VideoCrafter2: Overcoming Data Limitations for High-Quality Video Diffusion Models

TL;DR

VideoCrafter2 introduces a novel training scheme that leverages low-quality videos and synthesized high-quality images to develop high-quality video diffusion models without requiring large-scale high-quality datasets.

Contribution

The paper proposes a new training approach extending Stable Diffusion, enabling high-quality video generation from low-quality data by finetuning spatial modules with high-quality images.

Findings

High-quality video models can be trained using low-quality videos and synthesized images.

Finetuning spatial modules improves video quality without motion degradation.

The method outperforms existing approaches in picture quality, motion, and concept composition.

Abstract

Text-to-video generation aims to produce a video based on a given prompt. Recently, several commercial video models have been able to generate plausible videos with minimal noise, excellent details, and high aesthetic scores. However, these models rely on large-scale, well-filtered, high-quality videos that are not accessible to the community. Many existing research works, which train models using the low-quality WebVid-10M dataset, struggle to generate high-quality videos because the models are optimized to fit WebVid-10M. In this work, we explore the training scheme of video models extended from Stable Diffusion and investigate the feasibility of leveraging low-quality videos and synthesized high-quality images to obtain a high-quality video model. We first analyze the connection between the spatial and temporal modules of video models and the distribution shift to low-quality videos. We observe that full training of all modules results in a stronger coupling between spatial and temporal modules than only training temporal modules. Based on this stronger coupling, we shift the distribution to higher quality without motion degradation by finetuning spatial modules with high-quality images, resulting in a generic high-quality video model. Evaluations are conducted to demonstrate the superiority of the proposed method, particularly in picture quality, motion, and concept composition.