🤖 AI Summary
To address distance and color distortion in real-time underwater rendering caused by medium absorption and scattering, this work pioneers the integration of a physics-driven underwater imaging model into the 3D Gaussian Splatting (3DGS) framework. We introduce a differentiable medium attenuation module that jointly enables structure-aware reconstruction, photorealistic color recovery, and medium-free rendering. Our approach synergistically combines physically grounded underwater light transport modeling with end-to-end neural rendering optimization, preserving 3DGS’s real-time performance while substantially improving novel-view synthesis quality: PSNR increases by 2.1 dB on real-world underwater datasets (e.g., SeaThru-NeRF), depth map accuracy improves, and true-color fidelity is enhanced. The core contribution is the first successful coupling of explicit physical models with 3DGS—establishing a new paradigm for underwater visual reconstruction that unifies realism, physical interpretability, and computational efficiency.
📝 Abstract
We introduce SeaSplat, a method to enable real-time rendering of underwater scenes leveraging recent advances in 3D radiance fields. Underwater scenes are challenging visual environments, as rendering through a medium such as water introduces both range and color dependent effects on image capture. We constrain 3D Gaussian Splatting (3DGS), a recent advance in radiance fields enabling rapid training and real-time rendering of full 3D scenes, with a physically grounded underwater image formation model. Applying SeaSplat to the real-world scenes from SeaThru-NeRF dataset, a scene collected by an underwater vehicle in the US Virgin Islands, and simulation-degraded real-world scenes, not only do we see increased quantitative performance on rendering novel viewpoints from the scene with the medium present, but are also able to recover the underlying true color of the scene and restore renders to be without the presence of the intervening medium. We show that the underwater image formation helps learn scene structure, with better depth maps, as well as show that our improvements maintain the significant computational improvements afforded by leveraging a 3D Gaussian representation.