About The Project

This is an unofficial implementation of Training-free Diffusion Model Adaptation for Variable-Sized Text-to-Image Synthesis. Further details can be viewed in the papers

Samples

A black cat sitting on top of a pot for a plant. [768x768]
BMW X5 3.0d Swap or Sell. [768x768]
Colorful house against nature background. [1024x1024]
Kids doing homeworks. [224x224]
A person on Moon. [224x224]
A red stop sign that says PARE near a brick intersection. [224x224]

Getting Started

For the ease of installation, you should use anaconda to setup this repo.

Installation

conda create --name <env> --file req.txt
conda activate <env>

Inference

# Usage:
python trainingfreediffuser.py --help

# Example:
## With reweighting scheme
python trainingfreediffuser.py "A black cat sitting on top of a pot for a plant." --wh 768 -n 8 -s 0 -e
## Without reweighting scheme (Normal sampling)
python trainingfreediffuser.py "A black cat sitting on top of a pot for a plant." --wh 768 -n 8 -s 0

Acknowledgements

This project would not be possible without the generous work from Stability AI and HuggingFace