Video World Models · 2026-10-09

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

You-Zhe Xie, Ting-Wei Chou, Yu-Hsuan Li, Kaipeng Zhang, Zhixiang Wang, Yu-Lun Liu

AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.

TL;DR

OuroWorld converts any static 3D Gaussian Splatting (3DGS) scene into a 3D cinemagraph with vivid, diverse motion that loops seamlessly from any viewpoint. A vision-language model prompts a video generator to make an approximately looping reference video, which is lifted to multi-view videos. The method then fits an Inconsistency-Robust Periodic 4DGS using a Fourier-series deformation field (looping by construction) and a Grounded Drift Field (absorbs cross-view inconsistency).

Why it matters

The authors address 3D world models that are photorealistic but remain frozen in time by producing an endlessly looping, view-consistent dynamic 3D representation. They emphasize mask-free conversion from static 3DGS, support for general deformation/object motion/illumination change beyond Eulerian fluid-like motion, and propose a ground-truth-free evaluation for vividness, naturalness, loop seam coherence, and scene quality.

Method

Mask-free 3-stage pipeline: use a VLM to prompt a video model for a reference loop from a rendered 3DGS view; lift to multi-view videos; complete inpainted multi-view videos.
Inconsistency-Robust Periodic 4DGS: represent the dynamic 3D cinemagraph with a canonical 3DGS plus a Fourier-series Periodic Deformation Field that guarantees looping by construction.
Learn from inconsistent generated multi-view supervision via a Grounded Drift Field anchored at the reference view (drift discarded at inference) plus scene-view consistent optimization (L1 on reference frames and LPIPS on refined frames).

Abstract (from arXiv)

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

Publication record

Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.

First public release: 2026-10-08 · Indexed by Pub-AI: 2026-10-09 UTC

preprint