SVG: 3D Stereoscopic Video Generation via Denoising Frame Matrix

TL;DR

This paper introduces a novel pose-free, training-free method for generating 3D stereoscopic videos from monocular videos using a frame matrix inpainting framework and depth estimation, without scene optimization or model fine-tuning.

Contribution

It presents a new approach that warps monocular videos into stereoscopic views and employs a frame matrix inpainting framework with disocclusion boundary re-injection, improving 3D stereoscopic video quality.

Findings

Significant improvement over previous methods in generating stereoscopic videos.

Effective in maintaining semantic coherence and consistency in generated videos.

Validated on multiple generative models including Sora, Lumiere, WALT, and Zeroscope.

Abstract

Video generation models have demonstrated great capabilities of producing impressive monocular videos, however, the generation of 3D stereoscopic video remains under-explored. We propose a pose-free and training-free approach for generating 3D stereoscopic videos using an off-the-shelf monocular video generation model. Our method warps a generated monocular video into camera views on stereoscopic baseline using estimated video depth, and employs a novel frame matrix video inpainting framework. The framework leverages the video generation model to inpaint frames observed from different timestamps and views. This effective approach generates consistent and semantically coherent stereoscopic videos without scene optimization or model fine-tuning. Moreover, we develop a disocclusion boundary re-injection scheme that further improves the quality of video inpainting by alleviating the negative effects propagated from disoccluded areas in the latent space. We validate the efficacy of our proposed method by conducting experiments on videos from various generative models, including Sora [4 ], Lumiere [2], WALT [8 ], and Zeroscope [ 42]. The experiments demonstrate that our method has a significant improvement over previous methods. The code will be released at \url{https://daipengwa.github.io/SVG_ProjectPage}.