JosephTiTan / FreePCA Public

Notifications You must be signed in to change notification settings
Fork 2
Star 20

Code of the paper "FreePCA：Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis", accepted by CVPR 2025 (highlight).

20 stars 2 forks Branches Tags Activity

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 18 Commits
config		config
lvdm		lvdm
prompts		prompts
scripts		scripts
utils		utils
README.md		README.md
gradio_app.py		gradio_app.py
overview.jpg		overview.jpg
predict.py		predict.py

Repository files navigation

FreePCA：Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis(arxiv link)

Overview

Setup (based on Videocrafter2)

1. Install Environment via Anaconda (Recommended)

conda create -n freepca python=3.8.5
conda activate freepca
pip install -r requirements.txt

2. Download pretrained T2V models via Hugging Face, and put the `model.ckpt` in `checkpoints/base_512_v2/model.ckpt`.

3. Input the following commands in terminal.

sh scripts/run_text2video.sh

About

Code of the paper "FreePCA：Integrating Consistency Information across Long-short Frames in Training-free Long Video Generation via Principal Component Analysis", accepted by CVPR 2025 (highlight).

Report repository

Releases

No releases published

Packages

No packages published

Languages