📽️ See more video samples in our project page!
![](https://private-user-images.githubusercontent.com/63445348/333781031-aafafa52-5ddf-4093-9d29-681fe469e447.gif?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.HSMqsgQ9QEGECoJk8pxcLa5uHcKs_DnnlEaObX6_nbw)
"An astronaut floating in space, high quality, 4K resolution.",
VideoCrafter2, 100 frames, 320 X 512 resolution
![](/jjihwan/FIFO-Diffusion_public/raw/main/assets/opensora_fifo.gif)
"A corgi vlogging itself in tropical Maui."
Open-Sora Plan, 512 X 512 resolution
[2024.06.06] 🔥🔥🔥 We are excited to release the code for Open-Sora Plan v1.1.0. Thanks to the authors for open-sourcing the awesome baseline!
[2024.05.25] 🥳🥳🥳 We are thrilled to present our official PyTorch implementation for FIFO-Diffusion. We are releasing the code based on VideoCrafter2.
[2024.05.19] 🚀🚀🚀 Our paper, FIFO-Diffusion: Generating Infinite Videos from Text without Training, has been archived.
git clone git@github.com:jjihwan/FIFO-Diffusion_public.git
cd FIFO-Diffusion_public
☀️ Start with VideoCrafter
python3 -m venv .fifo
source .fifo/bin/activate
pip install -r requirements.txt
Model | Resolution | Checkpoint |
---|---|---|
VideoCrafter2 (Text2Video) | 320x512 | Hugging Face |
Store them as following structure:
cd FIFO-Diffusion_public
.
└── videocrafter_models
└── base_512_v2
└── model.ckpt # VideoCrafter2 checkpoint
Requires less than 9GB VRAM with Titan XP.
python3 videocrafter_main.py --save_frames
May consume slightly more memory than the single GPU inference (11GB with Titan XP). Please note that our implementation for parallel inference might not be optimal. Pull requests are welcome! 🤓
python3 videocrafter_main_mp.py --num_gpus 8 --save_frames
Comming soon.
☀️ Start with Open-Sora Plan v1.1.0
For simple implementation, we use the DDPM scheduler for Open-Sora Plan v1.1.0. Since Open-Sora Plan recommends using the PNDM scheduler, the results might not show the optimal performance. Multi-processing (parallelizable inference) and adapting PNDM scheduler are our next plan.
cd FIFO-Diffusion_public
git clone git@github.com:PKU-YuanGroup/Open-Sora-Plan.git
python -m venv .sora
source .sora/bin/activate
cd Open-Sora-Plan
pip install -e .
pip install deepspeed
Requires about 40GB VRAM with A6000. It uses n=8 by default.
sh scripts/opensora_fifo_65.sh
Requires about 40GB VRAM with A6000. It uses n=4 by default.
sh scripts/opensora_fifo_221.sh
Comming soon.
@article{kim2024fifo,
title = {FIFO-Diffusion: Generating Infinite Videos from Text without Training},
author = {Jihwan Kim and Junoh Kang and Jinyoung Choi and Bohyung Han},
journal = {arXiv preprint arXiv:2405.11473},
year = {2024},
}
Our codebase builds on VideoCrafter, Open-Sora Plan, zeroscope. Thanks to the authors for sharing their awesome codebases!