Dehaze-then-Splat: Generative Dehazing with Physics-Informed 3D Gaussian Splatting for Smoke-Free Novel View Synthesis

• URL: http://arxiv.org/abs/2604.13589v3
• Date: Thu, 23 Apr 2026 03:15:40 GMT
• Title: Dehaze-then-Splat: Generative Dehazing with Physics-Informed 3D Gaussian Splatting for Smoke-Free Novel View Synthesis
• Abstract summary: We present Dehaze-then-Splat, a two-stage pipeline for multi-view smoke removal and novel view synthesis developed for Track2 of the NTIRE 2026 3D Restoration and Reconstruction Challenge.<n>In the first stage, we produce pseudo-clean training images via per-frame generative dehazing using Nano Banana Pro, followed by brightness normalization.<n>In the second stage, we train 3D Gaussian Splatting with physics-informed auxiliary losses.
• Score: 1.8997372468370612
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We present Dehaze-then-Splat, a two-stage pipeline for multi-view smoke removal and novel view synthesis developed for Track~2 of the NTIRE 2026 3D Restoration and Reconstruction Challenge. In the first stage, we produce pseudo-clean training images via per-frame generative dehazing using Nano Banana Pro, followed by brightness normalization. In the second stage, we train 3D Gaussian Splatting (3DGS) with physics-informed auxiliary losses -- depth supervision via Pearson correlation with pseudo-depth, dark channel prior regularization, and dual-source gradient matching -- that compensate for cross-view inconsistencies inherent in frame-wise generative processing. We identify a fundamental tension in dehaze-then-reconstruct pipelines: per-image restoration quality does not guarantee multi-view consistency, and such inconsistency manifests as blurred renders and structural instability in downstream 3D reconstruction.Our analysis shows that MCMC-based densification with early stopping, combined with depth and haze-suppression priors, effectively mitigates these artifacts. On the Akikaze validation scene, our pipeline achieves 20.98\,dB PSNR and 0.683 SSIM for novel view synthesis, a +1.50\,dB improvement over the unregularized baseline.

Related papers

• DualDiff3D: Dual Structure-Appearance Diffusion Priors for Reliability-Enhanced 3D Gaussian Splatting [71.316050612276]
  3D reconstruction and novel-view synthesis often lead to poor reconstruction quality and artifacts in rendered novel views.<n>Recent efforts attempt to utilize powerful diffusion priors, yet they typically process rendered and reference views along an additional dimension in a single network.<n>We propose DualDiff, a novel pipeline that leverages dual diffusion priors with a Structure-Appearance Attention (SAA) module.<n>Our approach outperforms state-of-the-art methods even in the inference-only setting, with further performance gains achievable through training.
  arXiv  Detail & Related papers  (2026-09-01T16:45:53Z)
• KC-3DGS: Kurtosis-Constrained Gaussian Splatting for High-Fidelity View Synthesis [28.10448224660243]
  3D Gaussian Splatting (3DGS) enables real-time novel view synthesis by representing scenes as collections of anisotropic Gaussians optimized via differentiableization.<n>Standard pixel-space losses (L1, SSIM) constrain only aggregate reconstruction error, permitting the optimization to redistribute error across frequency scales.<n>We propose KC-3DGS, which augments 3DGS training with wavelet-domain supervision based on natural image statistics.
  arXiv  Detail & Related papers  (2026-06-02T04:04:24Z)
• VidSplat: Gaussian Splatting Reconstruction with Geometry-Guided Video Diffusion Priors [79.88300861391114]
  VidSplat is a training-free generative reconstruction framework.<n>We tackle two key challenges that enable the effective integration of generation and reconstruction.<n>Experiments on widely used benchmarks demonstrate our superior performance in sparse-view scene reconstruction.
  arXiv  Detail & Related papers  (2026-05-12T02:20:31Z)
• DualSplat: Robust 3D Gaussian Splatting via Pseudo-Mask Bootstrapping from Reconstruction Failures [11.192603832832212]
  3D Gaussian Splatting (3DGS) achieves real-time rendering, but its performance degrades significantly when training images contain transient objects.<n>Existing methods face a circular dependency: accurate transient detection requires a well-reconstructed static scene, while clean reconstruction itself depends on reliable transient masks.<n>We address this challenge with DualSplat, a Failure-to-Prior framework that converts first-pass reconstruction failures into explicit priors for a second reconstruction stage.
  arXiv  Detail & Related papers  (2026-04-23T12:50:13Z)
• PoInit-of-View: Poisoning Initialization of Views Transfers Across Multiple 3D Reconstruction Systems [55.84012796735184]
  Poisoning input views of 3D reconstruction systems has been recently studied.<n>We propose PoInit-of-View, which optimize adversarial perturbations to intentionally introduce cross-view inconsistencies.<n> Experimental results demonstrate the effectiveness of our PoInit-of-View on diverse 3D reconstruction systems and datasets.
  arXiv  Detail & Related papers  (2026-04-17T00:48:12Z)
• 3D Gaussian Splatting for Annular Dark Field Scanning Transmission Electron Microscopy Tomography Reconstruction [13.051560405222775]
  ADF-STEM tomography reconstructs nanoscale materials in 3D by integrating multi-view tilt-series images.<n>We introduce a coefficient $$ to stabilize scattering across tilt angles, ensuring consistent denza via scattering view normalization.<n>Experiments on 45-view and 15-view tilt series show that DenZa-Gaussian produces high-fidelity reconstructions and 2D projections that align more closely with original tilts.
  arXiv  Detail & Related papers  (2026-04-06T14:01:43Z)
• S2D: Sparse to Dense Lifting for 3D Reconstruction with Minimal Inputs [53.597239706600455]
  We introduce Sparse to Dense lifting (S2D), a novel pipeline that bridges the two representations and achieves high-quality 3DGS reconstruction with minimal inputs.<n>Experiments show that S2D achieves the best consistency in generating novel view guidance and first-tier sparse view reconstruction quality under different input sparsity.
  arXiv  Detail & Related papers  (2026-03-11T15:37:20Z)
• ERGO: Excess-Risk-Guided Optimization for High-Fidelity Monocular 3D Gaussian Splatting [63.138778159026934]
  We propose an adaptive optimization framework guided by excess risk decomposition, termed ERGO.<n> ERGO dynamically estimates the view-specific excess risk and adaptively adjust loss weights during optimization.<n>Experiments on the Google Scanned Objects dataset and the OmniObject3D dataset demonstrate the superiority of ERGO over existing state-of-the-art methods.
  arXiv  Detail & Related papers  (2026-02-10T20:44:43Z)
• One-Shot Refiner: Boosting Feed-forward Novel View Synthesis via One-Step Diffusion [57.824020826432815]
  We present a novel framework for high-fidelity novel view synthesis (NVS) from sparse images.<n>We design a Dual-Domain Detail Perception Module, which enables handling high-resolution images without being limited by the ViT backbone.<n>We develop a feature-guided diffusion network, which can preserve high-frequency details during the restoration process.
  arXiv  Detail & Related papers  (2026-01-20T17:11:55Z)
• HBSplat: Robust Sparse-View Gaussian Reconstruction with Hybrid-Loss Guided Depth and Bidirectional Warping [11.035994094874141]
  HBSplat is a framework that seamlessly integrates robust structural cues, virtual view constraints, and occluded region completion.<n> HBSplat sets a new state-of-the-art, achieving up to 21.13 dB PSNR and 0.189 LPIPS, while maintaining real-time inference.
  arXiv  Detail & Related papers  (2025-09-29T15:03:31Z)
• From Restoration to Reconstruction: Rethinking 3D Gaussian Splatting for Underwater Scenes [13.730810237133822]
  We propose textbfR-Splatting, a unified framework that bridges underwater image restoration (UIR) with 3D Gaussian Splatting (3DGS)<n>Our method integrates multiple enhanced views produced by diverse UIR models into a single reconstruction pipeline.<n>Experiments on Seathru-NeRF and our new BlueCoral3D dataset demonstrate that R-Splatting outperforms strong baselines in both rendering quality and geometric accuracy.
  arXiv  Detail & Related papers  (2025-09-22T13:50:20Z)
• DSplats: 3D Generation by Denoising Splats-Based Multiview Diffusion Models [67.50989119438508]
  We introduce DSplats, a novel method that directly denoises multiview images using Gaussian-based Reconstructors to produce realistic 3D assets.<n>Our experiments demonstrate that DSplats not only produces high-quality, spatially consistent outputs, but also sets a new standard in single-image to 3D reconstruction.
  arXiv  Detail & Related papers  (2024-12-11T07:32:17Z)
• T-3DGS: Removing Transient Objects for 3D Scene Reconstruction [83.05271859398779]
  Transient objects in video sequences can significantly degrade the quality of 3D scene reconstructions.<n>We propose T-3DGS, a novel framework that robustly filters out transient distractors during 3D reconstruction using Gaussian Splatting.
  arXiv  Detail & Related papers  (2024-11-29T07:45:24Z)
• CVT-xRF: Contrastive In-Voxel Transformer for 3D Consistent Radiance Fields from Sparse Inputs [65.80187860906115]
  We propose a novel approach to improve NeRF's performance with sparse inputs. We first adopt a voxel-based ray sampling strategy to ensure that the sampled rays intersect with a certain voxel in 3D space. We then randomly sample additional points within the voxel and apply a Transformer to infer the properties of other points on each ray, which are then incorporated into the volume rendering.
  arXiv  Detail & Related papers  (2024-03-25T15:56:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.