Seeing World Dynamics in a Nutshell

TL;DR

NutWorld is a novel framework that converts monocular videos into continuous 3D Gaussian representations, enabling efficient, high-fidelity, and real-time scene modeling without optimization.

Contribution

It introduces the STAG representation and a single-pass transformation method for monocular videos into dynamic 3D Gaussian models, improving coherence and efficiency.

Findings

High-fidelity video reconstruction achieved

Real-time scene modeling demonstrated

Effective depth and flow regularization implemented

Abstract

We consider the problem of efficiently representing casually captured monocular videos in a spatially- and temporally-coherent manner. While existing approaches predominantly rely on 2D/2.5D techniques treating videos as collections of spatiotemporal pixels, they struggle with complex motions, occlusions, and geometric consistency due to absence of temporal coherence and explicit 3D structure. Drawing inspiration from monocular video as a projection of the dynamic 3D world, we explore representing videos in their intrinsic 3D form through continuous flows of Gaussian primitives in space-time. In this paper, we propose NutWorld, a novel framework that efficiently transforms monocular videos into dynamic 3D Gaussian representations in a single forward pass. At its core, NutWorld introduces a structured spatial-temporal aligned Gaussian (STAG) representation, enabling optimization-free scene modeling with effective depth and flow regularization. Through comprehensive experiments, we demonstrate that NutWorld achieves high-fidelity video reconstruction quality while enabling various downstream applications in real-time. Demos and code will be available at https://github.com/Nut-World/NutWorld.