[Project Page] [Paper]
Refer to our latest work for more discussion about initial noise in diffusion!
Our codebase is built on CompVis/stable-diffusion and has shared dependencies and model architecture.
conda env create -f environment.yaml
conda activate ldm
Download the StableDiffusion weights from the CompVis organization at Hugging Face
(download the sd-v1-4.ckpt file), and link them:
mkdir -p models/ldm/stable-diffusion-v1/
ln -s <path/to/model.ckpt> models/ldm/stable-diffusion-v1/model.ckpt
Play with hands-on to try our approach right away, refer to the Initializer in the utils.py for the implementation.
@article{mao2023guided,
title={Guided Image Synthesis via Initial Image Editing in Diffusion Model},
author={Mao, Jiafeng and Wang, Xueting and Aizawa, Kiyoharu},
journal={ACM MM},
year={2023}
}