Awesome

DEVIL Protocol

This respotory is the offical implementation of following paper:

Evaluation of Text-to-Video Generation Models: A Dynamics Perspective
Mingxiang Liao*, Hannan Lu*, Xinyu Zhang*, Fang Wan, Tianyu Wang, Yuzhong Zhao, Wangmeng Zuo, Qixiang Ye, Jingdong Wang
Neural Information Processing Systems, (2024)

Please refer to the Project Page to view illustrations depicting the dynamics of various videos.

Table of Contents

• DEVIL Protocol
• Table of Contents
• Overview
• Installation
• Model Weights
• Gemini API Key
• Usage
• Acknowledgement
• Citation
• TODO

Overview

Comprehensive and constructive evaluation protocols play an important role in the development of sophisticated text-to-video (T2V) generation models. Existing evaluation protocols primarily focus on temporal consistency and content continuity, yet largely ignoring the dynamics of video content. Dynamics are an essential dimension for measuring the visual vividness and the honesty of video content to text prompts. In this study, we propose an effective evaluation protocol, termed DEVIL, which centers on the dynamics dimension to evaluate T2V models. For this purpose, we establish a new benchmark comprising text prompts that fully reflect multiple dynamics grades, and define a set of dynamics scores corresponding to various temporal granularities to comprehensively evaluate the dynamics of each generated video. Based on the new benchmark and the dynamics scores, we assess T2V models with the design of three metrics: dynamics range, dynamics controllability, and dynamics-based quality. Experiments show that DEVIL achieves a Pearson correlation exceeding 90% with human ratings, demonstrating its potential to advance T2V generation models.

Installation

cd geminiplayground 
pip install -e .
cd ..
pip install -r requirements.txt

Model Weights

Download model weights from Google Drive or Baidu Disk(extract code: 2gjp) and put them in the model_weights directory.

Gemini API Key

The naturalness metric relies on the Gemini 1.5 Pro model. So please turn to Gemini to obtain your gemini_api_key before evaluating the metric.

Usage

• Generate videos with prompts provided in prompts/, all videos should be named with dynamics prefix, e.g. 'high_1.mp4', 'very_high_123.mp4'...
• Evalute videos dynamics and model metrics:

  bash eval_dynamics.dist.sh \
    --video_dir dir_to_your_videos \
    --gemini_api_key your_gemini_api_key \
    --num_gpus 8
  

Acknowledgement

We gratefully acknowledge the following repositories, whose resources were instrumental in our evaluation: Vbench, EvalCrafter, geminiplayground, and ViClip. Our sincere thanks to the contributors of these projects.

Citation

Please consider citing our paper in your publications if the project helps your research.

@article{liao2024evaluation,
  title={Evaluation of text-to-video generation models: A dynamics perspective},
  author={Liao, Mingxiang and Lu, Hannan and Zhang, Xinyu and Wan, Fang and Wang, Tianyu and Zhao, Yuzhong and Zuo, Wangmeng and Ye, Qixiang and Wang, Jingdong},
  journal={arXiv preprint arXiv:2407.01094},
  year={2024}
}

TODO

• New version of the prompt
• Demo website
• Pypi pakege