The repository for Can Diffusion Model Conditionally Generate Astrophysical Images? (Zhao et al. 2023). This script is mainly built upon:Conditional Diffusion MNIST, guided-diffusion and improved-diffusion.
The requirements to run the codes are rather simple, please check the requirements.txt, where I list the versions of the packages I used (with py3.8), while probably your existed PyTorch environment can work. In this repository (mainly in a single file main.py), you can know all about the basics of the paper DDPM.
You can directly run the main.py which will give some outputs to outputs.
DDPM for image generation conditional on the astrophysical parameter
The wavelet scattering transform used for quantification is calculated with this repository by Sihao Cheng.
The StyleGAN2 implementation is based on the related branch in this repository.