π’ This work has been accepted to ICCV 2025.
This repository provides a differentiable wave optics simulator for point spread function (PSF) generation, image formation, and end-to-end optimization of computational imaging systems.
It is designed for research in computational imaging, holography, and optics-aware machine learning.
-
Differentiable Wave Optics Simulation
Forward model with gradient support for optimization. -
PSF Generation
Simulate diffraction-limited or aberrated point spread functions. -
Image Formation
Render images under custom optical systems. -
End-to-End Optimization
Jointly optimize optical parameters and downstream tasks. -
Config-Driven Pipeline
All experiments are configured via JSON files for reproducibility.
Clone the repository:
git clone https://github.com/JerryHoTaiwan/DeepWaveOptics.git
cd DeepWaveOpticsCreate a virtual environment and install dependencies:
python -m venv .venv source .venv/bin/activate # Linux/Mac .venv\Scripts\activate # Windows pip install -r requirements.txt
DeepWaveOptics/
βββ configs/ # JSON configs (default/experiments/env)
βββ src/ # Core implementation
β βββ functions.py # Callable blocks (PSF, render, optimize, etc.)
β βββ utils.py # Config + folder utilities
β βββ main.py # Entry point
β βββ initialization.py
βββ results/ # Outputs (PSFs, images, logs, checkpoints)
βββ README.md
All experiments are controlled via JSON configs:
python main.py --config configs/experiments/show_psfs.json
- Generate PSFs
python main.py --config configs/show_psfs.json
- Render Images
python main.py --config configs/display_triplet.json
- End-to-End Optimization
python main.py --config configs/train_cellphonelens.json
If you use this code in your work, please cite:
@article{ho2024differentiable,
title={A Differentiable Wave Optics Model for End-to-End Computational Imaging System Optimization},
author={Ho, Chi-Jui and Belhe, Yash and Rotenberg, Steve and Ramamoorthi, Ravi and Li, Tzu-Mao and Antipa, Nicholas},
journal={arXiv preprint arXiv:2412.09774},
year={2024}
}
This project builds on the excellent work of the DeepLens and DiffOptics frameworks developed by Xinge Yang et al.. Their differentiable rayβwave model provided the foundation for extending this project toward end-to-end optimization of computational imaging systems. We gratefully acknowledge their contributions and encourage readers to also cite their work:
@article{yang2024end,
title={End-to-End Hybrid Refractive-Diffractive Lens Design with Differentiable Ray-Wave Model},
author={Yang, Xinge and Souza, Matheus and Wang, Kunyi and Chakravarthula, Praneeth and Fu, Qiang and Heidrich, Wolfgang},
journal={arXiv preprint arXiv:2406.00834},
year={2024}
}