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

TL;DR

This paper introduces a two-stage pipeline combining generative dehazing and physics-informed 3D Gaussian splatting to improve smoke removal and novel view synthesis in multi-view 3D reconstruction.

Contribution

It proposes a novel dehazing-then-splat pipeline with auxiliary losses for better multi-view consistency in smoke removal and view synthesis.

Findings

Achieved 20.98 dB PSNR and 0.683 SSIM on validation scene

Demonstrated that priors and early stopping mitigate artifacts in 3D reconstruction

Improved over baseline by +1.50 dB PSNR in novel view synthesis

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.