaussian Splatting,
PhGS,2609.20623,arXiv 2026,2026,Feed-Forward,Post-hoc pruning and recurrent refinement pipeline that compacts single-view feed-forward 3DGS without retraining the base model.,
PM-Loss,2506.05327,3DV 2026,2026,Feed-Forward,Pointmap-supervised depth regularization for smoother feed-forward 3DGS boundaries,https://github.com/aim-uofa/PM-Loss
ReconSplat,2608.28895,arXiv 2026,2026,Feed-Forward,Feed-forward model for 3D scene reconstruction that aims to address the longstanding trade-off between plausible view generation for unobserved regions and geometric consistency,
Reliev3R,2604.00548,CVPR 2026,2026,Feed-Forward,Reduces dependency on dense multi-view geometric annotations,
RoSplat,2605.13093,arXiv 2026,2026,Feed-Forward,"Generalizable 3D Gaussian Splatting has recently emerged as an efficient approach for novel-view synthesis, enabling feed-forward synthesis from only a few input views. However, existing pixel-wise...",
Spark3R,2605.06270,arXiv,2026,Feed-Forward,"Feed-forward 3D reconstruction models based on Vision Transformers can directly estimate scene geometry and camera poses from a small set of input images, but scaling them to video inputs with hund...",
SparseSplat,,CVPR 2026,2026,Feed-Forward,"Entropy-based adaptive density, SOTA with 22% of Gaussians (150K vs 688K)",https://github.com/victkk/SparseSplat-page
SplatWeaver,2605.07287,arXiv,2026,Feed-Forward,Generalizable novel view synthesis aims to render unseen views from uncalibrated input images without requiring per-scene optimization. Recent feed-forward approaches based on 3D Gaussian Splatting...,
SR3R,,CVPR 2026,2026,Feed-Forward,Rethinking Super-Resolution 3D Reconstruction with Feed-Forward Gaussian Splatting; multi-resolution feature fusion + progressive optimization for high-fidelity 3D from low-res input,
SubSplat,2607.20813,arXiv 2026,2026,Feed-Forward,Subpixel Gaussian reparameterization (SPGR); subdivide main Gaussians from low-res features,
TokenGS,2604.15239,arXiv,2026,Feed-Forward,Learnable tokens decoupling 3D Gaussian prediction from pixels,
TriSplat,2605.26115,arXiv,2026,Feed-Forward,Triangle primitives replace Gaussians; simulation-ready mesh export from single forward pass,
UniqueSplat,2608.02145,arXiv 2026,2026,Feed-Forward,View-conditioned feed-forward 3D Gaussian Splatting model to reconstruct customized 3D radiance fields for each view query,
UniSHARP,,CVPR,2026,Feed-Forward,First unified monocular 3DGS across pinhole/fisheye/360 cameras,
VG2GT,2606.01573,arXiv 2026,2026,Feed-Forward,Voxel-Gaussian splatting visual geometry grounded transformer unifying voxel and Gaussian representations for feed-forward reconstruction,
VoxelTTO,2609.21498,arXiv 2026,2026,Feed-Forward,Feed-forward framework for reconstructing geometrically accurate 3DGS scenes from an arbitrary number of images and optional camera parameters,
VS-Splat,2609.12343,arXiv 2026,2026,Feed-Forward,Voxel-selective feed-forward 3DGS for end-to-end 3D object reconstruction from sparse views,
Wild3R,2606.11894,arXiv 2026,2026,Feed-Forward,Feed-forward approach for unconstrained sparse photo collections,
WildSplat,,arXiv,2026,Feed-Forward,First feed-forward 3DGS from unposed in-the-wild images; dual-branch architecture decoupling geometry and appearance for appearance-conditioned NVS,
WildSplatter,2604.21182,arXiv,2026,Feed-Forward,"Unconstrained images with appearance embeddings, <1s reconstruction",https://github.com/yfujimura/WildSplatter
Z-Order GS,2605.13465,CVPR 2026 Oral,2026,Feed-Forward,Z-order strategy for spatially coherent Gaussian sequence with sparse attention,
AnySplat,2505.23716,SIGGRAPH 2025,2025,Feed-Forward,In-the-wild feed-forward with appearance/lighting variation handling,
CUT3R,2501.12387,CVPR 2025,2025,Feed-Forward,,https://github.com/CUT3R/CUT3R
DepthSplat,2410.13862,CVPR 2025,2025,Feed-Forward,Stereo-guided depth regularization for feed-forward 3DGS,https://github.com/yzhq97/depthsplat
Flash3D,2406.04343,3DV 2025,2025,Feed-Forward,,https://github.com/eldar/flash3d
MonST3R,2410.03825,ICLR 2025,2025,Feed-Forward,,https://github.com/Junyi42/monst3r
NoPoSplat,2410.24207,ICLR 2025,2025,Feed-Forward,,https://github.com/cvg/NoPoSplat
OmniSplat,2412.16604,CVPR 2025,2025,Feed-Forward,Taming feed-forward 3D Gaussian Splatting for omni-view novel view synthesis with cross-view feature matching,
Spann3R,2408.16061,3DV 2025,2025,Feed-Forward,,https://github.com/HengyiWang/spann3r
SplatFormer,2411.06390,CVPR 2025,2025,Feed-Forward,Transformer-based Gaussian prediction for generalizable 3D reconstruction,
VGGT,2503.11651,CVPR 2025,2025,Feed-Forward,,https://github.com/facebookresearch/vggt
VolSplat,2509.19297,arXiv 2025,2025,Feed-Forward,Voxel-aligned Gaussian prediction replacing pixel-aligned splats for multi-view consistency,https://github.com/ziplab/VolSplat
ZPressor,2505.23734,NeurIPS 2025,2025,Feed-Forward,Bottleneck-aware latent compression enabling scalable feed-forward 3DGS from 100+ input views,https://github.com/ziplab/ZPressor
CAT3D,2405.10314,NeurIPS 2024,2024,Feed-Forward,,
EpipolarFree-GS,2410.22817,NeurIPS 2024,2024,Feed-Forward,"Removing epipolar constraint for generalizable NVS, stronger cross-domain generalization",
FreeSplat,2405.17958,NeurIPS 2024,2024,Feed-Forward,Generalizable feed-forward indoor 3DGS with pixel-aligned Gaussian prediction,https://github.com/wangys16/FreeSplat
GeoLRM,2406.15333,NeurIPS 2024,2024,Feed-Forward,Geometry-aware attention for large reconstruction model generating high-quality 3D Gaussians,https://github.com/alibaba-yuanjing-aigclab/GeoLRM
GGN,2503.16338,NeurIPS 2024,2024,Feed-Forward,Gaussian Graph Network modeling inter-Gaussian relationships with graph neural networks,https://github.com/shengjun-zhang/GGN
GS-LRM,2404.19702,ECCV 2024,2024,Feed-Forward,1B-parameter transformer with zero-shot generalization,https://github.com/sunnyuvion/GS-LRM
InstantSplat,2403.20309,arXiv,2024,Feed-Forward,Pose-free sparse-view reconstruction in ~40 seconds,https://github.com/NVlabs/InstantSplat
LatentSplat,2403.16292,ECCV 2024,2024,Feed-Forward,,https://github.com/Chrixtar/latentsplat
MVSplat,2403.14627,ECCV 2024,2024,Feed-Forward,Cost-volume-based 3DGS from 3 sparse views,https://github.com/donydchen/mvsplat
MVSplat360,2411.04924,NeurIPS 2024,2024,Feed-Forward,Feed-forward 360-degree scene synthesis from sparse views,https://github.com/donydchen/mvsplat360
PixelSplat,2312.12337,CVPR 2024,2024,Feed-Forward,Epipolar Transformer for feed-forward stereo GS reconstruction from image pairs,https://github.com/davidtvs/pixelsplat
ReconFusion,2312.02981,CVPR 2024,2024,Feed-Forward,,
Splatt3R,2408.13912,arXiv preprint,2024,Feed-Forward,,https://github.com/btsmart/splatt3r
3D-GIMP,2607.20789,arXiv 2026,2026,Language & Semantic,3DGS inpainting + PatchMatch hybrid; single generative inpainting on key view then propagate,
CapFrame,2608.30342,ECCV 2026,2026,Language & Semantic,Text-instructed viewpoint localization in 3D Gaussian scenes; geometric pseudo-labels bridge language queries and spatial grounding,
CausalSplat,2608.11150,arXiv,2026,Language & Semantic,Hierarchical causal reasoning in 3DGS; implicit intent understanding + complex spatial relationship comprehension for scene-level causal inference,
CDSeg,2608.05482,arXiv 2026,2026,Language & Semantic,"Cross-Domain Segmentation via Gaussian Splatting (CDSeg), a label-transfer interface that requires no task-specific 3D segmentation training and uses Gaussian primitives as a renderable label carrier",
Consistent Scene Understanding in 3DGS,,arXiv,2026,Language & Semantic,Multi-cue mask refinement with semantic/depth/edge fusion + cross-view mask matching for consistent 3DGS instance segmentation,
CoRef-GS,2609.20586,arXiv 2026,2026,Language & Semantic,Cooperative referring Gaussian splatting framework,
DGSG-Mind,2605.29879,arXiv,2026,Language & Semantic,Dynamic 3D Gaussian scene graphs with probabilistic voxel grid + 3D Gaussian Mind for embodied reasoning,
Gaga,,arXiv,2026,Language & Semantic,Group Any Gaussians via 3D-aware Memory Bank; leverages inconsistent 2D masks from zero-shot segmentation models; 3D-aware memory bank for robust cross-pose mask association,
GeoSAM-3D,2606.00447,arXiv 2026,2026,Language & Semantic,Open-vocabulary 3D scene segmentation on monocular video: 3DGS reconstruction + differentiable graph-geodesic propagation kernel (SAM prompt propagates along heat-kernel distance),
GLMap,2605.01736,CVPR 2026,2026,Language & Semantic,Multi-scale Gaussian-Language map for zero-shot navigation,https://github.com/sx-zhang/GLMap
GroupForward,2608.17535,arXiv 2026,2026,Language & Semantic,"To this end, we propose GroupForward, an instance-grouped feed-forward Gaussian splatting model that reconstructs geometry, appearance, instance structure, and semantics from sparse, unposed, and uncalibrated multi-view images",
Ilov3Splat,2605.04506,ICPR 2026,2026,Language & Semantic,Instance-level open-vocabulary 3DGS via multi-resolution hash embedding for CLIP features + SAM contrastive instance field; two-stage 3D clustering for natural language-driven 3D object retrieval (CSIRO),
LB-Edit,2607.19777,arXiv 2026,2026,Language & Semantic,"Attention-guided editing camera placement (ACP) + multi-view attention alignment (MAA); 5 views, 7x lower latency",
LEGO,2608.10057,ECCV 2026,2026,Language & Semantic,Hierarchical language gaussian splatting; multi-view SAM granularity-adaptive re-grading into 3D-consistent hierarchy; per-level language scene graph supports LLM spatial reasoning,
NG-GS,2604.14706,arXiv,2026,Language & Semantic,NeRF-guided Gaussian-level semantic assignment,
NormLift,2609.18898,arXiv 2026,2026,Language & Semantic,3D-side feature lifting formulating per-Gaussian assignment as cosine alignment for semantically reliable Gaussians.,
OP2GS,2605.20044,arXiv 2026,2026,Language & Semantic,Dual-opacity primitives decoupling visual existence from instance occupancy for object-aware 3DGS,
OpenGaFF,2605.06088,arXiv,2026,Language & Semantic,Understanding open-vocabulary 3D scenes with Gaussian-based representations remains challenging due to fragmented and spatially inconsistent semantic predictions across multi-view observations. In ...,
OpenSplatGraph,2610.07569,arXiv 2026,2026,Language & Semantic,"Unified framework that constructs persistent 3D scene graphs directly from an online Gaussian-based open-vocabulary semantic map, augmenting it with a reliability-aware semantic field for query-conditioned object extraction and language-guided grounding.",https://csiro-robotics.github.io/OpenSplatGraph
OutLangSplat,2608.04560,arXiv,2026,Language & Semantic,UAV outdoor 3D language GS; 2D-3D dual-branch representation; training-free contribution and consistency-aware gaussian feature aggregation; first UAV open-vocabulary 3D scene dataset,
PointGS,2605.11520,CVPR 2026,2026,Language & Semantic,"Unsupervised point cloud segmentation is critical for embodied artificial intelligence and autonomous driving, as it mitigates the prohibitive cost of dense point-level annotations required by full...",
ReLaGS,2603.17605,CVPR 2026,2026,Language & Semantic,Relational Language Gaussian Splatting for open-vocabulary 3D scene understanding,
Robust Prior-Guided Seg.,2605.16065,ICIP,2026,Language & Semantic,SAM-HQ + prior-guided label reassignment for multiview-consistent 3DGS segmentation; object removal/extraction/recoloring,
SAGO,,arXiv,2026,Language & Semantic,Online segment 3D Gaussians via virtual drones; reframes interactive 3DGS segmentation as Next-Best-View planning; 50x faster than prior methods,
SCOUP,2605.13600,arXiv 2026,2026,Language & Semantic,Sparse Code Uplifting for efficient 3D language GS with 400x training speedup,
Semantic Foam,2604.26262,CVPR 2026 Highlight,2026,Language & Semantic,Volumetric Voronoi mesh for spatial + semantic decomposition,http://semanticfoam.github.io/
TrackRef3D,2605.26576,arXiv,2026,Language & Semantic,Track-then-label paradigm with TSCM for open-world referring segmentation in 3DGS,
VCAR,2608.30870,ACM MM 2026,2026,Language & Semantic,Training-free 3DGS segmentation via view completeness and axis-aware boundary refinement; no additional training required for high-quality 3D segmentation,
VLRC,,arXiv,2026,Language & Semantic,Vision-Language Reprojection Consistency as scalable signal for better feed-forward 3D pretraining,
X-GS,2603.09632,arXiv,2026,Language & Semantic,Extensible GS framework: X-GS-Perceiver for SLAM + semantic distillation for downstream tasks,
ZeroSplat,2607.18801,ECCV 2026,2026,Language & Semantic,"Generalized referring 3DGS segmentation (GR3DGS); zero-feature, training-free, 0/1/N target support",
GaussianGraph,2503.04034,arXiv,2025,Language & Semantic,3DGS-based scene graph generation with Control-Follow clustering + 3D spatial relation correction for open-world understanding,
GS-LLM,2412.09176,CVPR 2025,2025,Language & Semantic,LLM-guided GS for reasoning-driven 3D scene understanding and manipulation,
ReferSplat,,ICML 2025 Oral,2025,Language & Semantic,Referring segmentation in 3D Gaussian Splatting with language-guided instance-level features for open-vocabulary 3D understanding,https://github.com/heshuting555/ReferSplat
SceneSplat,,ICCV 2025 Oral,2025,Language & Semantic,Gaussian Splatting-based scene understanding with vision-language pretraining; SceneSplat-7K dataset; native 3DGS encoder for open-vocabulary segmentation,
Feature 3DGS,2312.03203,CVPR 2024,2024,Language & Semantic,Distilled DINO/SAM features for 3D segmentation/detection,https://github.com/D%20Charles2/feature-3dgs
LangSplat,2312.16084,CVPR 2024,2024,Language & Semantic,CLIP features stored per-Gaussian for open-vocabulary 3D queries,https://github.com/minghanqin/LangSplat
OpenGaussian,2406.02058,NeurIPS 2024,2024,Language & Semantic,Per-Gaussian feature distillation for point-level open-vocabulary 3D understanding,
AniGen,,SIGGRAPH 2026 TOG,2026,Generation,Unified S3 Fields for single-image animatable 3D asset generation with skeletal-guided canonical space and diffusion prior,https://github.com/VAST-AI-Research/AniGen
AnySurf,2605.26149,arXiv,2026,Generation,Any surface generation with directed edge representation,
AssetGen,2605.26137,arXiv,2026,Generation,Deployable 3D asset generation at interactive speed,
DeG,2605.16355,SIGGRAPH 2026,2026,Generation,Density-Sampled Gaussians: learnable probability density on octree for adaptive density control; fully differentiable analogue to densification/pruning; latent diffusion with VecSeq canonical re-indexing,
DENSER,2606.01419,arXiv 2026,2026,Generation,Depth-guided ensemble with staged EFA-GS reconstruction for generative 3D scene synthesis,
DualDiff3D,2609.01516,arXiv 2026,2026,Generation,Novel pipeline that leverages dual diffusion priors with a Structure-Appearance Attention (SAA) module to introduce reference guidance for refining low-quality novel views rendered from flawed 3D representations,
FillGauss,2607.17773,arXiv 2026,2026,Generation,"Building on this dataset, we propose a novel generative framework (FillGauss) that integrates 3D Gaussian Splatting (3DGS) with internal state conditioning for sound generation",
FillGS,2607.29284,arXiv 2026,2026,Generation,4D Gaussian Splatting (4DGS) can render dynamic scenes photorealistically,
Filling the Unseen,2609.13262,arXiv 2026,2026,Generation,Scene extrapolation via 3DGS; fills unobserved regions and OOD novel views beyond training view distribution,
FlowObject,2606.19019,arXiv 2026,2026,Generation,"Framework that reformulates sparse-view 3D reconstruction as a training-free, guided inverse problem",
FLUX3D,2606.24874,arXiv 2026,2026,Generation,"To address these issues, we propose FLUX3D, a scalable image-to-3DGS framework that boosts both representation learning and cross-modal alignment during generation",
GaussianZoom,2605.18252,arXiv 2026,2026,Generation,Progressive zoom-in generative 3D reconstruction with multi-scale semantic reasoning + expandable continuous LoD,
GaussVid,2608.21849,arXiv 2026,2026,Generation,Novel 3D-aware video restoration framework designed to enhance the quality of sparse 3DGS reconstruction,
GS-Voxel,2608.17988,arXiv 2026,2026,Generation,"Fitting-free structured latent framework, and evaluate it for large-scale aerial 3D Gaussian scene generation",
GSComplete,2609.08449,arXiv 2026,2026,Generation,Gaussian splat completion with 2D diffusion priors; fills missing regions of incomplete Gaussian splats,
InfiniSplat,2608.02437,arXiv 2026,2026,Generation,Feed-forward single-image 3DGS framework that moves from a pixel-aligned representation toward a surface-aligned representation,
Lucida,2608.30821,arXiv,2026,Generation,Parse-Generate-Place paradigm for Real-to-Sim composable scene modeling; structured scene generation with physical plausibility for embodied AI training,
MeshFlow,2606.04621,CVPR,2026,Generation,Efficient artistic mesh generation via MeshVAE + flow-based Diffusion Transformer; compact latent space,
PanoWorld,2605.17916,arXiv 2026,2026,Generation,Generative spatial world model for whole-house panorama synthesis with floorplan-derived 3D shell and dynamic 3DGS cache,
PhysX-Omni,2605.21572,arXiv 2026,2026,Generation,Unified simulation-ready physical 3D generation for rigid + deformable + articulated objects; PhysXVerse dataset,
PixGS,,arXiv,2026,Generation,Single-stage pixel-space diffusion directly generating 3D Gaussians; bypasses lossy latent compression; denoise Gaussian attributes per timestep; inference ~1s/A100,
ROAR-3D,2605.21121,arXiv 2026,2026,Generation,Routing arbitrary views for multi-view 3D generation; token-wise view router establishes 2D-to-3D correspondences without explicit pose input; dual-stream attention preserves pretrained single-view behavior,
SceneAssistant,2603.12238,arXiv 2026,2026,Generation,Visual-feedback-driven agent for open-vocabulary 3D scene generation; VLM + 3D object generation with atomic operations (Scale/Rotate/FocusOn); iterative visual feedback loop for spatial coherence; supports natural-language scene editing,https://github.com/ROUJINN/SceneAssistant
SceneConductor,2606.08402,arXiv 2026,2026,Generation,Multi-agent orchestration for single-image 3D scene generation; three-stage pipeline: scene initialization + environment construction + multi-agent refinement; geometry-aware layout predictor from point maps; planner agent dispatches specialist agents for localized revisions,
SceneFrom3D,,arXiv,2026,Generation,Geometry-conditioned outdoor 3D scene generation via view scheduling with object-level control; directed generation graph with anchor views + interpolation trajectories,
Scenethesis,2505.02836,ICLR 2026,2026,Generation,Training-free agentic framework integrating LLM scene planning with vision-guided layout refinement; vision module generates image guidance + extracts scene structure; optimization module enforces pose alignment and physical plausibility; judge module verifies spatial coherence,
SIC3D,2604.08760,arXiv,2026,Generation,Style image conditioned text-to-3D Gaussian Splatting generation,
SPAR3S,2609.03931,ECCV 2026,2026,Generation,"Sparse voxel-aligned 3D latent generative model for scene completion; learns compact latent space via differentiable 3DGS, masked autoregressive transformer",
SymTRELLIS,2606.04108,arXiv,2026,Generation,Symmetry-enforced voxel latents for 3D generation; velocity symmetrization at each ODE step,
TRELLIS.2,,CVPR 2026 Best Student Paper,2026,Generation,4B-parameter native 3D large model; 17s PBR asset generation with integrated texture and material synthesis,
X-Splat,2607.02099,arXiv 2026,2026,Generation,First Gaussian Splatting framework for generating CBCT-like 3D dental volumes from a single PXR,
HoGS,2503.19232,CVPR 2025,2025,Generation,Unified near and far object reconstruction via homogeneous Gaussian Splatting,
SceneGenAgent,2410.21909,ACL 2025,2025,Generation,LLM-based coding agent for precise industrial scene generation via C# code; structured calculable format + layout verification + iterative refinement; 81% success rate on real industrial tasks; SceneInstruct dataset for fine-tuning open-source LLMs,
DiffGS,2410.19657,NeurIPS 2024,2024,Generation,Functional Gaussian Splatting diffusion in function space (not original space),https://github.com/weiqi-zhang/DiffGS
Director3D,2406.17601,NeurIPS 2024,2024,Generation,Text to progressive 3D scene GS generation with camera trajectory planning,https://github.com/imlixinyang/director3d
DreamGaussian,2309.16653,ICLR 2024 Oral,2024,Generation,SDS text-to-3D with 3DGS prior for orders-of-magnitude speedup,https://github.com/dreamgaussian/dreamgaussian
GaussianCube,2403.19655,NeurIPS 2024,2024,Generation,Optimal Transport reorders Gaussians into voxel grid structure enabling 3D diffusion generation,https://github.com/GaussianCube/GaussianCube
GaussianDreamer,2310.08529,CVPR 2024,2024,Generation,Fast 3DGS-based text-to-3D generation coupling SDS with structured Gaussian initialization,https://github.com/hustvl/GaussianDreamer
GSGAN,2406.02968,NeurIPS 2024,2024,Generation,Hierarchical GAN for direct 3D Gaussian generation,https://github.com/hse1032/GSGAN
MVGamba,2406.06367,NeurIPS 2024,2024,Generation,State Space Model (Mamba) replacing Transformer for 3D generation via sequence modeling,https://github.com/SkyworkAI/MVGamba
Tetrahedron Splatting,2406.01579,NeurIPS 2024,2024,Generation,Novel tetrahedron primitive replacing Gaussian ellipsoid for 3D generation tasks,https://github.com/fudan-zvg/tet-splatting
Asset Harvester,2604.18468,arXiv,2026,Autonomous Driving,AV log → simulation-ready 3D assets via SparseViewDiT,https://research.nvidia.com/labs/sil/projects/asset-harvester/
ConFixGS,2605.09688,arXiv,2026,Autonomous Driving,"Feedforward 3D Gaussian Splatting (3DGS) often struggles in trajectory-based sparse-view driving scenes. Existing Gaussian repair methods mainly target optimization-based 3DGS, while diffusion-base...",
DecoupleGS,2608.01761,arXiv 2026,2026,Autonomous Driving,"Extensive experiments demonstrate that DecoupleGS achieves a balanced fidelity-efficiency trade-off, improves metric and photometric consistency",
DeGO,2605.28587,CVPR,2026,Autonomous Driving,Deformable Gaussian occupancy decoupling rigid and non-rigid motion with factorized 4D VGGT distillation,
DLWM,,CVPR 2026,2026,Autonomous Driving,Dual Latent World Model: 3DGS-centric self-supervised pretraining for autonomous driving via two-stage Gaussian-based world modeling,
GaussianDWM,2512.23180,CVPR 2026,2026,Autonomous Driving,3D Gaussian Driving World Model unifying scene understanding + multi-modal generation; task-aware language-guided sampling for compact 3D tokens into LLM,
GEM,2605.17682,arXiv 2026,2026,Autonomous Driving,Gaussian Evolution Model: non-autoregressive 4D Gaussian occupancy forecasting for motion planning,
Ground4D,2605.04435,arXiv,2026,Autonomous Driving,Spatially-grounded feedforward 4D for off-road reconstruction,https://github.com/wsnbws/Ground4D
P2GS,2605.16925,CVPR 2026,2026,Autonomous Driving,Physical prior-guided 3DGS: joint HDR radiance + exposure decomposition for photometrically consistent urban reconstruction,
ParkingWorld,2605.25029,arXiv 2026,2026,Autonomous Driving,"Autonomous parking demands precise low-speed maneuvering within narrow, cluttered, and highly constrained environments, where vehicles must navigate tight spaces while avoiding static obstacles and complex geometric boundaries",
PointForward,2605.11594,arXiv,2026,Autonomous Driving,Sparse 3D queries in world space with spatial-temporal fusion + scene graphs for feedforward driving reconstruction,
RAD,,NeurIPS 2025,2026,Autonomous Driving,3DGS-based RL for end-to-end driving: 3DGS twin digital world + closed-loop RL training overcoming IL causal confusion,
RealityBridge,2606.16278,arXiv 2026,2026,Autonomous Driving,"Task-oriented curation pipeline to construct training data, and design a four-stage supervised training strategy followed by reward-guided post-training",
REAP,2605.08713,arXiv 2026,2026,Autonomous Driving,Soft predictive collision penalty mechanism to reduce collision rates by penalizing obstacle-approaching actions,
RoGS,2607.15048,arXiv 2026,2026,Autonomous Driving,Adaptive Meshgrid Gaussian for large-scale road surface mapping; combines mesh-based and Gaussian representations for HD map generation and lane-level perception,
SparseStreet,2606.03909,arXiv 2026,2026,Autonomous Driving,"Motivated by this, we propose SparseStreet, a general compression framework specifically designed for street scenes",
StreetNVS,2606.01590,arXiv 2026,2026,Autonomous Driving,Multi-sensor NVS framework (LiDAR + multi-camera + ego-motion) via Reference-Enhanced Camera Attention + relative ray-level positional encoding; matches 10-100x denser methods on Waymo,
X-World,,arXiv,2026,Autonomous Driving,Controllable multi-view generative world model (video diffusion + 3DGS simulation) for autonomous driving,
GaussianLSS,,CVPR 2025,2025,Autonomous Driving,BEV perception via Gaussian Splatting,
GaussianSSC,2603.21487,CVPR 2025,2025,Autonomous Driving,GS-based 3D semantic scene completion with Gaussian-anchored feature lifting,
SplatAD,2411.16816,CVPR 2025,2025,Autonomous Driving,Autonomous driving GS with dynamic object decomposition and sensor simulation,
ADS-GS,,arXiv,2024,Autonomous Driving,Static + dynamic decomposition for driving scenes,
GaussianBeV,2407.14108,WACV 2025,2024,Autonomous Driving,BEV perception via Gaussian Splatting: lifting 2D features into 3D Gaussian BEV representation,
Street-GS,,arXiv,2024,Autonomous Driving,LiDAR-camera fusion + multi-view optimization,
3DGS³,2605.11489,arXiv,2026,Dynamic & 4D,Joint super sampling and frame interpolation for real-time large-scale 3DGS rendering; unified SS+FI pipeline enabling high-fidelity frame generation at scale,
4DGS-Fixer,2609.21176,arXiv 2026,2026,Dynamic & 4D,Generative sparse-view 4D Gaussian Splatting with iterative refinement guided by video diffusion priors.,
ACA-GS,2608.04581,arXiv 2026,2026,Dynamic & 4D,"Recent advances in 4D Gaussian Splatting (4DGS) enable high-fidelity, real-time spatiotemporal rendering, but expose a fundamental trade-off between motion expressiveness and storage efficiency",
Amortized Anchor Refinement,2608.30218,arXiv,2026,Dynamic & 4D,Deployable continuous-time 4D Gaussian reconstruction with amortized anchor refinement; enables real-time temporal coherent 4D reconstruction from multi-view video,
AniGS,2607.18539,arXiv 2026,2026,Dynamic & 4D,Scene-level 3DGS animation via diffusion prior; canonical 3DGS + time-conditioned deformation; iterative dataset-model update,
ATGS,2608.30184,SIGGRAPH 2026,2026,Dynamic & 4D,Anchored Temporal Gaussian Splatting for long volumetric video representation; temporal anchor points stabilize Gaussian trajectories over extended video sequences,
ChainSplat,2608.28570,arXiv 2026,2026,Dynamic & 4D,"Physics-inspired framework that jointly learns the 3D geometry, appearance, kinematics, and dynamics of DLOs solely from multi-view RGB videos",
ClipGStream,2604.13746,CVPR 2026,2026,Dynamic & 4D,Clip-stream any-length any-motion multi-view dynamic scene reconstruction,
Color-Encoded Illumination,2604.26920,CVPR 2026 Highlight,2026,Dynamic & 4D,High-speed volumetric reconstruction via color-coded illumination,
Comp4D,2403.16993,WACV 2026,2026,Dynamic & 4D,,https://github.com/VITA-Group/Comp4D
CubifyGS,2606.28720,arXiv 2026,2026,Dynamic & 4D,Object-level mapping framework that shifts dynamic maintenance from passive re-optimization to active asset management,
D2-4DGS,2608.01588,arXiv 2026,2026,Dynamic & 4D,"To exploit their complementarity, we propose D2-4DGS, a sparse-camera dynamic 4D Gaussian Splatting framework guided by dual-source depth priors",
D4RT,2512.08924,CVPR 2026 Best Paper,2026,Dynamic & 4D,Unified query mechanism for 4D reconstruction + tracking from monocular video; 200+ FPS,
DeGenseGS,,arXiv,2026,Dynamic & 4D,Geometrically and semantically decoupled 4DGS for surgical scene understanding; HexPlane spatiotemporal entanglement module synchronizes semantic changes with scene dynamics,
DynActiveGS,2608.01178,arXiv 2026,2026,Dynamic & 4D,Dynamic-aware active reconstruction framework based on 3D Gaussian Splatting (3DGS) for autonomous exploration in dynamic environments,
EdMCGS,2609.08332,arXiv 2026,2026,Dynamic & 4D,Event-driven Markov chain 3DGS for extreme-low-frame-rate dynamic scene reconstruction from RGB + event stream,
ERF-GS,2608.08531,arXiv 2026,2026,Dynamic & 4D,Event-RGB fusion Gaussian splatting (ERF-GS) framework that integrates event information into both optimization and densification stages of the Gaussian splatting pipeline,
FastFlowGS,2609.16310,arXiv 2026,2026,Dynamic & 4D,Streaming 4DGS for fast-moving subjects from sparse external cameras; fuses sparse matches + optical flow via Kalman temporal update; +12.6% VMAF on Panoptic,
FreeForm,2605.29318,CVPR 2026,2026,Dynamic & 4D,Reduced-order deformable simulation from particle-based skinning eigenmodes; 40x faster,
FreeTimeGS++,2605.03337,arXiv,2026,Dynamic & 4D,Principled analysis + gated marginalization for 4DGS stability,
FRUC,,arXiv,2026,Dynamic & 4D,Feed-forward 3DGS for dynamic scene reconstruction from uncalibrated collaborative driving views; ego-centric causal occlusion field + zero-initialized injection,
GeoRect4D,2604.20784,arXiv,2026,Dynamic & 4D,"Reconstructing dynamic 3D scenes from sparse multi-view videos is highly ill-posed, often leading to geometric collapse, trajectory drift, and floating artifacts. Recent attempts introduce generati...",
GrainGS,2607.21448,arXiv 2026,2026,Dynamic & 4D,"Hierarchical anchor skeleton + per-Gaussian deformation with stop-gradient; 36.98 dB PSNR, 435.6 FPS, 4.67 MB",
Grassmannian Splatting,2607.10489,arXiv,2026,Dynamic & 4D,Rank-2 spacetime surfels via Grassmannian parameterization; closed-form motion model without deformation field; fastest among compared methods (4.9-5.6x over quality baselines),
HDR-NSFF,2603.08313,ICLR 2026,2026,Dynamic & 4D,Dynamic HDR radiance fields from alternating-exposure video,https://shin-dong-yeon.github.io/HDR-NSFF/
L2D2-GS,2606.29374,arXiv 2026,2026,Dynamic & 4D,"This paper presents L2D2-GS, a unified framework that reformulates generalizable reconstruction not as a one-shot regression, but as a robust iterative process of optimization and densification",
LagrangeGS,2608.22773,arXiv 2026,2026,Dynamic & 4D,Formulates dynamic 3DGS as a non-conservative Lagrangian system,
LetCamsGo,2606.04593,CVPR,2026,Dynamic & 4D,4D reconstruction from sparse dynamic cameras; inter-camera matching + intra-camera tracking,
MORPHOS,2606.02491,arXiv 2026,2026,Dynamic & 4D,Autoregressive 4D generation with Temporal Structured Latents (T-SLAT) unifying mesh/Gaussian/radiance field; causal attention ensures temporal consistency,
MotionDreamer,2405.20155,arXiv 2026,2026,Dynamic & 4D,Category-agnostic skeletal animation from 2D video; structural-semantic injection maps visual dynamics to heterogeneous joint hierarchies; 20;000+ 3D model dataset with rigging and animation,
Multi4D,2606.22197,arXiv 2026,2026,Dynamic & 4D,"To resolve this, we introduce Multi4D, a framework for high-fidelity dynamic Gaussian Splatting based on multi-level competitive allocation",
MVFusion-GS,,arXiv,2026,Dynamic & 4D,Motion-variance guided refinement + MotionFormer temporal attention for dynamic 3DGS; explicit motion awareness improves foreground motion + static background,
NeuROK,2605.30347,CVPR,2026,Dynamic & 4D,Generative 4D Neural Object Kinematics; learns latent kinematic state for simulative dynamics,
OuroWorld,2610.12461,arXiv 2026,2026,Dynamic & 4D,"Mask-free framework that turns any static 3DGS scene into a 3D cinemagraph with diverse, seamlessly looping motion from any viewpoint; Inconsistency-Robust Periodic 4DGS guarantees looping by construction while a grounded drift field absorbs cross-view inconsistency.",https://ouroworld.userwei.com
OVOW,2606.31388,ECCV 2026,2026,Dynamic & 4D,Monocular video to instance-level 4D mesh scene for physics simulation; unifies static/rigid/non-rigid; gravity/contact/support correction; 120-scene benchmark,
PaMoSplat,2605.10307,TCSVT,2026,Dynamic & 4D,"Dynamic scene reconstruction represents a fundamental yet demanding challenge in computer vision and robotics. While recent progress in 3DGS-based methods has advanced dynamic scene modeling, obtai...",
ParticleGS,2505.20270,CVPR 2026 Highlight,2026,Dynamic & 4D,The ability to extrapolate dynamic 3D scenes beyond the observed timeframe is fundamental to advancing physical world understanding and predictive modeling. Existing dynamic 3D reconstruction metho...,
PD-4DGS,2605.11427,arXiv,2026,Dynamic & 4D,"4D Gaussian Splatting (4DGS) enables high-quality dynamic novel view synthesis, yet current models remain monolithic bitstreams that clients must download in full before any frame can be rendered, ...",
PersistGS,2606.03479,arXiv 2026,2026,Dynamic & 4D,"$textbfPersistGS$, a method that restores object permanence during occlusion by coupling differentiable rigid body simulation with 3D Gaussian Splatting",
R5DGS,2605.25909,arXiv,2026,Dynamic & 4D,Identity Encoding + rigid-body centroid constraint for semantic-aware 4DGS; 11 FPS speedup,
RiGS,2605.23672,arXiv,2026,Dynamic & 4D,Static/rigid/transient Gaussian decomposition + scene flow guidance for monocular 4DGS,
SemDynReg,2606.28656,arXiv 2026,2026,Dynamic & 4D,Deformable 3D Gaussian Splatting (3DGS) has emerged as an efficient approach for rendering dynamic scenes in a wide range of 3D applications,
SMG,2608.31023,ECCV 2026,2026,Dynamic & 4D,Semantic Motion Graph for monocular dynamic Gaussian splatting; semantic-motion coupling enables coherent decomposition of complex non-rigid scenes into motion groups,
T2Mo,2606.05162,arXiv,2026,Dynamic & 4D,Controllable dynamic 3D shape generation via 3D trajectories + text; shape-grounded trajectory embedding,
TIDES,2606.02058,arXiv 2026,2026,Dynamic & 4D,Continuous-time event camera simulator from dynamic Gaussian Splatting; derives per-pixel intensity dynamics from scene model (not frame differencing); multi-threshold crossing + adaptive time-stepping + tile-level bandwidth modeling,
UniFusion,2609.05888,ECCV 2026,2026,Dynamic & 4D,Unified spatio-temporal depth alignment for sparse-view 4D reconstruction; spatio-temporal neural fields align cross-view/cross-time depth without masks,
Velox,2605.04527,arXiv 2026,2026,Dynamic & 4D,"We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and appearance; compressive, aiding in downstream efficiency; ...",
WebSpline,2606.02096,arXiv 2026,2026,Dynamic & 4D,Structure-Informed Spline (learnable Hermite spline) models Gaussian trajectories + Structural Proxy Graph organizes motion; SOTA on iPhone dataset,
World from Motion,,arXiv,2026,Dynamic & 4D,Generative dynamic GS from monocular video; conditions video model to fix rendering artifacts + fill missing regions; distills back to consistent 4D Gaussians,
BARD-GS,2503.15835,CVPR 2025,2025,Dynamic & 4D,Bundle-Adjustable and Rational Deformation Gaussian Splatting; rational motion model with bundle adjustment for complex non-rigid dynamics; rational deformation model for complex non-rigid dynamics,
DynOMo,2409.02104,3DV 2025,2025,Dynamic & 4D,Online motion adaptation for dynamic Gaussian Splatting; adaptive per-frame Gaussian motion estimation handles changing scene dynamics; online motion adaptation for dynamic scenes,
GauFRE,2312.11458,WACV 2025,2025,Dynamic & 4D,"Gaussian Frequency-domain Radial basis Expansion; radial basis function networks for dynamic Gaussian deformation in frequency domain; evaluated on D-NeRF, HyperNeRF, NeRF-DS",
GaussianFlow,2403.12365,CVPR 2025,2025,Dynamic & 4D,Optical flow-guided 4DGS for temporally consistent dynamic scene reconstruction,
GaussianWorld,2412.10373,CVPR 2025,2025,Dynamic & 4D,World-model GS: dynamic scene forecasting with Gaussian-based future prediction,
GFlow,2405.18426,AAAI 2025,2025,Dynamic & 4D,"Gaussian Flow for dynamic scene reconstruction; flow-based temporal motion modeling with warp refinement for continuous-time rendering; evaluated on D-NeRF (PSNR 34.27), HyperNeRF (PSNR 26.30, 12 min training), Plenoptic Video (PSNR 32.00, SSIM 0.970)",
IGS,2503.16979,CVPR 2025,2025,Dynamic & 4D,Incremental Gaussian Splatting; incremental reconstruction pipeline with per-frame Gaussian management for streaming dynamic scenes; incremental Gaussian management for streaming dynamic scenes,
MoDec-GS,2501.03714,CVPR 2025,2025,Dynamic & 4D,"Motion Decomposition Gaussian Splatting; decomposes scene motion into global and local components with separate deformation networks; evaluated on HyperNeRF (PSNR 27.78, rank 1)",
MoDGS,2406.00434,ICLR 2025,2025,Dynamic & 4D,Motion-aware 3D Gaussian Splatting for dynamic scenes; explicit motion field modeling with smoothness constraints for high-quality temporal coherence; motion-aware approach for monocular dynamic scenes,
ReconDreamer++,2503.18438,arXiv 2025,2025,Dynamic & 4D,Incremental reconstruction and rendering for dynamic driving scenarios; integrates temporal Gaussian management with driving scene expansion; dynamic driving scenario reconstruction,
RobustSplat,,ICCV 2025,2025,Dynamic & 4D,Decouples densification from dynamics for transient-free 3DGS,
Shape of Motion,2407.13764,ICCV 2025,2025,Dynamic & 4D,,https://github.com/vye16/shape-of-motion
SpectroMotion,2410.17249,CVPR 2025,2025,Dynamic & 4D,"Spectral temporal analysis for motion-aware 3DGS; frequency-domain decomposition of Gaussian motions handles both slow and fast dynamics; evaluated on HyperNeRF and NeRF-DS (PSNR 24.65, rank 1 on NeRF-DS)",
SplineGS,2412.09982,CVPR 2025,2025,Dynamic & 4D,Spline-based sparse control for dynamic Gaussian Splatting; B-spline trajectories of sparse control points model continuous scene deformation; B-spline sparse control for dynamic deformation,
STC-GS,2502.14895,ICLR 2025,2025,Dynamic & 4D,Spatio-Temporal Consistent Gaussian Splatting; temporal consistency loss with Gaussian re-projection for artifact-free dynamic rendering; spatio-temporal consistent dynamic rendering,
STG,2510.22140,CVPR 2025,2025,Dynamic & 4D,Spatiotemporal Gaussians with decomposed spatial-temporal attention for 4D rendering,
3DGStream,2403.01444,CVPR 2024,2024,Dynamic & 4D,3D Gaussian Stream for real-time radiance field streaming; per-frame 3DGS optimization with temporal consistency for long dynamic video sequences; per-frame streaming for long dynamic video sequences,
4D-rotor GS,2402.03307,SIGGRAPH 2024,2024,Dynamic & 4D,"4D rotor-based Gaussian Splatting; geometric algebra rotors for compact 4D motion representation suitable for articulated and non-rigid dynamics; evaluated on D-NeRF (PSNR 34.26) and Plenoptic Video (277.5 FPS, highest frame rate)",
4DGS,2310.08528,CVPR 2024,2024,Dynamic & 4D,4D anisotropic Gaussians (3D + time) with regularized deformation,https://github.com/hustvl/4DGaussians
CoGS,2312.05664,CVPR 2024,2024,Dynamic & 4D,Controllable Gaussian Splatting for dynamic scenes; invertible deformation network enables both reconstruction and real-time editing of dynamic content; deformation field approach for dynamic scenes,
Deformable-3DGS,2309.13101,CVPR 2024,2024,Dynamic & 4D,Deformation field network for 3DGS enabling high-fidelity dynamic scene rendering,https://github.com/ingra14m/Deformable-3DGS
DN-4DGS,2410.13607,NeurIPS 2024,2024,Dynamic & 4D,Denoised deformable network with temporal-spatial aggregation for dynamic scene rendering,https://github.com/peoplelu/DN-4DGS
DreamMesh4D,2410.06756,NeurIPS 2024,2024,Dynamic & 4D,Sparse-controlled Gaussian-Mesh hybrid 4D generation,https://github.com/WU-CVGL/DreamMesh4D
Dual-GS,2409.08353,ACM ToG 2024,2024,Dynamic & 4D,Dual-stream Gaussian Splatting; separates global and local geometry streams for stable reconstruction of long-duration dynamic scenes; dual-stream for long-duration dynamic scenes,
DynMF,2312.00112,CVPR 2024,2024,Dynamic & 4D,Dynamic neural motion fields decomposing scene motion into compact basis functions for 4D GS,
GPS-Gaussian,2312.02155,CVPR 2024,2024,Dynamic & 4D,Generalizable Pixel-aligned Gaussian Splatting for novel view synthesis of dynamic scenes; feed-forward pipeline without per-scene optimization; generalizable feed-forward approach for dynamic scenes,
Grid4D,2410.20815,NeurIPS 2024,2024,Dynamic & 4D,4D decomposed hash encoding for efficient spatiotemporal Gaussian queries in dynamic GS,https://github.com/JiaweiXu8/Grid4D
HDR-GS,2405.15125,NeurIPS 2024,2024,Dynamic & 4D,HDR-specific GS luminance encoding + fast tonemapping for 1000x HDR view synthesis,https://github.com/caiyuanhao1998/HDR-GS
HiCoM,2411.07541,NeurIPS 2024,2024,Dynamic & 4D,Hierarchical coherent motion for streamable dynamic scene with 3DGS,https://github.com/gqk/HiCoM
L4GM,2406.10324,NeurIPS 2024,2024,Dynamic & 4D,Large-scale feed-forward 4D Gaussian reconstruction from video,
LoopGaussian,2404.08966,ACM MM 2024,2024,Dynamic & 4D,Loop-based motion representation for dynamic 3DGS; decomposes SE(3) transformations into meaningful sub-operations via loop primitives; loop-based motion decomposition for dynamic 3DGS,
MotionGS,2410.07707,NeurIPS 2024,2024,Dynamic & 4D,"Explicit motion field guiding deformable 3DGS, decoupling motion from appearance",
NeuroGauss4D-PCI,2405.14241,NeurIPS 2024,2024,Dynamic & 4D,4D neural field + Gaussian deformation fields for point cloud interpolation,https://github.com/jiangchaokang/NeuroGauss4D-PCI
PhysGaussian,2311.12198,CVPR 2024,2024,Dynamic & 4D,,https://github.com/XPandora/PhysGaussian
Real-time 4DGS,2310.10642,ICLR 2024,2024,Dynamic & 4D,"Real-time 4D Gaussian Splatting; 4DGS with temporal slicing for real-time rendering of dynamic scenes at interactive frame rates; evaluated on D-NeRF (PSNR 34.09) and Plenoptic Video (PSNR 32.01, 114 FPS)",
SAGD,2401.17857,ECCV 2024,2024,Dynamic & 4D,Self-supervised articulated Gaussian discovery: automatic part segmentation + articulation estimation without supervision,
SC-GS,,arXiv,2024,Dynamic & 4D,Spatial-temporal compression for dynamic Gaussians,
SP-GS,2406.03697,ICML 2024,2024,Dynamic & 4D,"Gaussian Splatting with Sparse Pixels; sparse-pixel control points drive deformation field for efficient dynamic scene reconstruction; evaluated on D-NeRF (PSNR 37.98), HyperNeRF (PSNR 26.78), NeRF-DS",
TC4D,2403.17920,ECCV 2024,2024,Dynamic & 4D,,https://github.com/sherwinbahmani/tc4d
Video-3DGS,2406.02541,arXiv 2024,2024,Dynamic & 4D,Video-based 3D Gaussian Splatting; per-frame Gaussian initialization with inter-frame motion priors for video-to-3D dynamic reconstruction; video-based per-frame dynamic reconstruction,
Vidu4D,2405.16822,NeurIPS 2024,2024,Dynamic & 4D,Dynamic Gaussian Surfels for single-video to 4D reconstruction,https://github.com/yikaiw/vidu4d
4DGen,2312.17225,arXiv preprint,2023,Dynamic & 4D,,https://github.com/VITA-Group/4DGen
DreamGaussian4D,2312.17142,arXiv preprint,2023,Dynamic & 4D,,https://github.com/jiawei-ren/dreamgaussian4d
PVG,2311.18561,arXiv 2023,2023,Dynamic & 4D,Periodic Vibration Gaussian; models temporal dynamics via periodic vibration functions enabling 4D representation without explicit deformation networks; periodic vibration function for 4D representation,
AEGIR,2606.28635,arXiv 2026,2026,HDR & Relighting,Framework that explicitly models local area emitters within a relightable Gaussian Splatting representation,
AIGS-Net,2606.17998,arXiv 2026,2026,HDR & Relighting,"To address this issue, this paper proposes an Adaptive Illumination Gaussian Splatting Network (AIGS-Net), an ultra-lightweight architecture for fast low-light enhancement",
AlbedoEdit,2606.01362,arXiv 2026,2026,HDR & Relighting,Unified generative video editing (insertion/removal/texture) via intrinsic albedo map guidance; single-frame albedo edit auto-generates harmonized RGB video,
Ambient-Robust IR,2605.30250,arXiv,2026,HDR & Relighting,Active RGB-NIR imaging for ambient-robust inverse rendering,
BRF-GS,2608.31159,arXiv,2026,HDR & Relighting,Hyperspectral bidirectional reflectance factor modeling with 3DGS; joint hyperspectral image generation and BRDF reconstruction from multi-view hyperspectral imagery,
DR-GS,2606.29379,arXiv 2026,2026,HDR & Relighting,"To address these challenges, we propose Deformable and Relightable GS (DR-GS), a unified Gaussian framework that integrates physically-based inverse rendering, relighting, and deformation-aware manipulation",
F-RNG,2605.25975,arXiv,2026,HDR & Relighting,Feed-forward relightable 3DGS via LRM + residual,
GLUT,2605.19889,arXiv 2026,2026,HDR & Relighting,3D Gaussian Lookup Table for continuous color transformation; replaces grid-based 3D LUT with learnable Gaussian primitives; compact conditional generator CGLUT for multi-style LUT blending; supports localized editing without global retraining,
GS-PI,2609.19907,arXiv 2026,2026,HDR & Relighting,"Multi-scale cross-view conditioning mechanism that integrates three complementary components: a global semantic prior, source-anchored photometric cues, and an absolute spatial learned view-direction conditioning signal",
Instant Colorization,2604.17155,arXiv,2026,HDR & Relighting,Visibility-weighted least squares for per-Gaussian colorization,
LightBridge,2609.02543,arXiv 2026,2026,HDR & Relighting,"Feed-forward generative relighting for 3DGS; latent bridge diffusion + Gaussian propagation transformer, single-pass",
LIT-GS,2606.20424,arXiv 2026,2026,HDR & Relighting,LiDAR-inertial-thermal Gaussian Splatting framework that injects LiDAR-derived plane geometry as an explicit constraint in both pose/structure refinement and Gaussian optimization,
LumiMotion,2604.10994,CVPR 2026,2026,HDR & Relighting,Improving Gaussian relighting with scene dynamics,
MaterialClusterGS,2606.09018,arXiv 2026,2026,HDR & Relighting,Palette-based material decomposition framework for 2D Gaussian Splatting that enables physically based relighting and material editing,
PAM-ToD,2610.11572,arXiv 2026,2026,HDR & Relighting,Lightweight plug-in appearance model for cross-time-of-day 3DGS that learns per-Gaussian color corrections (scale plus additive brightness) while keeping pre-trained parameters fixed; introduces the CARLA-ToD benchmark.,
Relit-LiVE,2605.06658,SIGGRAPH 2026,2026,HDR & Relighting,Relight video by jointly learning environment video,https://github.com/zhuxing0/Relit-LiVE
SRUG,2605.24700,arXiv,2026,HDR & Relighting,Shadow-guided relightable urban scene with 3DGS,
SSD-GS,2604.13333,ICLR 2026,2026,HDR & Relighting,Scattering and shadow decomposition for relightable 3DGS,
VIRGi,2603.02986,IEEE TPAMI 2026,2026,HDR & Relighting,View-dependent instant recoloring with single edited image,
Large Material Gaussian Model,2509.22112,arXiv 2025,2025,HDR & Relighting,Large-scale material-aware Gaussian model for relightable 3D asset generation,
GRF,,arXiv,2024,HDR & Relighting,Material decomposition + relighting in Gaussian space,
GS-IR,,arXiv,2024,HDR & Relighting,Inverse rendering: Gaussians → geometry + BRDF + lighting,
GStex,2409.12954,ECCV 2024,2024,HDR & Relighting,Texture-tiled Gaussians with UV-parameterized appearance for editable material and relighting,
HumanGaussian,2311.17061,CVPR 2024,2024,HDR & Relighting,Text-driven 3D human generation with Gaussian Splatting,
NeuMA,2410.08257,NeurIPS 2024,2024,HDR & Relighting,Neural Material Adaptor replacing SH with physics-constrained material decomposition,https://github.com/XJay18/NeuMA
Spec-Gaussian,2402.15870,NeurIPS 2024,2024,HDR & Relighting,Anisotropic Spherical Gaussians replacing SH for view-dependent specular appearance,https://github.com/ingra14m/Specular-Gaussians
GaussianShader,2311.17977,arXiv,2023,HDR & Relighting,Shading functions for reflective/refractive surfaces,
2DGS-SLAM,,TRO 2026,2026,SLAM,Globally consistent RGB-D SLAM with 2DGS; loop closure + global optimization,https://github.com/PRBonn/2DGS-SLAM
Anchor3R,2606.05035,arXiv,2026,SLAM,Streaming 3D reconstruction with current-centric local prediction; transient anchor + loop closure,
Anythingreality,2607.09260,arXiv,2026,SLAM,"Robust online GS SLAM for VR scene exploration + VLM interaction; ORB-SLAM3-based pose + real-time Gaussian reconstruction; +14.5% PSNR, 88% VLM recognition",
Cube-Splat,2609.21347,arXiv 2026,2026,SLAM,First panoramic GS-SLAM framework that factorizes each 360° frame into a cubemap of four fixed-orientation virtual pinhole views sharing a single optical center,
DL-SLAM,,arXiv,2026,SLAM,Dual-level probability framework for dynamic environment GS-SLAM; semantic+geometric pixel/object-level dynamic probability; artifact-free static map + semantic map,
EliGSiR,2609.20348,arXiv 2026,2026,SLAM,Continual Gaussian mapper that controls how the available optimization budget is used as the reconstruction evolves,
EvFlow-GS,2604.22183,ICME 2026,2026,SLAM,Event camera + optical flow for motion blur handling,
Flow4DGS-SLAM,2604.22339,CVPR 2026,2026,SLAM,Optical flow-guided 4D Gaussian SLAM for dynamic scenes; category-agnostic motion mask; GMM temporal opacity/rotation,
GAVIS,2605.30342,CVPR,2026,SLAM,Uncertainty-driven 3DGS active mapping via anisotropic visibility field; Bayesian Network rasterizer,
GeoGS-SLAM,2607.07452,arXiv,2026,SLAM,Geometry-Only Gaussian Splatting for dense monocular SLAM; removes appearance modeling reducing per-primitive params 80%+; SOTA online mapping + geometry quality,
GeoGS-SLAM (Geometric Priors),2607.11184,arXiv,2026,SLAM,"Online monocular dense reconstruction combining a 3DGS scene map with learned geometric priors from a feed-forward visual geometry model; Gaussian primitives are sampled from both RGB input and priors, then jointly optimized with camera poses",
GGD-SLAM,2604.12837,ICRA 2026,2026,SLAM,Generalizable motion model for monocular dynamic 3DGS SLAM; FIFO sequential attention; no semantic labels/depth required,
GLAM-SLAM,2607.21416,IROS 2026,2026,SLAM,Large-scale outdoor decoupled GS SLAM; feature tracking front-end + sparse anchor grid; 15% better on KITTI/Oxford,
GS-Map-SLAM,,CVPR,2026,SLAM,Gaussian-based dense mapping for SLAM with view-dependent quality; real-time dense mapping via GS optimization,
ImprovedVBGS,2607.15542,arXiv 2026,2026,SLAM,Real-time continual Variational Bayes GS; improved CAVI with per-frame subset updates instead of full-observation iterations; enables on-the-fly SLAM without replay buffers,
LightSplat,2609.07274,arXiv 2026,2026,SLAM,Real-time high-fidelity 3DGS SLAM with loop closure; meets real-time constraints with loop closure support,
LiteLoc,2605.17777,IEEE/CAA JAS 2026,2026,SLAM,Color-free decoupled 3DGS feature field for visual localization: 94% storage reduction + 5% match condensing,
MAGICIAN,2603.22650,CVPR 2026 Oral,2026,SLAM,Active mapping with imagined Gaussians + beam search,https://shiyao-li.github.io/magician/
MAGS-SLAM,2605.10760,arXiv,2026,SLAM,Collaborative photorealistic 3D reconstruction from multiple agents enables rapid large-scale scene capture for virtual production and cooperative multi-robot exploration. While recent 3D Gaussian ...,
MMD-SLAM,2606.19874,arXiv 2026,2026,SLAM,"To address these limitations, we propose MMD-SLAM, a structure-enhanced Visual SLAM framework that leverages the Atlanta World (AW) assumption to guide a Multi-Meta Gaussian representation for photorealistic mapping",
MonoEM-GS,2604.10593,arXiv,2026,SLAM,Monocular expectation-maximization GS SLAM,
MoonSplat,2606.17935,arXiv 2026,2026,SLAM,Our code and data are available at https://github.com/TrickyGo/MoonSplat,
MoPe,2606.29237,arXiv 2026,2026,SLAM,"We realize this principle in MoPe, a memory-aware uncertainty filter for monocular Gaussian mapping",
MyGO-Splat,2606.29738,arXiv 2026,2026,SLAM,"Closed-loop Gaussian SLAM framework that analytically rasterizes Gaussian primitives into pixel-wise depth and surface normals, allowing the map to actively supervise camera pose optimization",
PanoGS-SLAM,2609.17387,arXiv 2026,2026,SLAM,"First panoramic dense SLAM on 3DGS; spherical-domain differentiable rendering, sphere-consistent photometric loss, depth-guided Gaussian init",
Pocket-SLAM,2606.24796,arXiv 2026,2026,SLAM,"Rendering-area-aware pruning strategy that selectively removes Gaussians based on their contribution to the effective rendering area, rather than solely relying on Gaussian-level heuristics such as opacity or gradient magnitude",
RawSLAM,2609.20589,arXiv 2026,2026,SLAM,"To the best of our knowledge, the first online Gaussian SLAM framework that tracks and maps directly on single-exposure 16-bit linear HDR imagery",
Real-Time LiDAR GS-SLAM,,arXiv,2026,SLAM,First real-time LiDAR Gaussian Splatting SLAM; tight G-ICP + spherical rasterization dense mapping; Newer College F-score 86.78%@>20FPS,
RoSe-SLAM,2608.29003,arXiv 2026,2026,SLAM,"Robust Semantic-aware Gaussian Splatting SLAM (RoSe-SLAM), to address the dynamic challenge by a holistic semantic scene understanding from uncalibrated monocular inputs",
SCOUT-SLAM,2609.14634,arXiv 2026,2026,SLAM,Structurally-coupled dual-uncertainty 3DGS SLAM; shared-base LoAd tracking uncertainty independent of reconstruction quality,
SplatlessDF,2606.13990,arXiv 2026,2026,SLAM,"In this paper, building on this principle, we introduce SplatlessDF, a continuous distance field (DF) mapping framework that uses anisotropic Gaussian elements from a spatial rather than photometric perspective",
SplitGS-Loc,2605.07351,arXiv 2026,2026,SLAM,Mixture-of-Gaussians splitting to disambiguate 2D-3D correspondences in GS feature fields,
Structured-Li-GS,2606.27509,arXiv 2026,2026,SLAM,"In this study, we develop a Structured framework for Gaussian Splatting (3DGS) with LiDAR integration (Structured-Li-GS)",
Track2Map,2607.08408,MICCAI 2026,2026,SLAM,Online deformable GS SLAM for robotic surgery; track-anchored deformation initialization; joint camera + scene optimization from surgical video,
Triangle Splatting SLAM,2605.31419,arXiv 2026,2026,SLAM,First dense RGB-D SLAM with differentiable triangles as 3D map representation; Delaunay triangulation converts triangle soup to connected mesh online,
ULF-Loc,2605.04730,CVPR 2026 Highlight,2026,SLAM,Unbiased landmark feature for robust visual localization with 3DGS; geometry-weighted aggregation replaces alpha-compositing features,
VGGT-GS SLAM,2609.19628,arXiv 2026,2026,SLAM,Monocular 3D Gaussian Splatting SLAM system designed for uncalibrated videos,
VIGS-SLAM,2512.02293,ECCV 2026,2026,SLAM,Visual-inertial GS SLAM with iPhone real-time demo; feature tracking + Gaussian mapping; Docker + TensorRT,https://github.com/cvg/VIGS-SLAM
CGS-SLAM,,IROS 2025,2025,SLAM,Compact voxel-based 3DGS for dense visual SLAM,
CoMapGS,2503.20998,CVPR 2025,2025,SLAM,Covisibility map-based Gaussian Splatting for sparse-view novel view synthesis and rendering,
S3PO-GS,,ICCV 2025,2025,SLAM,Global scale-consistent outdoor monocular 3DGS SLAM,
SplatLoc,2409.14067,CVPR 2025,2025,SLAM,GS-based visual localization with Gaussian-anchored map representation,
WildGS-SLAM,2504.03886,CVPR 2025,2025,SLAM,Dynamic environment SLAM with uncertainty-aware mapping,https://github.com/JokerJohn/WildGS-SLAM
DG-SLAM,2411.08373,NeurIPS 2024,2024,SLAM,Dynamic Gaussian SLAM with hybrid pose optimization for dynamic environments,https://github.com/fudan-zvg/DG-SLAM
Gaussian Splatting SLAM,2312.06741,CVPR 2024 (Highlight),2024,SLAM,First real-time monocular 3DGS SLAM,https://github.com/muskie82/MonoGS
Photo-SLAM,2311.16728,CVPR 2024,2024,SLAM,Hyper primitives map with explicit geometric features for localization + implicit photometric features; Gaussian-Pyramid training; runs on Jetson AGX Orin,https://github.com/hjr37/Photo-SLAM
SplaTAM,2312.02126,CVPR 2024,2024,SLAM,First real-time GS-SLAM: online incremental Gaussians with silhouette mask and rendering-based pose optimization,https://github.com/spla-tam/SplaTAM
AugSplat,2606.31556,arXiv 2026,2026,Sparse-View,Simple framework for improving Gaussian Splatting in sparse-view regimes using radiance-field-based view augmentation,
Bi-FlowGS,2609.17039,arXiv 2026,2026,Sparse-View,Bidirectional flow co-refinement bridging generative view completion and Gaussian geometry; V2G/G2V distillation alleviates Geometry Cheating,
D3GS,2609.22941,arXiv 2026,2026,Sparse-View,Depth-DINO-Diffusion co-guided sparse-view 3DGS framework jointly enhancing geometry and appearance.,
DropAnSH-GS,,CVPR 2026,2026,Sparse-View,Dropping Anchor and Spherical Harmonics for sparse-view Gaussian Splatting,
FrameTwin,2605.09362,arXiv,2026,Sparse-View,"We present FrameTwin, a curve-anchored Gaussian alignment framework that uses sparse-view images to close the control loop for adaptive wireframe 3D printing. Our key idea is to capture the deforma...",
GeoQuery,2605.12399,SIGGRAPH 2026,2026,Sparse-View,"3D Gaussian Splatting (3DGS) has emerged as a prominent paradigm for 3D reconstruction and novel view synthesis. However, it remains vulnerable to severe artifacts when trained under sparse-view co...",
GSCompleter,2604.20155,arXiv,2026,Sparse-View,Distillation-free sparse-view completion via Stereo-Anchor,
HeroGS,,CVPR 2026,2026,Sparse-View,Hierarchical image→region→pixel guidance for sparse-view robustness,
OC-GS,2609.31572,arXiv 2026,2026,Sparse-View,"Object-centric Gaussian splatting for irregular turntable capture; orbit-consistent refinement jointly optimizes image-derived geometry and angles with shared camera, rotation axis, and pivot; 12/8/6 irregular views achieve 21.26/19.36/15.83 dB foreground PSNR, exceeding four pose-free baselines",
PairDropGS,2605.12072,arXiv,2026,Sparse-View,Dropout-based sparse-view 3D Gaussian Splatting (3DGS) methods alleviate overfitting by randomly suppressing Gaussian primitives during training. Existing methods mainly focus on designing increasi...,
PanoPlane,2605.14135,arXiv 2026,2026,Sparse-View,"Plane-aware panoramic completion for indoor sparse-view 3DGS, +17.8% PSNR",
Pi-GS,2602.03327,arXiv,2026,Sparse-View,Reference-free π³ initialization for sparse-view 3DGS,
RRTO-CF3DGS,2609.30865,arXiv 2026,2026,Sparse-View,Reliability-regulated trajectory optimization for progressive COLMAP-free 3DGS; self-supervised bidirectional cycle-consistency with forward motion propagation + retrospective trajectory correction in sliding window; outperforms unposed baselines on Tanks & Temples and CO3D-V2,https://github.com/Zijian1026/RRTO-CF3DGS
Sparse-View Surface via GS,,arXiv,2026,Sparse-View,Normal-guided depth propagation from high- to low-confidence regions + outlier depth edge-aware regularization; resolves Gaussian discreteness-induced depth discontinuities,
Sparse2DGS,2505.19854,arXiv 2026,2026,Sparse-View,Sparse-view surface reconstruction using 2DGS with dense point cloud init; confidence-aware depth,
StructSplat,2606.28321,ECCV 2026,2026,Sparse-View,Sparse-view generalizable 3DGS without camera parameters; structure-guided reconstruction from HITSZ/DU/BUTD,
TRACE-GS,2608.10286,arXiv,2026,Sparse-View,On-policy trajectory distillation plus privileged geometric conditioning; first to derive on-policy supervision from privileged geometry for sparse-view 3DGS recovery; LUPI setting,
TV-SGS,2609.07734,arXiv 2026,2026,Sparse-View,Gaussian splatting with geometric information propagation via tensor voting under sparse views,
VidSplat,2605.11424,SIGGRAPH 2026,2026,Sparse-View,"Gaussian Splatting has achieved remarkable progress in multi-view surface reconstruction, yet it exhibits notable degradation when only few views are available. Although recent efforts alleviate th...",
Binocular3DGS,2410.18822,NeurIPS 2024,2024,Sparse-View,Binocular disparity-guided depth + GS joint optimization for sparse views,https://github.com/hanl2010/Binocular3DGS
CoR-GS,2405.12110,ECCV 2024,2024,Sparse-View,Co-regularization of two randomly initialized GS fields: co-pruning + pseudo-view augmentation for sparse views,
FewViewGS,2411.02229,NeurIPS 2024,2024,Sparse-View,Multi-stage coarse-to-fine training strategy for few-view Gaussian Splatting,
FSGS,2312.00451,ECCV 2024,2024,Sparse-View,SRF geometric prior + 3DGS for few-shot view synthesis,https://github.com/VITA-Group/FSGS
GaussianObject,2402.10259,CVPR 2024,2024,Sparse-View,Object-centric GS from sparse views with depth-regularized Gaussian initialization,https://github.com/Chenyu-Yang-GOAT/GaussianObject
SCGaussian,2411.03637,NeurIPS 2024,2024,Sparse-View,Structure consistency constraint + geometric regularization for sparse-view GS,https://github.com/prstrive/SCGaussian
4DGS-WAM,2608.25956,arXiv 2026,2026,World Models & Spatial Intelligence,Object-centric World Action Model based on 4D Gaussian Splatting; separately models dynamic objects (policy-predicted future actor states) and static background; bridges past 4DGS reconstruction and future action prediction,
ABot-3DWorld 0,2607.11673,arXiv,2026,World Models & Spatial Intelligence,Universal world model → 3DGS; Spatial Generative Primitive (SGP) + panoramic video reconstruction; multimodal input → explorable 3DGS world; SOTA among open-source methods,
APEIRIA,,arXiv (unverified: arXiv ID not found in arXiv API; pending source check),2026,World Models & Spatial Intelligence,Neuro-symbolic 3D spatial reasoning framework combining 3D MLLM open-vocabulary understanding with programmatic verification; bridges black-box neural reasoning and interpretable symbolic spatial reasoning,
F3DGS,2604.01605,arXiv 2026,2026,World Models & Spatial Intelligence,Federated 3D Gaussian Splatting for decentralized multi-agent world modeling; shared geometric scaffold from registered LiDAR point clouds; enables distributed 3DGS reconstruction without centralized data aggregation,
FlashWorld,2510.13678,arXiv 2025,2026,World Models & Spatial Intelligence,Feed-forward 3DGS world model: single forward pass generates dynamic 3D scenes with real-time interaction,
GWM,2508.17600,arXiv 2025,2026,World Models & Spatial Intelligence,Gaussian World Model: 3DGS as environment dynamics modeling primitive with autoregressive future state prediction for interactive world simulation,
Holi-Spatial,2603.07660,ICML 2026 Oral,2026,World Models & Spatial Intelligence,"Evolving Video Streams into Holistic 3D Spatial Intelligence; fully automated spatial data construction framework from raw video; 4M+ spatial multimodal samples; covers 3D reconstruction, depth, 2D mask, 3D bbox, instance description, 3D grounding, spatial QA (Shanghai AI Lab, NWPU, SJTU)",
HY-World 2.0,2604.14268,arXiv 2026,2026,World Models & Spatial Intelligence,Multi-modal world model framework: text/single-view/multi-view/video inputs → navigable 3DGS world via panorama generation (HY-Pano 2.0) + trajectory planning (WorldNav) + world expansion (WorldStereo 2.0),
MRO-GWM,2606.01950,arXiv 2026,2026,World Models & Spatial Intelligence,Object-centric Gaussian world model for rigid bodies; canonical Gaussian + spatio-temporal transformer predicts rigid body motion; supports model-predictive control,
Spatial-TTT,2603.12255,ECCV 2026,2026,World Models & Spatial Intelligence,Test-time training for streaming spatial intelligence; 2B-parameter model surpasses GPT-5 and Gemini-3-pro on spatial benchmarks; continuous spatial memory update from long-form video streams; 40%+ savings in VRAM and compute,
S²AM3D,,arXiv (unverified: arXiv ID not found in arXiv API; pending source check),2026,World Models & Spatial Intelligence,Scale-controllable 3D point cloud arbitrary part segmentation; first model with continuous resolution adaptation,
Cloth-Splatting,2501.01715,arXiv 2025,2025,World Models & Spatial Intelligence,3D cloth state estimation via prediction-update framework; action-conditioned dynamics model predicts future states + 3DGS updates predicted states; couples mesh-based representation with Gaussian Splatting for differentiable state-to-image map,
ArtMesh,2605.16582,arXiv 2026,2026,Human & Avatar,Part-aware articulated mesh field with restricted Delaunay remeshing + bidirectional vertex-wise motion consistency,
CapTalk,2605.29316,arXiv,2026,Human & Avatar,Text-guided stylization and speech-driven 3D head animation; separate control over style and emotion; dynamic emotion during inference,
COSY,2605.24114,arXiv,2026,Human & Avatar,Compositional GAN generates hair/skin/glasses/torso independently for disentangled head editing,
D-Rex,2604.27871,SIGGRAPH 2026,2026,Human & Avatar,Diffusion post-process relighting for expressive avatars,
EmoZone-Talker,2606.15848,arXiv 2026,2026,Human & Avatar,Novel framework that reformulates audio-driven facial animation as a structured spatial-temporal coordination problem under cross-modal conflicts,
FlexiAvatar,2607.19100,ECCV 2026,2026,Human & Avatar,Unified framework optimizing only visible body regions; SMPL-X tracking + diffusion completion,
G-Skin,2608.01726,arXiv 2026,2026,Human & Avatar,"To address this challenging problem, we propose G-Skin, a novel generative skinning framework designed for expressive and high-fidelity animation with 3D Gaussian representation",
GaussianEmoTalker,2607.00959,arXiv 2026,2026,Human & Avatar,Audio-driven framework for real-time emotional talking head synthesis based on 3D Gaussian Splatting,
GradRig,2609.05127,arXiv 2026,2026,Human & Avatar,Differentiable weights for skinned Gaussian splat deformation; spatial gradient of skinning weights for mesh-free deformation,
HairGPT,2605.08824,SIGGRAPH 2026 Journal Track,2026,Human & Avatar,Strand-as-Language autoregressive modeling for 3D hairstyle synthesis,
Hand-4DGS,2606.19156,arXiv 2026,2026,Human & Avatar,"To address these challenges, we introduce Hand-4DGS, the first feed-forward framework for reconstructing dynamic 4D hands directly from egocentric videos, enabling both fast (~60 FPS) inference and strong generalization",
HeadsUp,2605.04035,ECCV 2026,2026,Human & Avatar,"UV-parameterized 3D Gaussian representation decoupling primitive count from input resolution; trained on 10,000+ subjects for large-scale, feed-forward 3D Gaussian head reconstruction from multi-view captures",https://apple.github.io/ml-headsup/
High-Fidelity Human GS,2604.21714,arXiv,2026,Human & Avatar,SMPL-X geometric priors + region-aware initialization,
HumanSplatHMR,2605.02784,arXiv,2026,Human & Avatar,Joint pose refinement + Gaussian avatar optimization,
Latent Dynamics,2605.21478,arXiv 2026,2026,Human & Avatar,Pose-driven 3DGS avatar with transformer decoder + dynamics residual latent; learned force decomposition (driving/restoring/dissipative) for temporally coherent clothing animation,
Multi-view Consistent 3D Gaussian Head Avatars,2605.25220,CVPR,2026,Human & Avatar,Multi-view consistent 3D Gaussian head avatars from single-view input without multi-view generation,
OASIS,2607.29633,arXiv 2026,2026,Human & Avatar,Tailored 3D Gaussian Splatting framework for single-image hand avatar reconstruction,
PD-GS,2608.05218,arXiv 2026,2026,Human & Avatar,"TextbfPhoneme-Driven Gaussian Splatting (PD-GS), which augments a 3DGS talker with time-aligned phoneme tokens obtained from an automatic ASR and forced-alignment pipeline",
PEAR,,SIGGRAPH 2026,2026,Human & Avatar,Single-image animatable 3D human avatar at 100 FPS; full-body + hand + face reconstruction from single natural image,
PiG-Avatar,2605.20185,arXiv 2026,2026,Human & Avatar,Hierarchical neural-field-guided Gaussian avatars in volumetric canonical space decoupled from template topology; 3D barycentric anchor transport for kinematic coherence; emergent self-organization of anchor density,
ProgressiveAvatars,,SIGGRAPH 2026 TOG,2026,Human & Avatar,Progressive animatable 3D Gaussian avatar generation,
RAGA,2606.29329,arXiv 2026,2026,Human & Avatar,"RAGA, a Ray-Traced Gaussian Shadow Casting formulation based on exact ray-Gaussian line integrals",
S-Avatar,2607.28164,arXiv 2026,2026,Human & Avatar,Novel method for generating photorealistic 3D head avatars from a single image using a diffusion-guided 3D model generation module and strategies for animating 3D Gaussian Splatting (3DGS),
SDTalk,2605.09956,arXiv,2026,Human & Avatar,"High-quality, real-time talking head synthesis remains a fundamental challenge in computer vision. Existing reconstruction- and rendering-based methods typically rely on identity-specific models, l...",
SpatialAvatar-0,2606.15659,arXiv 2026,2026,Human & Avatar,10K-iter layout-preserving per-subject refinement loop that freezes the FLAME-binding and Gaussian count and replaces densification with a three-component anti-spike regularization,
SplitAvatar,2605.25751,arXiv,2026,Human & Avatar,Autoregressive Gaussian splitting via GNN for one-shot head avatar with progressive detail,
GaussianBody,2401.09720,CVPR 2025,2025,Human & Avatar,SMPL-X aligned Gaussian body with part-aware densification,
GaussianTalker,2404.16012,CVPR 2025,2025,Human & Avatar,Audio-driven Gaussian talking head with facial prior and emotion control,
SplatFace,2403.18784,CVPR 2025,2025,Human & Avatar,Face-specific GS with identity-preserving Gaussian anchoring from single image,
SplatPose,2503.05174,CVPR 2025,2025,Human & Avatar,Pose-conditioned Gaussian Splatting for monocular human reconstruction,
SplatTalk,2503.06271,CVPR 2025,2025,Human & Avatar,Audio-driven 3D talking face GS with emotion and style control,
3DGS-Avatar,2312.09228,CVPR 2024,2024,Human & Avatar,Deformable 3DGS for animatable human avatars with pose-conditioned Gaussian deformation,https://github.com/mikeqzy/3DGS-Avatar
ExpressiveGaussianHuman,2407.03204,NeurIPS 2024,2024,Human & Avatar,Expression-coefficient-driven Gaussian deformation fields for expressive human avatars,
GAGAvatar,2410.07971,NeurIPS 2024,2024,Human & Avatar,Generalizable and animatable Gaussian head avatar from monocular video,https://github.com/xg-chu/GAGAvatar
GAS,,arXiv,2024,Human & Avatar,Compression + caching for real-time avatar rendering,
GauHuman,2312.02973,ECCV 2024,2024,Human & Avatar,Human-specific GS with SMPL-constrained Gaussian initialization and pose-aware densification,
GaussianAvatar,,arXiv,2024,Human & Avatar,Pose-driven human body Gaussian representation,
HeadGaS,2312.02902,ECCV 2024,2024,Human & Avatar,Dynamic head GS with blendshape-driven Gaussian deformation for real-time reenactment,
Human3Diffusion,2406.08475,NeurIPS 2024,2024,Human & Avatar,Diffusion + GS dual-driven 3D human avatar reconstruction,https://github.com/YuxuanSnow/Human3Diffusion/
HumanSplat-NIPS,2406.12459,NeurIPS 2024,2024,Human & Avatar,Single-image human GS with SMPL-guided Gaussian binding,
SplatArmor,2311.10812,CVPR 2024,2024,Human & Avatar,LBS-based articulated Gaussian Splatting for human body with twist-aware deformation,
SplattingAvatar,,arXiv,2024,Human & Avatar,Expression-conditioned Gaussian deformation,
BEA-GS,2605.09662,CVPR,2026,Editing,Visible boundary + non-visible extraction losses for near-perfect object extraction from 3DGS,
Capacity-Controlled Stylization,,ECCV,2026,Editing,Capacity-controlled multi-view stylization of 3DGS; style-structure disentanglement,
CoGeo-GS,2608.26656,arXiv 2026,2026,Editing,Concept-driven framework for controllable multi-object removal in 3D scenes,
CoIn,2606.27584,arXiv 2026,2026,Editing,"CoIn, a novel framework that bridges 2D inpainting models and 3DGS through a multi-stage consistency pipeline",
DeSplat,,arXiv 2026,2026,Editing,Decomposed Gaussian Splatting for distractor-free rendering; separates transient objects from static scene,
DiffSoup,2603.27151,arXiv,2026,Editing,Triangle soup as extreme radiance field simplification,https://github.com/kenji-tojo/diffsoup
DReSG,2608.29048,arXiv 2026,2026,Editing,3D-grounded residual-feedback framework for stylized Gaussian splatting,
DS-StyleGaussian,2605.30065,IEEE IVMSP,2026,Editing,Data-sufficient zero-shot 3D style transfer via feature Gaussian splatting + deferred stylization,
Edit3DGS,2606.17432,arXiv 2026,2026,Editing,Unified framework for dynamic 3D head editing that integrates 2D instruction-guided diffusion with 3D Gaussian splatting,
Edit3r,2512.25071,ECCV 2026,2026,Editing,Instant 3D scene editing from sparse unposed images; SAM2-based recoloring strategy for cross-view-consistent supervision; single-pass feed-forward editing,
FaceParts,2605.13853,arXiv 2026,2026,Editing,Unsupervised segmentation and editing of GS avatars with FLAME-anchored transfer,
FluSplat,2604.20038,arXiv,2026,Editing,Feed-forward sparse-view editing without per-scene optimization,
FTSplat,2603.05932,arXiv,2026,Editing,Feed-forward triangle primitive generation,
GeM-NR,2606.05142,arXiv,2026,Editing,Geometry-aware multi-view editing for nonrigid scene changes; training-free depth alignment + query refinement,
GOR-IS,2605.00498,arXiv,2026,Editing,Physically consistent object removal via intrinsic decomposition,https://applezyh.github.io/GOR-IS-project-page/
GS-DIFF,2605.07203,arXiv 2026,2026,Editing,"Scene change detection directly on 3D Gaussian primitives, mapping pixel-level differences to primitive-level updates",
IRIS,2603.15368,arXiv,2026,Editing,Hybrid proxy with analytical ray-primitive intersection,https://github.com/gwilczynski95/iris
MeGAS,2606.23455,arXiv 2026,2026,Editing,Novel framework that incorporates thermomechanical phase-change dynamics into 3D Gaussian Splatting (3DGS),
ObjectMorpher,2603.28152,CVPR 2026,2026,Editing,Deformable 3DGS for 3D-aware interactive image editing,
RoVES,2605.25373,arXiv,2026,Editing,Physics-aware road geometry insertion + 4-DOF vehicle dynamics for driving scene editing,
Semantic-Guided Progressive Object Removal,,arXiv,2026,Editing,DINOv2 semantic-guided block matching + region-wise progressive refinement for high-quality 3D object removal in 3DGS scenes,
SketchFaceGS,2604.19202,CVPR 2026 Highlight,2026,Editing,Sketch-driven face editing via Transformer UV prediction,https://github.com/gogoneural/SketchFaceGS_jittor
SplatShot,2606.01493,arXiv 2026,2026,Editing,Training-free 3D face avatar from single in-the-wild photo; per-step 3D feedback loop: predict clean image -> refit 3DGS -> photometric difference backpropagated to noise estimate,
Super-Gaussian,2608.04475,arXiv,2026,Editing,3DGS interactive scene editing plus NLI-based volume rendering visualization in VR,
SVGS-Edit,2603.28126,arXiv,2026,Editing,"Text-driven 3D scene editing has attracted considerable interest due to its convenience and user-friendliness. However, methods that rely on implicit 3D representations, such as Neural Radiance Fie...",https://amateurc.github.io/svgs.github.io
TOM-GS,2607.22717,arXiv 2026,2026,Editing,Editable video representation that forgoes complex deformations in favor of regular 3D Gaussians equipped with a continuous temporal opacity formulation,
TransSplat,2604.19571,arXiv,2026,Editing,Language-driven editing as unbalanced semantic transport,
WeatherCity,2602.22096,arXiv,2026,Editing,Weather Gaussian representation for 4D urban scene weather editing with physics-driven particle simulation,
VEGS,2407.02945,CVPR 2025,2025,Editing,Video-driven editing of Gaussian Splatting with temporal consistency propagation,
Align Your Gaussians,2312.13763,CVPR 2024,2024,Editing,Text-to-4D with dynamic 3D Gaussians and composed diffusion models,
BAD-Gaussians,2403.11831,CVPR 2024,2024,Editing,Bundle-adjusted deformation Gaussians for consistent editing across views,https://github.com/yccyencheng/BAD-Gaussians
D-MiSo,2405.14276,NeurIPS 2024,2024,Editing,Multi-Gaussians Soup representation for editing dynamic 3D scenes,
FlashSplat,2409.08270,ECCV 2024,2024,Editing,Alpha blending linearity enables 2D-to-3D GS segmentation as linear programming with closed-form solution (50x faster),https://github.com/florinshen/FlashSplat
Frosting,,arXiv,2024,Editing,Decoupled geometry/appearance editing,
GaussCtrl,2403.08733,ECCV 2024,2024,Editing,Depth-conditioned attention + progressive editing for controllable GS generation from text/depth,
Gaussian Grouping,2312.00732,ECCV 2024,2024,Editing,Identity encoding per Gaussian + SAM supervision + 3D spatial consistency for open-world 3D segmentation and editing,
GaussianCut,2411.07555,NeurIPS 2024,2024,Editing,Graph cut algorithm for interactive 3DGS segmentation enabling instance-level editing,
GaussianEditor,,arXiv,2024,Editing,CLIP-guided text/geometry-driven editing,
GeoGaussian,,arXiv,2024,Editing,Mesh-prior-guided Gaussian manipulation,
GScream,2404.13679,ECCV 2024,2024,Editing,Cross-attention feature propagation bridging visible/invisible regions for 3D object removal,
GSDeformer,2405.15491,arXiv,2024,Editing,Cage-based deformation for 3DGS; proxy point cloud bridges cage deformation to Gaussian transform,
InFusion,2404.11613,CVPR 2024,2024,Editing,Inpainting-guided Gaussian Splatting for 3D content insertion and scene completion,
ProEdit,2411.05006,NeurIPS 2024,2024,Editing,Progressive local editing with global consistency maintenance for 3D scenes,
StylizedGS,2404.05220,NeurIPS 2024,2024,Editing,Reference-based controllable scene stylization with Gaussian Splatting,
VR-GS,2401.16663,ECCV 2024,2024,Editing,Physical-based GS editing in VR: real-time Gaussian manipulation with haptic feedback,
3DCodeBench,2606.01057,arXiv 2026,2026,CAD & Reverse Engineering,Systematic benchmark for VLM agents on procedural 3D generation (text/image -> code -> 3D); evaluates 12 VLMs; includes 3DCodeArena human preference ranking platform,
ASSEMCAD,,arXiv,2026,CAD & Reverse Engineering,Axiom-driven framework for production-ready CAD assembly generation from natural language; port-and-mate-based library with executable mate relations + engineering axioms; outperforms code-centric CAD baselines on AssemBench,
BRepCLIP,2606.05515,arXiv 2026,2026,CAD & Reverse Engineering,Contrastive multi-modal pre-training with BRep primitives for CAD understanding; cross-modal retrieval,
BrepGaussian,2602.21105,CVPR 2026,2026,CAD & Reverse Engineering,3DGS + B-rep CAD reconstruction to parametric STEP models,
CADDreamer,,CVPR 2025 Highlight,2026,CAD & Reverse Engineering,CAD object generation from single-view images; parametric primitive sequence generation from 2D input,
CADFit,2605.01171,arXiv,2026,CAD & Reverse Engineering,"Despite recent progress, recovering parametric CAD construction sequences from geometric input, such as meshes or point clouds, is a key challenge for design and manufacturing, as existing CAD reco...",
CADFS,2605.01925,CVPR 2026,2026,CAD & Reverse Engineering,Large-scale CAD program dataset + LLM-assisted CAD understanding,
CADSplat,2609.18473,arXiv 2026,2026,CAD & Reverse Engineering,"Framework that reconstructs photorealistic, geometrically accurate digital twins from sparse (<15 views), wide-baseline posed images of an object by regularizing 3D Gaussian Splatting (3DGS) with an explicit CAD shape prior",
KDH-CAD,2606.01702,arXiv 2026,2026,CAD & Reverse Engineering,Knowledge-data hybrid CAD learning: pretrained foundation model + structured domain knowledge from textbooks + small labeled CAD data; 92.6% accuracy with only 250 samples,
MidSurfNet,2606.01891,arXiv 2026,2026,CAD & Reverse Engineering,Learning-augmented mid-surface abstraction: neural face pairing module + interference implicit field (SDF intersection) for arbitrary offset control; 87.32% pairing accuracy,
RealCAD,2608.30617,arXiv,2026,CAD & Reverse Engineering,Real-world image to CAD reconstruction; handles domain gap and parametric deviation from imperfect photographs; code and dataset released,
RORA,2608.04842,arXiv,2026,CAD & Reverse Engineering,Single static video to articulated object end-to-end pipeline; mixed 3DGS plus mesh; convex decomposition plus automatic joint suggestion algorithm; deployed in Unreal Engine and Isaac Sim,
SEIG,2606.02580,arXiv 2026,2026,CAD & Reverse Engineering,VLM generates executable Blender programs from single image via staged decomposition (geometry->materials->composition->lighting); no differentiable rendering or multi-view supervision needed,
UniCAD,2606.05058,arXiv,2026,CAD & Reverse Engineering,Unified multi-modal multi-task CAD benchmark + UniCAD-MLLM for point/text/image to CAD reconstruction,
UniMGS,,AAAI 2026,2026,CAD & Reverse Engineering,Single-pass rasterization for both mesh and Gaussians simultaneously,
Zero-to-CAD,2604.24479,arXiv 2026,2026,CAD & Reverse Engineering,Agentic synthesis of parametric CAD programs; 2B VLM trained on 1M+ synthetic CAD programs,
BrepGiff,,arXiv 2026,2025,CAD & Reverse Engineering,Lightweight generation of complex B-rep with 3D GAT Diffusion; graph attention network diffusion for B-rep topology+geometry (IEEE Xplore),
MaGS,,ICCV 2025,2025,CAD & Reverse Engineering,Mesh-adsorbed Gaussians; deform mesh → Gaussians follow,
Vol3DGS,,CVPR 2025,2025,CAD & Reverse Engineering,Physically accurate volume-consistent rendering resolving splatting/volume inconsistency via volume rendering integral for 3DGS,
2DGS,2403.17888,SIGGRAPH 2024,2024,CAD & Reverse Engineering,2D disks on surfaces enabling direct mesh extraction via Poisson reconstruction,
3DTV,2604.11211,arXiv,2026,Cross-Domain,Feedforward 3-camera interpolation at 40 FPS,https://stefanmschulz.github.io/3DTV_webpage/
Aes3D,2605.05155,arXiv,2026,Cross-Domain,First aesthetic assessment framework for 3DGS scenes; Aesthetic3D dataset + Aes3DGSNet that predicts scores directly from Gaussian primitives,
AquaFlow,2608.22906,arXiv 2026,2026,Cross-Domain,"To address these challenges, we propose AquaFlow, a monocular Gaussian Splatting streaming reconstruction framework for efficient and high-fidelity underwater reconstruction",
AsyncEvGS,2605.07192,arXiv 2026,2026,Cross-Domain,Asynchronous event camera + RGB dual-system for motion-blurred 3DGS via cross-domain pose estimation,
AudioGS,2604.08967,arXiv,2026,Cross-Domain,Spectrogram-based audio Gaussian Splatting for sound field reconstruction,
BiSplat-WRF,2604.25945,IEEE ICC 2026 Workshop,2026,Cross-Domain,Gaussian Splatting for wireless radiance field reconstruction,
CoherentRaster,2605.04509,arXiv,2026,Cross-Domain,Subpixel-level 3DGS rasterization for light field displays,
CORF-GS,2607.25569,arXiv 2026,2026,Cross-Domain,"To bridge this gap, we propose CORF-GS, a real-time WRF reconstruction framework that processes sequential optical and radio frequency (RF) keyframes",
CVQPG,2609.11434,arXiv 2026,2026,Cross-Domain,Complex-valued quadratic phase Gaussian splatting for hologram representation; replaces standard 2D Gaussian in 2DGS,
DVSM,2605.29891,arXiv,2026,Cross-Domain,Decoder-only view synthesis model; shared reconstruction/rendering weights; SOTA on NVS benchmarks,
EmoTaG,2603.21332,CVPR 2026,2026,Cross-Domain,Few-shot emotion-aware talking head on Gaussian Splatting,https://emotag26.github.io/
EndoGSim,2605.16022,MICCAI 2026,2026,Cross-Domain,MLLM-guided 4DGS + differentiable MPM for physics-aware endoscopic scene reconstruction and simulation,
EndoMD-SLAM,2608.08949,arXiv 2026,2026,Cross-Domain,"To address this limitation, we propose EndoMD-SLAM, a framework designed to maintain stability under optical degradation through specialized tracking and mapping mechanisms",
ESVR,2608.05564,IEEE VIS 2026,2026,Cross-Domain,3D ellipsoid sparse volume rendering; learns directly from raw volume data; per-primitive ray sampling; up to 4 orders of magnitude compression; 43-223 FPS,
ExtraGS,2607.12785,arXiv 2026,2026,Cross-Domain,Framework for enhancing endoscopic view extrapolation through diffusion-guided 3D Gaussian Splatting,
Fake3DGS,2604.27590,ICPR 2026,2026,Cross-Domain,Benchmark for 3D manipulation detection in neural rendering,https://github.com/iot-unimore/Fake3DGS
FieryGS,2605.00177,ICLR 2026,2026,Cross-Domain,Physics-integrated combustion simulation + 3DGS rendering,https://pku-vcl-geometry.github.io/FieryGS/
GaussianPile,2603.20611,CVPR 2026,2026,Cross-Domain,Slice-based volumetric reconstruction via sparse 3DGS + imaging system-aware focus model; 11x faster than NeRF; 16x compression,
GeoFovea-GS,2607.12641,arXiv 2026,2026,Cross-Domain,"To address this issue, this paper proposes GeoFovea-GS as a geometry-aware cross-layer framework for communication-efficient wireless aerial VR",
GS-DOT,2604.23675,arXiv,2026,Cross-Domain,Diffuse optical tomography via Gaussian Splatting,
GS-STVSR,2604.18047,arXiv,2026,Cross-Domain,"Continuous Spatio-Temporal Video Super-Resolution (C-STVSR) aims to simultaneously enhance the spatial resolution and frame rate of videos by arbitrary scale factors, offering greater flexibility t...",
GS-Underwater,2608.25483,arXiv 2026,2026,Cross-Domain,Controlled cross-regime study of Gaussian splatting underwater; benchmarks five systems under shared protocol across turbidity levels; finds restoration pre-pass beats medium-aware systems,
GSBF,2608.05896,arXiv 2026,2026,Cross-Domain,"Recognizing that radio propagation is intrinsically governed by the physical geometry, we develop a 3D Gaussian splatting for environment-aware beamforming (GSBF) pipeline based on multi-modal data",
gsplat,,arXiv,2026,Cross-Domain,CUDA-accelerated 3DGS rasterization library; 4x VRAM savings; 10% training speedup; production-grade real-time rendering; part of nerfstudio project,
LagrangianSplats,2605.09299,arXiv,2026,Cross-Domain,"Reconstructing 3D fluid velocity fields from sparse 2D video observations is a highly ill-posed inverse problem, demanding both transport consistency with observed motion and physical validity unde...",
LINGO,2609.22849,arXiv 2026,2026,Cross-Domain,Latent initialization and gradient optimization framework for sparse-view X-ray novel-view synthesis and CT reconstruction with 3DGS.,
MedGSSR,2609.06874,arXiv 2026,2026,Cross-Domain,Generalizable medical image super-resolution 3D reconstruction via hierarchical feed-forward Gaussian splatting,
MSGS,2604.13340,arXiv,2026,Cross-Domain,Multispectral 3D Gaussian Splatting,
NAB-GS,2508.15151,MICCAI,2026,Cross-Domain,Negative Alpha Blending for zero-shot 3D medical segmentation,
NemoSplat,2608.22888,arXiv 2026,2026,Cross-Domain,"To overcome these limitations, we present NemoSplat, the first feed-forward 4D Gaussian Splatting framework tailored for media-aware dynamic reconstruction directly from uncalibrated marine videos",
OCH3R,2605.13018,arXiv,2026,Cross-Domain,Object-centric scene understanding is a fundamental challenge in computer vision. Existing approaches often rely on multi-stage pipelines that first apply pre-trained segmentors to extract individu...,
OctCGS,2605.22961,arXiv 2026,2026,Cross-Domain,Octree-contextual Gaussian splatting framework that explicitly models the order of bounce jointly over Tx/Rx positions and carrier frequencies,
OpenSpatial,2604.07296,arXiv,2026,Cross-Domain,Principled open-source data engineering for 3DGS,
PAGS,2608.25472,arXiv 2026,2026,Cross-Domain,"PAGS, a differentiable framework for blind autofocusing PACT via speed-of-sound-adaptive Gaussian splatting",
PanoLess,2607.25362,arXiv 2026,2026,Cross-Domain,Gaussian-splat-based framework that reconstructs the surrounding environment as a distant illumination map from images captured on only one side of a reflective surface,
PanoLOG,2607.08769,arXiv,2026,Cross-Domain,Geometry and gradient-based partitioning for panoramic outdoor 3DGS reconstruction; first panoramic outdoor benchmark Pano360,
PatchPoison,2604.13153,arXiv,2026,Cross-Domain,Poisoning multi-view datasets to degrade 3D reconstruction (Security),
PG-3DGS,2605.11266,arXiv,2026,Cross-Domain,"Recent advances in Gaussian Splatting have enabled fast, high-fidelity 3D scene generation, yet these methods remain purely visual and lack an understanding of how shapes behave in the physical wor...",
RESPIRE,2604.28179,arXiv,2026,Cross-Domain,CT-informed mesh-anchored GS for dynamic bronchoscopy,https://asdunnbe.github.io/RESPIRE/
RGS,2604.27552,arXiv,2026,Cross-Domain,Residual wavelet-GS for ultra sparse-view CBCT reconstruction,
RPC-GS,2606.06690,arXiv 2026,2026,Cross-Domain,First Gaussian Splatting framework for satellite imagery that operates natively with Rational Polynomial Camera (RPC) models,
RxGS,2605.24290,arXiv 2026,2026,Cross-Domain,"RxGS, which achieves receiver-generalizable synthesis within a single unified model",
SandSim,2604.27572,arXiv,2026,Cross-Domain,Curve-guided GS for sand painting process reconstruction,
SatSplat,2606.28581,arXiv 2026,2026,Cross-Domain,"First framework to adapt 2D Gaussian Splatting (2DGS) to satellite photogrammetry, with online camera adjustment",
SatSplatDiff,2606.27223,arXiv 2026,2026,Cross-Domain,"To address these limitations, we propose SatSplatDiff, which aims to minimize geometric degradation prevalent in generative refinement",
Scene2Sound,2608.00463,arXiv 2026,2026,Cross-Domain,"Task of generating a spatially consistent soundscape for a given 3DGS world through auditory grounding, identifying which objects in the world should emit sound and anchoring each to a persistent 3D position",
Sensor2Sensor,2605.22809,CVPR,2026,Cross-Domain,Cross-embodiment sensor conversion via 4DGS; translates dashcam to AV sensor suite using diffusion,
Shape-guided X-ray GS,2609.10376,arXiv 2026,2026,Cross-Domain,Shape-guided 3DGS for sparse-view X-ray 3D reconstruction; reduces radiation exposure via shape priors,
Splat-based,2608.13159,arXiv 2026,2026,Cross-Domain,"Novel physics-inspired, self-calibrating metal artifact reduction method that efficiently reconstructs 3D CBCT volumes while correcting beam hardening artifacts",
SplAttN,2605.01466,ICML 2026 Spotlight,2026,Cross-Domain,Gaussian soft splatting for point cloud completion,https://github.com/zay002/SplAttN
StereoSplat+,2607.08808,IROS 2026,2026,Cross-Domain,Feed-forward stereo GS with diffusion-assisted progressive inference; one-step diffusion enhancer + stereo cost-volume + triplane 3D volume fusion,
Swimm3R,2608.00950,arXiv 2026,2026,Cross-Domain,Unified framework that combines medium-aware structure-from-motion (SfM) with Underwater Beta Splatting to address scattering- and attenuation-induced failures in underwater 3D reconstruction,
SyncSpace,2607.10050,arXiv,2026,Cross-Domain,Layout-conditioned 3DGS for MR space reskinning; coarse-to-fine registration + hand-tracked engulfment interaction,
TR-GS,2608.16042,arXiv 2026,2026,Cross-Domain,Gaussian-splatting framework for sparse view CT volumetric rendering,
TwinPose,,SIGGRAPH 2026 TOG,2026,Cross-Domain,Multi-view 3D pose estimation via person-specific subspaces,https://github.com/zgspose
Visionary,2512.08478,arXiv 2025,2026,Cross-Domain,WebGPU + 3DGS world model carrier: first Web-native platform supporting 4DGS/neural avatars/generative post-processing in real-time,
Wat3R,2607.08772,ECCV 2026,2026,Cross-Domain,Cross-domain semi-supervised underwater 3D geometry learning without annotations; teacher-student architecture adapts feed-forward reconstruction from air to underwater,
XClipGS,2608.07760,arXiv 2026,2026,Cross-Domain,Treats these as two separate problems: the render-time clip operator and supervision of the hidden interior,
XFreq-GS,2605.11432,arXiv,2026,Cross-Domain,"Channel modeling is fundamental to the analysis, design, and optimization of wireless communication systems, which, however, accurate wireless channel modeling remains challenging, especially given...",
EndoGS,2401.11535,CVPR 2025,2025,Cross-Domain,Endoscopic scene reconstruction with GS for surgical navigation,
RDSplat,2512.06774,arXiv,2025,Cross-Domain,Robust watermarking against diffusion editing,
SurfFill,2512.03010,arXiv,2025,Cross-Domain,LiDAR point cloud completion via Gaussian surfel splatting; ambiguity heuristic for LiDAR beam divergence artifacts; divide-and-conquer for building-sized completion,
Thermal3D-GS,2409.08042,ECCV 2024,2025,Cross-Domain,"Thermal3D-GS: physics-induced 3D Gaussians for thermal-infrared novel-view synthesis; models atmospheric transmission and thermal conduction with a temperature-consistency constraint; introduces TI-NSD benchmark (6,664 thermal frames)",
DDGS-CT,2406.02518,NeurIPS 2024,2024,Cross-Domain,Direction-disentangled X-ray volume rendering with Gaussian acceleration for CT,
Habitat-GS,,arXiv,2024,Cross-Domain,3DGS-rendered simulator for robot navigation training,
R2-Gaussian,2405.20693,NeurIPS 2024,2024,Cross-Domain,GS adapted for Radon transform + X-ray volume rendering for tomographic reconstruction,https://github.com/Ruyi-Zha/r2_gaussian
3DThinkVLA,2606.04436,arXiv,2026,Simulation,Endowing VLA models with latent 3D priors via 3D-thinking-guided co-training; SOTA on LIBERO,
AGILE,2602.04672,SIGGRAPH,2026,Simulation,Agentic generation for hand-object interaction reconstruction from video; VLM guides generative model,
ArtiTwinSplat,,arXiv,2026,Simulation,Interactable Digital Twin Reconstruction via Gaussian Splatting from RGB-D videos; enables interaction with reconstructed digital twins; agent-system integration for articulated object manipulation,
DISCOVERSE,,arXiv,2026,Simulation,3DGS + MuJoCo unified Real2Sim2Real robot learning framework; scene-level + object-level digital twin,
Dynamic Mesh-Gaussian,2606.00444,arXiv 2026,2026,Simulation,Dual-representation framework: fixed-topology mesh for physics + Gaussian splatting for rendering; 4.65x speedup over varying-topology baseline,
FLUIDSPLAT,2605.18866,arXiv 2026,2026,Simulation,Physical flow field reconstruction via anisotropic GS partition-of-unity; proven Sobolev approximation rate,
LaGSplat,2608.16324,arXiv 2026,2026,Simulation,"Framework that infers interactive, physics-governed dynamics from one or a few monocular videos",
MonoPhysics,2605.30320,arXiv,2026,Simulation,Monocular inverse physics using differentiable MPM + 3DGS; three visual-physical bridges,
PhyGenHOI,2605.30268,arXiv,2026,Simulation,Physically-aware 4D HOI generation; MDM + MPM + 3DGS unified; windowed attraction + re-simulation,
SimuScene,2606.03994,arXiv,2026,Simulation,Physics-in-the-loop compositional 3D scene reconstruction from single image; physics engine as diagnostic,
WildFireGS,2608.11100,arXiv 2026,2026,Simulation,"Particle-based combustion model that operates natively on Gaussian representations, simulating ignition, heat transfer, combustion, and flame propagation across complex forest structures",
2DGS-Planner,2610.11752,arXiv 2026,2026,Embodied AI & Robotics,"Rasterization-based path planner for ground robots that reads planning-relevant geometry from a 2DGS map; multi-view attribution converts rendered normal dispersion into structural scores, with orthographic and cylindrical queries guiding adaptive node sampling and clearance estimation.",https://2dgs-planner.github.io/
AirSplan,2609.21226,arXiv 2026,2026,Embodied AI & Robotics,Adopts a normalized variant of 3D Gaussian Splatting that encodes high-fidelity scene geometry,
ArtGS,,IEEE 2026,2026,Embodied AI & Robotics,3D Gaussian Splatting for interactive visual-physical modeling and manipulation of articulated objects; bone-driven deformation + contact-aware rendering,
FastBridge,2607.01200,arXiv 2026,2026,Embodied AI & Robotics,Fast quadrotor flight requires safe obstacle avoidance under tight onboard compute limits,
FlyMirage,2605.19600,arXiv 2026,2026,Embodied AI & Robotics,"To address these challenges, we introduce FlyMirage, a highly scalable and fully automated data generation pipeline for aerial VLN",
Forecast-GS,2605.11144,arXiv,2026,Embodied AI & Robotics,"We introduce Forecast-aware Gaussian Splatting (Forecast-GS), a predictive 3D representation framework for language-conditioned robotic manipulation. While recent manipulation systems have made pro...",
FreeArtGS,2603.22102,arXiv 2026,2026,Embodied AI & Robotics,Articulated Gaussian Splatting under free-moving scenario; scalable articulated object reconstruction,
GaussLite,2606.30809,arXiv 2026,2026,Embodied AI & Robotics,Task-driven 3DGS mapping system that conditions its representation density on a natural-language task specification,
GaussMemory,2608.14986,arXiv 2026,2026,Embodied AI & Robotics,"Paradigm shift from passive storage to active, task-driven spatial memory",
GS-Playground,2604.25459,RSS 2026,2026,Embodied AI & Robotics,Batch 3DGS + parallel physics at 10^4 FPS for robot learning,https://gsplayground.github.io
GS-Surrogate,2604.06358,arXiv,2026,Embodied AI & Robotics,Deformable GS surrogate for ensemble simulation exploration,
GS-World,,arXiv 2025,2026,Embodied AI & Robotics,World model generative simulation + Engine-driven Sim2Real VLA: 3DGS as differentiable simulation engine for robot skill learning,
GSDrive,2604.28111,arXiv,2026,Embodied AI & Robotics,Reinforcing driving policies with 3DGS environment for high-fidelity RL training,
GSMem,2603.19137,arXiv,2026,Embodied AI & Robotics,"Effective embodied exploration requires agents to accumulate and retain spatial knowledge over time. However, existing scene representations, such as discrete scene graphs or static view-based snap...",
LeGS,2605.00408,arXiv 2026,2026,Embodied AI & Robotics,Loco-manipulation via Embodied Gaussian Splatting; mesh foreground (robot/objects) + 3DGS background; procedural motion primitive generator; VLA policy trained purely in LEGS,
LiftNav,2605.31376,arXiv 2026,2026,Embodied AI & Robotics,"Hybrid navigation framework built on GSFusion's TSDF+GS dual map, augmented with a real-time pipeline of YOLO-based detection, TSDF-based 3D lifting, and B-spline trajectory optimization",
LOGOS,2606.21527,arXiv 2026,2026,Embodied AI & Robotics,"To overcome these bottlenecks, we propose LOGOS, a LiDAR-only unified tiny obstacle segmentation system, which models the road surface as a continuous mixture of 2D Gaussian primitives and distinguishes tiny obstacles via",
ManiSplat,2606.10645,arXiv 2026,2026,Embodied AI & Robotics,"To address these challenges, we introduce ManiSplat, a unified framework that reconstructs controllable and decoupled Gaussian digital twins directly from monocular ego-view robotic videos",
NovaSplat,2610.13309,arXiv 2026,2026,Embodied AI & Robotics,Model-free 6D object pose estimation pipeline that trains a 3DGS model from a short onboarding video to serve as both template source and differentiable refinement backbone; introduces the faster NovaSplat-Lite inference variant.,https://github.com/POSE-Lab/novasplat
PARTICULATE,,CVPR 2026,2026,Embodied AI & Robotics,"Feed-forward 3D object articulation from static mesh; auto-infer movable structure including parts, connections, and motion axes; Oxford / Cambridge / NTU",
PG-Pose,2609.07231,arXiv 2026,2026,Embodied AI & Robotics,PG-Pose: geometry-aware framework combining Planar-based Gaussian Splatting (PGS) reconstruction with geometry-driven pose optimization; generalizable 6D pose of textureless objects without CAD priors,
PolyMerge,2606.16232,arXiv 2026,2026,Embodied AI & Robotics,"To address these limitations, we propose PolyMerge to convert a large, photorealistic 3D Gaussian Splatting (3DGS) model of a scene into a lightweight representation of convex polytopes whose union provably over-approximates all",
ProbSplat,2608.13143,arXiv 2026,2026,Embodied AI & Robotics,"This paper presents ProbSplat, a Compute-in-Memory (CIM)-inspired architecture based on programmable and energy efficient floating-gate inverter columns for probabilistic computing",
Real2Sim,2605.13591,arXiv 2026,2026,Embodied AI & Robotics,4DGS + differentiable MPM solver for physics-aware autonomous driving simulation,
RefGlass-GS,2606.28826,arXiv 2026,2026,Embodied AI & Robotics,"Therefore, we propose RefGlass-GS, a fusion framework that enables end-to-end UAV-based photorealistic, semantic, and interactive digitization of reflective glass facades",
RIDE,2609.11079,arXiv 2026,2026,Embodied AI & Robotics,Relocalization-informed depth estimation with 3DGS; render-match-PnP relocalization supports dense depth estimation,
SA-ResGS,2601.03024,ECCV 2026,2026,Embodied AI & Robotics,Self-augmented residual 3DGS for next-best-view selection; SA-Points via triangulation for coverage estimation; first residual learning for 3DGS to improve gradient flow in sparse regions,
SceneSmith,2602.09153,ICML 2026 Spotlight,2026,Embodied AI & Robotics,"Hierarchical agentic framework for simulation-ready indoor scene generation from natural language; designer-critic-orchestrator VLM agents; text-to-3D + dataset retrieval + physics property estimation; 3-6x more objects than prior methods, <2% collisions, 96% physics-stable",
SemSafe-3DGS,2609.19330,arXiv 2026,2026,Embodied AI & Robotics,Semantic risk-aware safe-active perception navigation framework in attributed 3D Gaussian maps.,
Splat-CBF,2609.23100,arXiv 2026,2026,Embodied AI & Robotics,Risk-aware control barrier function for safe next-best-view control in 3D Gaussian-Splat maps.,
Splat2Act,2610.13945,arXiv 2026,2026,Embodied AI & Robotics,"Automated real-to-sim-to-real pipeline that personalizes a pretrained VLA policy to a specific scene and object set using only a handheld phone scan and object images; builds a photorealistic digital twin with 3DGS and collision geometry, discovers receptacles, and generates pickup demonstrations with a privileged simulation oracle.",
SplatCtrl,2607.08948,ICRA 2026,2026,Embodied AI & Robotics,Perception-action coupling via GS + reactive robot control; isotropic Gaussian → continuous SDF → control barrier functions; collision-free manipulation,
SpotlessGS,2608.14713,arXiv 2026,2026,Embodied AI & Robotics,"Robots operating in dark or poorly lit environments rely on onboard lights, which often produce uneven illumination that degrades downstream perception tasks",
TAIL-Safe,2605.01195,arXiv,2026,Embodied AI & Robotics,Safety monitoring for IL policies using 3DGS digital twin,
GS-ID,,ICCV 2025,2025,Embodied AI & Robotics,GS-based illumination decomposition: real-scene light field factorization for embodied intelligence and autonomous driving,
Splat-Nav,2403.02751,CVPR 2025,2025,Embodied AI & Robotics,GS-based navigation with Gaussian-anchored topological maps,
SplatSim,2409.10161,CVPR 2025,2025,Embodied AI & Robotics,GS-based sim-to-real transfer for robotic manipulation with photorealistic rendering,
VR-Robo,2502.01536,RAL 2025,2025,Embodied AI & Robotics,"Recent success in legged robot locomotion is attributed to the integration of reinforcement learning and physical simulators. However, these policies often encounter challenges when deployed in rea...",https://github.com/zst1406217/VR-Robo
FreeMoCap,,arXiv,2024,Embodied AI & Robotics,"Open-source markerless motion capture from webcams (AGPL-3.0, 8.3k stars); drives 3DGS avatars; dual-use multi-cam for mocap + 3DGS training",https://github.com/freemocap/freemocap
GaussianGrasper,2403.09637,IEEE T-RO 2024,2024,Embodied AI & Robotics,"Constructing a 3D scene capable of accommodating open-ended language queries, is a pivotal pursuit, particularly within the domain of robotics. Such technology facilitates robots in executing objec...",
GaussNav,2403.11625,CVPR 2024,2024,Embodied AI & Robotics,GS-based navigation with language-guided semantic Gaussian maps for embodied agents,
GIC,2406.14927,NeurIPS 2024,2024,Embodied AI & Robotics,Gaussian-Informed Continuum for physical property identification and differentiable simulation,https://github.com/Jukgei/gic
GraspSplats,2409.02084,CoRL 2024,2024,Embodied AI & Robotics,The ability for robots to perform efficient and zero-shot grasping of object parts is crucial for practical applications and is becoming prevalent with recent advances in Vision-Language Models (VL...,
ManiGaussian,2403.08321,ECCV 2024,2024,Embodied AI & Robotics,Dynamic Gaussian Splatting for multi-task robotic manipulation — Gaussian world model predicts optimal actions,https://manigaussian.github.io
3DReflecNet,2605.10204,CVPR,2026,Robustness,Large-scale dataset for 3D reconstruction from reflection-corrupted images,
ClearGS,2609.31509,arXiv 2026,2026,Robustness,Reliability-aware Gaussian Splatting from handheld videos; Reliability-aware View Allocation (RVA) with graded supervision weights + Render-Guided In-Video Restoration (RIVR) with no-reference restoration expert + Full-Trajectory Repair Consolidation; SOTA on GS2E and GSOTM,
Dehaze-GaussianImage,2606.16163,arXiv 2026,2026,Robustness,"To address these issues, we propose Dehaze-GaussianImage, the first zero-shot framework that introduces 2D Gaussian Splatting (2DGS) into the image dehazing domain to break the traditional pixel-grid processing paradigm",
DelowlightSplat,2605.26629,arXiv,2026,Robustness,Lowlight Adapter + cost-volume inference predicts clean Gaussians from degraded inputs,
DualPhys-GS,2508.09610,arXiv 2026,2026,Robustness,Dual physically-guided 3DGS for underwater scene reconstruction; joint attenuation-scattering physics + lighting decomposition,
Fi-Gaussian,2606.16168,arXiv 2026,2026,Robustness,"To address these issues, we propose Fi-Gaussian, a frequency-aware implicit Gaussian splatting network for single image dehazing",
FreDeGS,,Springer 2026,2026,Robustness,Frequency-guided 3DGS for scene deblurring; frequency-aware decomposition for motion-blurred input reconstruction,
FujinSplat,2609.06017,arXiv 2026,2026,Robustness,RAW-domain Gaussian splatting seeing through smoke; separates participating medium and ISP in RAW domain,https://github.com/I2WM/FujinSplat
GlowGS,2605.23602,CVPR Findings,2026,Robustness,Diffusion + VFM semantic features for degraded-image 3DGS reconstruction,
JADE-GS,2607.14990,arXiv 2026,2026,Robustness,Formulates the combination of these priors as spatial evidence allocation,
Mind the Gap,,arXiv,2026,Robustness,Reveals 3-12 dB interpolation-extrapolation gap in standard 3DGS evaluation; proposes fair match count protocol + spatial preservation benchmark toolkit (16 scenes),
NIRRGB-GS,,AISY 2026,2026,Robustness,Near-infrared assisted low-light scene reconstruction and enhancement via multi-modal NIR+RGB Gaussian Splatting,
NOVA-GS,2609.12682,arXiv 2026,2026,Robustness,Noise-aware view-consistent 3DGS for low-light novel view synthesis; handles severe sensor noise and low SNR,
Per-View Distractor Filtering,2608.26951,arXiv 2026,2026,Robustness,Training-free distractor filtering in feed-forward 3DGS; excludes per-view Gaussians and renders with remaining representation to identify inconsistent content; feature similarity + rendering-based verification,
PriSplat,,ECCV 2026,2026,Robustness,Propagating reliable multi-view information for distractor-free 3DGS; 3D-aware inpainting engine from view synthesis prior; geometry-aware support view selection; mask-aware fast weight update,
SharpSplat,,arXiv,2026,Robustness,SAM3-based building edge extraction + rendered gradient-edge alignment; sharp building boundary reconstruction in UAV 3DGS scenes,
SSA-3DGS,,arXiv,2026,Robustness,Unsupervised screen-space artifact removal for 3DGS; joint optimization of 3D scene + learnable 2D overlay; cross-view geometric consistency decouples artifacts from geometry; up to +9 dB PSNR,
TDg,2605.30328,ISPRS,2026,Robustness,Thermal-to-Depth GS using only thermal images for depth estimation,
Tri-DehazeGS,2609.11223,arXiv 2026,2026,Robustness,Scene-medium decoupled 3DGS with transmittance-aware optimization for hazy multi-view reconstruction,
Underwater360,2605.26447,arXiv,2026,Robustness,Omnidirectional GS + physics-based appearance-medium decoupling for underwater panoramic scenes,
View-Structured Conformal Prediction,2609.10307,arXiv 2026,2026,Robustness,View-structured conformal prediction for 3DGS; certifies rendered view meets prediction coverage via statistical guarantee,
WilLaGS,2608.28240,arXiv 2026,2026,Robustness,"To address these limitations, we propose textbfWilLaGS, a unified framework for robust 3D scene reconstruction and generative appearance synthesis under unconstrained settings",
3DEditSafe,2605.15398,arXiv 2026,2026,Security,First safety-regularized 3D editing framework constraining NSFW semantic propagation; 3D safety regularization + safe semantic projection + residue suppression,
4D-GSW,2605.22342,arXiv,2026,Security,Kinematic-aware watermarking via STC metric + HMM-MRF energy for spatio-temporal consistency in 4DGS,
BitC-3DGS,2605.29583,arXiv,2026,Security,High-capacity 3DGS watermarking via bit-plane complement coding,
FlyMark,2610.11364,arXiv 2026,2026,Security,"Training-free invisible watermarking that writes a keyed message into 3DGS parameters using carrier directions derived from a fruit-fly connectome; extraction needs only cone queries, rounding, and majority voting while preserving geometry and appearance bit-identically.",
GaussTrace,2606.10612,ICML 2026,2026,Security,3DGS provenance analysis via evidence-driven LLM reasoning; directed provenance graphs from Gaussian scene attributes for IP protection and forensics,
GhostSplat,2608.29184,arXiv 2026,2026,Security,Input-triggered backdoor that installs such behavior in feed-forward 3DGS,
GuardMarkGS,2605.12919,arXiv 2026,2026,Security,"3D Gaussian Splatting (3DGS) is becoming a practical representation for novel view synthesis, but its growing adoption, together with rapid advances in instruction-driven 3DGS editing, also exposes...",
NGS-Marker,2608.17447,arXiv 2026,2026,Security,Novel native watermarking framework for 3DGS,
GaussianMarker,2410.23718,NeurIPS 2024,2024,Security,Uncertainty-aware watermark embedding + robust extraction for 3DGS copyright protection,
GeometryCloak,2410.22705,NeurIPS 2024,2024,Security,Geometric perturbation copyright watermark embedding into Gaussians preventing TGS-based 3D reconstruction,https://github.com/qsong2001/Geometry-Cloak
GS-Hider,2405.15118,NeurIPS 2024,2024,Security,"Steganography embedding into Gaussian parameters for 3D message hiding, visually lossless",
Graph-GSReg,2606.29782,arXiv 2026,2026,Ontology & Knowledge Representation,"Registers and merges multiple 3DGS scenes into one unified Gaussian representation by constructing and matching 3D scene graphs, without per-Gaussian correspondence training.",
GraphWrit3R,2609.31595,arXiv 2026,2026,Ontology & Knowledge Representation,"End-to-end 3D scene graph writing without multi-stage pipelines, ground-truth object annotations, or proprietary models; open-source and robust to error propagation.",
GRASP,2606.12910,arXiv 2026,2026,Ontology & Knowledge Representation,Grounded reasoning and symbolic planning for open-vocabulary tabletop manipulation: a pretrained VLM translates natural-language queries into PDDL goals for real-time task and motion planning.,
HiWE,2609.39323,arXiv 2026,2026,Ontology & Knowledge Representation,"Hierarchical world knowledge model with visual keypoint enhancement connecting point-based visual grounding, language planning, and depth-lifted semantic 3D representation for zero-shot path planning.",
KLDrive,2603.21029,arXiv 2026,2026,Ontology & Knowledge Representation,"Knowledge-graph-augmented LLM reasoning for fine-grained 3D scene question answering in autonomous driving, reducing hallucination with structured scene facts.",
NEUROSYMLAND,2607.02277,arXiv 2026,2026,Ontology & Knowledge Representation,Neuro-symbolic UAV landing-site assessment combining probabilistic semantic scene graphs with symbolic safety constraints; edge-deployable with transparent safety decisions.,
OntoPlan,2610.07649,arXiv 2026,2026,Ontology & Knowledge Representation,"Ontology-grounded scene representation aligning objects, spaces, relations and states in a shared symbolic vocabulary for scalable long-horizon robot task planning.",
Prior-SG,2608.06170,arXiv 2026,2026,Ontology & Knowledge Representation,"Casts hierarchical scene graph generation as probabilistic alignment of task priors, working in arbitrarily-structured open-plan environments beyond wall-separated geometric heuristics.",
SATURN,2606.22694,arXiv 2026,2026,Ontology & Knowledge Representation,Neuro-symbolic framework for perspective-aware compositional spatial reasoning: reconstructs approximate 3D scenes and composes soft spatial predicates via a training-free Pythonic symbolic executor.,
Vision-TL-Action,2607.26770,arXiv 2026,2026,Ontology & Knowledge Representation,"Neuro-symbolic trajectory generation from multi-view images and coordinate-free temporal-logic syntax graphs, binding perception to symbols through bidirectional cross-attention.",
DynamicGSG,2502.15309,arXiv 2025,2025,Ontology & Knowledge Representation,"Dynamic, high-fidelity, open-vocabulary 3D Gaussian scene graph construction that updates environment memory as scenes change, supporting long-term robot tasks.",
EmbodiedVSR,2503.11089,arXiv 2025,2025,Ontology & Knowledge Representation,Dynamic scene graph-guided chain-of-thought reasoning enabling zero-shot spatial understanding for embodied agents on complex long-horizon tasks.,
Artisan,2610.13947,arXiv 2026,2026,Super Agents & Agentic AI,"Multi-agent pipeline with physical simulation feedback for physically grounded articulated 3D assets — collision-free motion, stability under gravity, accurate mass and material properties.",
CADFather,2610.09127,arXiv 2026,2026,Super Agents & Agentic AI,"Autonomous agentic CAD reconstruction that coordinates complementary tools to recover parametric CAD programs from 3D meshes, deciding which tools to invoke and when to finish.",
CityPlanner,2609.09578,arXiv 2026,2026,Super Agents & Agentic AI,"Sandbox agent for executable urban planning: agents inspect task files, generate plans, run evaluators, and revise decisions in a unified file-based environment.",
COLMAR,2607.13524,arXiv 2026,2026,Super Agents & Agentic AI,"Cooperative view policy learning for multi-agent active 3D reconstruction, promoting overlap-aware coverage, team-level discovery, and collision-safe exploration under sensing budgets.",
DesignAgent3D,2608.21438,arXiv 2026,2026,Super Agents & Agentic AI,Interactive 3D scene editing agent with designer-like multimodal reasoning that disambiguates underspecified language requests and grounds precise edits in cluttered scenes.,
E-Ledger,2610.11552,arXiv 2026,2026,Super Agents & Agentic AI,"Multi-agent harness for safe, persistent execution in partially observable worlds: a code approval layer checks every action against policy, plus a world ledger of verified hidden rules.",
E3VS-Bench,2604.17969,arXiv 2026,2026,Super Agents & Agentic AI,"Benchmark for viewpoint-dependent active perception in 3DGS scenes under unrestricted 5-DoF control, beyond static-observation embodied QA.",
GASE,2606.17520,arXiv 2026,2026,Super Agents & Agentic AI,Gaussian-splatting-based automated system for reconstructing embodied-simulation environments with minimal sim-to-real gap for robot learning.,
GATOR,2610.11215,arXiv 2026,2026,Super Agents & Agentic AI,"Generative and agentic framework recovering textured 3D object assets and scene-relative pose from casual images, integrating sparse uncertain observations and inferring occluded surfaces.",
GISAgentBench,2608.01645,arXiv 2026,2026,Super Agents & Agentic AI,"Practitioner-sourced benchmark evaluating LLM agents on realistic multi-step GIS spatial-analysis workflows from urban planning, disaster response, and environmental monitoring.",
GN0,2606.03682,arXiv 2026,2026,Super Agents & Agentic AI,"Unified generation-evaluation-policy paradigm for vision-language navigation on a 3DGS simulation engine, with the GN-Matrix dataset and GN-Bench.",
MeMoREval,2610.12561,arXiv 2026,2026,Super Agents & Agentic AI,"Controlled experience-request benchmark studying which memory modalities (textual, visual, spatial) embodied agents should retain and reuse for subsequent tasks.",
PhotoAgent,2603.22796,arXiv 2026,2026,Super Agents & Agentic AI,"Robotic photographer agent integrating LMM chain-of-thought reasoning with geometric solvers, refined by visual reflection inside a photorealistic internal world model.",
SearchWorld,2610.09335,arXiv 2026,2026,Super Agents & Agentic AI,"Spatial value-grounded imagination via world models for UAV object search, prospectively predicting future states under partial observability.",
SkillScriptBench,2610.04008,arXiv 2026,2026,Super Agents & Agentic AI,"350-task benchmark separating documentation repair, script repair, and behavior preservation when executable agent skill packages self-evolve.",
SPLATIFY,2610.09116,arXiv 2026,2026,Super Agents & Agentic AI,Multi-agent framework converting 3DGS papers into trainable gsplat-based implementations via a context-free grammar over a modular method template with extension points.,
SwarmWorld,2608.26081,arXiv 2026,2026,Super Agents & Agentic AI,"Decentralized LLM-agent societies that self-organize via stigmergy without assigned roles, evolving functional technologies in a spatial environment.",
WorldBench,2610.10622,arXiv 2026,2026,Super Agents & Agentic AI,"Benchmark and judge for open-ended LLM-generated Three.js voxel worlds, resolving vision-vs-code judge disagreement on required scene items.",
GRAND-SLAM,2506.18885,arXiv 2025,2025,Super Agents & Agentic AI,Collaborative large-scale multi-agent Gaussian SLAM with implicit local-optimization tracking for globally consistent outdoor mapping and rapid exploration.,
Octopus v2,2404.01744,arXiv 2024,2024,Super Agents & Agentic AI,"On-device 2B-parameter language model for super-agent function calling that surpasses GPT-4 in accuracy and latency, enabling efficient context-aware execution on edge devices.",
CGGS,2607.03819,arXiv 2026,2026,AIGC Content Creation,Consistency-Augmented Geometric GS for ego-centric 3D scene generation; consistency-augmented loss + mutual information depth loss (MID) for hierarchical geometry-visual optimization,
Flex4DHuman,2606.13655,arXiv 2026,2026,AIGC Content Creation,Multi-view video diffusion transforming monocular or sparse multi-view video into synchronized dense multi-view sequences via relative camera-pose conditioning for 4D human reconstruction.,
LightHarmony3D,2603.29209,arXiv 2026,2026,AIGC Content Creation,Harmonizes illumination and shadows when inserting external mesh objects into reconstructed 3DGS scenes for interactive editing and content augmentation.,
OmniStyle-INR,2607.16362,arXiv 2026,2026,AIGC Content Creation,"Universal multimodal style transfer across implicit neural representations including Gaussian Splatting, conditioned by reference images and textual descriptions.",
PSGS,2602.00463,arXiv 2026,2026,AIGC Content Creation,"Text-driven panorama sliding scene generation via Gaussian Splatting for immersive VR/AR/gaming content, overcoming limited 3D-text data and inconsistent multi-view stitching.",
UniWorld-View,2608.04701,arXiv 2026,2026,AIGC Content Creation,Large-baseline novel view synthesis from casually captured monocular media via video diffusion models for immersive content creation.,
VirSqueezer,2609.01698,arXiv 2026,2026,AIGC Content Creation,Generates realistic localized deformations and squeezing dynamics in VR content creation from fine-grained per-finger squeezing controls.,
CSD,2505.04262,arXiv 2025,2025,AIGC Content Creation,Coupled Score Distillation couples multi-view joint distribution priors from diffusion models to ensure geometry-coherent text-to-3D generation with stable direct 3DGS optimization.,
DD3G,2504.00457,arXiv 2025,2025,AIGC Content Creation,"Distills a multi-view diffusion model into a 3D generator using Gaussian splatting, compressing visual and spatial knowledge via ODE simulation.",
DiffSplat,2501.16764,arXiv 2025,2025,AIGC Content Creation,"Repurposes pretrained image diffusion models for scalable 3D Gaussian splat generation from text or a single image, natively generating 3D Gaussians with 2D diffusion priors.",
DirectGaussian,2510.06967,arXiv 2025,2025,AIGC Content Creation,"Text-to-3D method directly generating object surfaces as 2D Gaussian splatting surfels, combining conditional text generation with surface representation.",
Drag4D,2509.21888,arXiv 2025,2025,AIGC Content Creation,Interactive framework integrating object motion control via user-defined 3D trajectories into text-driven 3D scene generation.,
Dragen3D,2502.16475,arXiv 2025,2025,AIGC Content Creation,Multiview geometry-consistent 3D Gaussian generation with drag-based control for single-image 3D content creation.,
DreamArt,2507.05763,arXiv 2025,2025,AIGC Content Creation,Generates interactable articulated objects with part decomposition and joint articulation from a single image for Embodied AI and AR/VR.,
GaussianMotion,2502.11642,arXiv 2025,2025,AIGC Content Creation,"End-to-end learning of animatable Gaussian avatars with pose guidance from text, generating fully animatable scenes aligned with textual descriptions.",
GaussVideoDreamer,2504.10001,arXiv 2025,2025,AIGC Content Creation,Single-image 3D scene generation combining video diffusion with inconsistency-aware Gaussian Splatting.,
GS-RGBN,2504.01512,arXiv 2025,2025,AIGC Content Creation,RGBN-volume Gaussian reconstruction model fixing geometric ambiguity and structural inconsistency in single-view 3D generation from multi-view diffusion images.,
GSV3D,2503.06136,arXiv 2025,2025,AIGC Content Creation,Gaussian Splatting-based geometric distillation with Stable Video Diffusion for single-image 3D object generation.,
InStyle,2512.03683,arXiv 2025,2025,AIGC Content Creation,"Instant appearance stylization of 3D shapes supporting fast, high-fidelity text-driven manipulation for game, VR and digital-arts asset pipelines.",
Lyra,2509.19296,arXiv 2025,2025,AIGC Content Creation,"Generative 3D scene reconstruction via video diffusion model self-distillation, producing virtual environments without captured references.",
SplatFont3D,2512.00413,arXiv 2025,2025,AIGC Content Creation,"Structure-aware text-to-3D artistic font generation with part-level style control, extending artistic font design from flat 2D to personalized 3D.",
T2Bs,2509.10678,arXiv 2025,2025,AIGC Content Creation,Text-to-character blendshapes framework combining static text-to-3D generation with video diffusion for animatable head morphable models.,