Disentangling Content and Motion for Text-Based Neural Video Manipulation

TL;DR

This paper presents DiCoMoGAN, a novel GAN-based method that disentangles content and motion to enable controllable, semantically meaningful video editing from natural language descriptions, outperforming existing methods in coherence and quality.

Contribution

The paper introduces a new GAN architecture with coupled networks for disentangling content and motion, facilitating semantic video manipulation from text descriptions.

Findings

DiCoMoGAN achieves superior temporal coherence compared to frame-based methods.

The method enables precise semantic edits aligned with natural language input.

Quantitative results show significant performance improvements over existing approaches.

Abstract

Giving machines the ability to imagine possible new objects or scenes from linguistic descriptions and produce their realistic renderings is arguably one of the most challenging problems in computer vision. Recent advances in deep generative models have led to new approaches that give promising results towards this goal. In this paper, we introduce a new method called DiCoMoGAN for manipulating videos with natural language, aiming to perform local and semantic edits on a video clip to alter the appearances of an object of interest. Our GAN architecture allows for better utilization of multiple observations by disentangling content and motion to enable controllable semantic edits. To this end, we introduce two tightly coupled networks: (i) a representation network for constructing a concise understanding of motion dynamics and temporally invariant content, and (ii) a translation network that exploits the extracted latent content representation to actuate the manipulation according to the target description. Our qualitative and quantitative evaluations demonstrate that DiCoMoGAN significantly outperforms existing frame-based methods, producing temporally coherent and semantically more meaningful results.