The Year in AI Papers
The year/Independent research

Paper 2512.09363

StereoWorld: Geometry-Aware Monocular-to-Stereo Video Generation

Ke Xing, Xiaojie Jin, Longfei Li, Yuyang Yin, Hanwen Liang, Guixun Luo, Chen Fang, Jue Wang, Konstantinos N. Plataniotis, Yao Zhao, Yunchao Wei

Published
Dec 2025
Research lab
Independent
Citations
2
GitHub
Not linked

01 In brief

Summary

StereoWorld is an end-to-end diffusion-based framework that converts monocular videos into high-fidelity stereo videos by adapting a pretrained video generator.

It conditions the model on the left-view video and uses a geometry-aware regularization combining disparity and depth supervision to ensure 3D structural fidelity.

A spatio-temporal tiling scheme enables efficient high-resolution, long-duration synthesis.

The authors curated StereoWorld-11M, a dataset of over 11 million frames from Blu-ray movies aligned to human interpupillary distance (IPD), addressing the lack of suitable training data.

Experiments show StereoWorld outperforms prior methods (GenStereo, SVG, StereoCrafter) in visual quality, geometric consistency, and temporal stability, with higher PSNR, SSIM, and lower LPIPS, EPE, and D1-all.

Human evaluations also rate it highest on stereo effect, visual quality, binocular consistency, and temporal consistency.

Limitations include limited control over stereo baseline and slow generation (~6 minutes per clip).

02 From the paper

Abstract

The growing adoption of XR devices has fueled strong demand for high-quality stereo video, yet its production remains costly and artifact-prone. To address this challenge, we present StereoWorld, an end-to-end framework that repurposes a pretrained video generator for high-fidelity monocular-to-stereo video generation. Our framework jointly conditions the model on the monocular video input while explicitly supervising the generation with a geometry-aware regularization to ensure 3D structural fidelity. A spatio-temporal tiling scheme is further integrated to enable efficient, high-resolution synthesis. To enable large-scale training and evaluation, we curate a high-definition stereo video dataset containing over 11M frames aligned to natural human interpupillary distance (IPD). Extensive experiments demonstrate that StereoWorld substantially outperforms prior methods, generating stereo videos with superior visual fidelity and geometric consistency. The project webpage is available at https://ke-xing.github.io/StereoWorld/.