OuroWorld: Bringing Any 3D World Alive as Diverse, Endlessly Looping 3D Cinemagraphs

2026-10-08

AI Summary

This study addresses the lack of dynamic expressiveness in static 3D scenes by proposing a method that transforms static 3D Gaussian Splatting into seamlessly looping 3D cinemagraphs viewable from arbitrary perspectives. Methodologically, it leverages vision-language models to guide video generation and introduces a mask-free framework alongside an inconsistency-resistant periodic 4D Gaussian Splatting technique. By employing Fourier series-based deformation fields, the approach transcends conventional fluid motion constraints, supporting general deformations, object movements, and illumination variations. Experimental results demonstrate that the proposed method comprehensively outperforms existing baselines across 39 scenes, achieving user preference rates ranging from 70.8% to 99.0%. These findings indicate significant improvements in both dynamic vividness and perceptual naturalness.

Abstract

Recent 3D world models generate photorealistic, explorable scenes that remain frozen in time. OuroWorld is a mask-free framework that turns any static 3D Gaussian Splatting scene into a 3D cinemagraph: a dynamic scene with vivid, diverse motion looping seamlessly from any viewpoint. A vision-language model infers plausible dynamics and guides a video model to synthesize a reference video, which we lift and complete into multi-view videos. To learn from this imperfect supervision, we propose Inconsistency-Robust Periodic 4DGS: a Fourier-series deformation field guarantees looping by construction, while a Grounded Drift Field anchored at the reference view absorbs cross-view inconsistency. Unlike prior Eulerian methods limited to fluid-like motion, we capture general deformation, object motion, and illumination change. We introduce a ground-truth-free evaluation covering vividness, naturalness, loop seam coherence, and scene quality. On 39 reconstructed and generated scenes, OuroWorld outperforms all baselines and wins 70.8%-99.0% of user-study comparisons. Project page: https://ouroworld.userwei.com

Problem

• 3D cinemagraphs
• static 3D scenes
• dynamic scene generation
• seamless looping
• 4D Gaussian Splatting

Innovation

• 3D Cinemagraphs
• 3D Gaussian Splatting
• Periodic 4DGS
• Vision-Language Model
• Mask-free Framework