Dual-Stream Diffusion Net for Text-to-Video Generation

TL;DR

This paper introduces a dual-stream diffusion network that enhances the quality and consistency of text-to-video generation by separately modeling content and motion, and aligning them through a cross-transformer, resulting in smoother videos.

Contribution

The paper presents a novel dual-stream diffusion architecture with cross-transformer interaction and motion decomposition for improved text-to-video synthesis.

Findings

Produces videos with fewer flickers and artifacts.

Demonstrates superior qualitative and quantitative results.

Enables personalized and consistent video variations.

Abstract

With the emerging diffusion models, recently, text-to-video generation has aroused increasing attention. But an important bottleneck therein is that generative videos often tend to carry some flickers and artifacts. In this work, we propose a dual-stream diffusion net (DSDN) to improve the consistency of content variations in generating videos. In particular, the designed two diffusion streams, video content and motion branches, could not only run separately in their private spaces for producing personalized video variations as well as content, but also be well-aligned between the content and motion domains through leveraging our designed cross-transformer interaction module, which would benefit the smoothness of generated videos. Besides, we also introduce motion decomposer and combiner to faciliate the operation on video motion. Qualitative and quantitative experiments demonstrate that our method could produce amazing continuous videos with fewer flickers.