NTIRE 2026 3D Restoration and Reconstruction in Real-world Adverse Conditions: RealX3D Challenge Results
Not on the map. Why?
Abstract (arXiv)
This paper presents a comprehensive review of the NTIRE 2026 3D Restoration and Reconstruction (3DRR) Challenge, detailing the proposed methods and results. The challenge seeks to identify robust reconstruction pipelines that are robust under real-world adverse conditions, specifically extreme low-light and smoke-degraded environments, as captured by our RealX3D benchmark. A total of 279 participants registered for the competition, of whom 33 teams submitted valid results. We thoroughly evaluate the submitted approaches against state-of-the-art baselines, revealing significant progress in 3D reconstruction under adverse conditions. Our analysis highlights shared design principles among top-performing methods and provides insights into effective strategies for handling 3D scene degradation.
Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Builds on 43 of its 79 references that are in the map. First posted on arXiv 5 Apr 2026.
Builds on
43 of 79 references.
References · 43
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,226
- 2025VGGT: Visual Geometry Grounded Transformercited by 1,539
- 2024Depth Anything V2cited by 1,422
- 20242D Gaussian Splatting for Geometrically Accurate Radiance Fieldscited by 1,076
- 2023Scaffold-GS: Structured 3D Gaussians for View-Adaptive Renderingcited by 695
- 2025Depth Anything 3: Recovering the Visual Space from Any Viewscited by 692
- 2020Zero-Reference Deep Curve Estimation for Low-Light Image Enhancementcited by 542
- 2023FSGS: Real-Time Few-shot View Synthesis using Gaussian Splattingcited by 279
- 2025MoGe-2: Accurate Monocular Geometry with Metric Scale and Sharp Detailscited by 261
- 2022Vision Transformers for Single Image Dehazingcited by 260
- 2023Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancementcited by 231
31 more
- 2024A Survey on 3D Gaussian Splattingcited by 224
- 2024Street Gaussians: Modeling Dynamic Urban Scenes with Gaussian Splattingcited by 221
- 2021Learning to Enhance Low-Light Image via Zero-Reference Deep Curve Estimationcited by 192
- 2025Difix3D+: Improving 3D Reconstructions with Single-Step Diffusion Modelscited by 161
- 20243D Gaussian as a New Era: A Surveycited by 104
- 2024SGS-SLAM: Semantic Gaussian Splatting For Neural Dense SLAMcited by 87
- 2022Practical Blind Image Denoising via Swin-Conv-UNet and Data Synthesiscited by 86
- 2023MB-TaylorFormer: Multi-branch Efficient Transformer Expanded by Taylor Formula for Image Dehazingcited by 81
- 2023SeaThru-NeRF: Neural Radiance Fields in Scattering Mediacited by 75
- 2025HVI: A New Color Space for Low-light Image Enhancementcited by 63
- 2024DarkIR: Robust Low-Light Image Restorationcited by 48
- 2024MaIR: A Locality- and Continuity-Preserving Mamba for Image Restorationcited by 45
- 2024WaterSplatting: Fast Underwater 3D Scene Reconstruction Using Gaussian Splattingcited by 41
- 2023Aleth-NeRF: Illumination Adaptive NeRF with Concealing Field Assumptioncited by 35
- 2025Luminance-GS: Adapting 3D Gaussian Splatting to Challenging Lighting Conditions with View-Adaptive Curve Adjustmentcited by 27
- 2026The Eleventh NTIRE 2026 Efficient Super-Resolution Challenge Reportcited by 25
- 2025DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Imagescited by 24
- 2024MG-SLAM: Structure Gaussian Splatting SLAM with Manhattan World Hypothesiscited by 22
- 2025LITA-GS: Illumination-Agnostic Novel View Synthesis via Reference-Free 3D Gaussian Splatting and Physical Priorscited by 21
- 2024DeRainGS: Gaussian Splatting for Enhanced Scene Reconstruction in Rainy Environmentscited by 19
- 2025RealX3D: A Physically-Degraded 3D Benchmark for Multi-view Visual Restoration and Reconstructioncited by 19
- 2025I2-NeRF: Learning Neural Radiance Fields Under Physically-Grounded Media Interactionscited by 17
- 2026Faster-GS: Analyzing and Improving Gaussian Splatting Optimizationcited by 17
- 2025DenseSplat: Densifying Gaussian Splatting SLAM with Neural Radiance Priorcited by 14
- 2024Efficient Diffusion as Low Light Enhancercited by 13
- 2026GenSmoke-GS: A Multi-Stage Method for Novel View Synthesis from Smoke-Degraded Images Using a Generative Modelcited by 11
- 2026Denoising the Deep Sky: Physics-Based CCD Noise Formation for Astronomical Imagingcited by 8
- 2026Unifying Color and Lightness Correction with View-Adaptive Curve Adjustment for Robust 3D Novel View Synthesiscited by 8
- 20253D-UIR: 3D Gaussian for Underwater 3D Scene Reconstruction via Physics Based Appearance-Medium Decouplingcited by 5
- 2026UDPNet: Unleashing Depth-based Priors for Robust Image Dehazingcited by 3
- 2025Efficient Real-World Deblurring using Single Images: AIM 2025 Challenge Reportcited by 2
Built on by
Citing papers per year: 2026: 11.
11 papers.
- 2026CLIP-Guided Data Augmentation for Night-Time Image Dehazingcited by 8
- 20263D Smoke Scene Reconstruction Guided by Vision Priors from Multimodal Large Language Modelscited by 7
- 2026Dual-Branch Remote Sensing Infrared Image Super-Resolutioncited by 7
- 2026SmokeGS-R: Physics-Guided Pseudo-Clean 3DGS for Real-World Multi-View Smoke Restorationcited by 6
- 2026RAWild: Sensor-Agnostic RAW Object Detection via Physics-Guided Curve and Grid Modelingcited by 3
- 2026FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolutioncited by 2
- 2026ELoG-GS: Dual-Branch Gaussian Splatting with Luminance-Guided Enhancement for Extreme Low-light 3D Reconstructioncited by 1
- 2026Dehaze-then-Splat: Generative Dehazing with Physics-Informed 3D Gaussian Splatting for Smoke-Free Novel View Synthesiscited by 1
- 2026DelowlightSplat: Feed-Forward Gaussian Splatting for Lowlight 3D Scene Reconstructioncited by 0
- 2026DeblurNVS: Geometric Latent Diffusion for Novel View Synthesis from Sparse Motion-Blurred Imagescited by 0
- 2026FujinSplat: Seeing Through Smoke with RAW-Domain Gaussian Splattingcited by 0
- 2026FujinSplat: Seeing Through Smoke with RAW-Domain Gaussian Splatting5 Sep 2026
- 2026DeblurNVS: Geometric Latent Diffusion for Novel View Synthesis from Sparse Motion-Blurred Images31 May 2026
- 2026DelowlightSplat: Feed-Forward Gaussian Splatting for Lowlight 3D Scene Reconstruction26 May 2026
- 2026RAWild: Sensor-Agnostic RAW Object Detection via Physics-Guided Curve and Grid Modeling7 May 2026
- 2026FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolution5 May 2026
- 2026Dehaze-then-Splat: Generative Dehazing with Physics-Informed 3D Gaussian Splatting for Smoke-Free Novel View Synthesis15 Apr 2026
- 2026ELoG-GS: Dual-Branch Gaussian Splatting with Luminance-Guided Enhancement for Extreme Low-light 3D Reconstruction14 Apr 2026
- 2026Dual-Branch Remote Sensing Infrared Image Super-Resolution11 Apr 2026
- 20263D Smoke Scene Reconstruction Guided by Vision Priors from Multimodal Large Language Models7 Apr 2026
- 2026CLIP-Guided Data Augmentation for Night-Time Image Dehazing7 Apr 2026
- 2026SmokeGS-R: Physics-Guided Pseudo-Clean 3DGS for Real-World Multi-View Smoke Restoration7 Apr 2026