Search papers, labs, and topics across Lattice.
This paper introduces Ocean4D, a novel generative framework for underwater 4D reconstruction that addresses the challenges posed by light transport degradation and dynamic water variations. By integrating a 4D geometrically consistent conditioning component with a Medium-Aware Block for implicit denoising, the method stabilizes underwater appearances and enhances cross-view consistency. Experimental results on dynamic and static benchmarks show that Ocean4D achieves state-of-the-art performance, significantly improving the reliability of underwater video reconstruction.
Ocean4D achieves unprecedented stability and consistency in underwater 4D reconstruction, overcoming the limitations of traditional methods that fail to account for medium effects.
Underwater 4D reconstruction remains challenging due to the coupling between degraded light transport in participating media and dynamic water variations. Most existing Methods are developed under in-air assumptions and do not explicitly account for underwater absorption and backscatter. Additionally, near-static assumptions make these approaches sensitive to drifting particles and dynamic distractors , leading to unstable geometry and inconsistent cross-view results. To address these issues, we propose a generative framework for underwater 4D reconstruction, named Ocean4D, which is built on two complementary components. Specifically, 4D-GCC constructs 4D geometrically consistent conditioning with improved cross-frame coverage, while the Medium-Aware Block performs implicit medium-aware denoising in the latent diffusion process to stabilize underwater appearance under absorption and scattering. Given a monocular video and target cameras, our method generates videos along the target trajectories while preserving global structure and cross-view consistency. Extensive experiments on both dynamic and static underwater benchmarks demonstrate state-of-the-art performance on underwater reconstruction.
