Researchers have introduced OuroWorld, a mask-free framework that transforms any static 3D Gaussian Splatting scene into a 3D cinemagraph — a dynamic scene featuring vivid, diverse motion that loops seamlessly from any viewpoint. The work addresses a key limitation of recent 3D world models, which generate photorealistic, explorable scenes that remain frozen in time.
The pipeline begins with a vision-language model that infers plausible dynamics and guides a video model to synthesize a reference video. This reference is then lifted and completed into multi-view videos. To learn from this imperfect supervision, the authors propose Inconsistency-Robust Periodic 4DGS: a Fourier-series deformation field that guarantees looping by construction, paired with a Grounded Drift Field anchored at the reference view to absorb cross-view inconsistency.
Unlike prior Eulerian methods limited to fluid-like motion, OuroWorld captures general deformation, object motion, and illumination change. The team also introduces a ground-truth-free evaluation covering vividness, naturalness, loop seam coherence, and scene quality.
On 39 reconstructed and generated scenes, OuroWorld outperformed all baselines and won between 70.8% and 99.0% of user-study comparisons. A project page is available at ouroworld.userwei.com.