Repository: https://github.com/electricalgorithm/hologen

See the first dataset (100k, 157.4GB!) generated with the HoloGen in HuggingFace 🤗

Use Cases

- Training physics-aware neural networks for holographic reconstruction
- Testing and benchmarking reconstruction algorithms
- Generating labeled datasets for twin image suppression models
- Educational demonstrations of computational holography
- Rapid prototyping of holography-based computer vision systems

Quick Start

from hologen import (
    HologramConfig,
    InlineHolography,
    CircleGenerator,
    generate_hologram
)

config = HologramConfig(
    image_size=512,
    pixel_size=3.45e-6, # 3.45 μm pixels
    wavelength=532e-9, # 532 nm green laser
    propagation_distance=0.01 # 1 cm
)

obj_field = CircleGenerator(radius=50).generate(config)
hologram = generate_hologram(obj_field, config, InlineHolography()) 

Installation (from source)

git clone https://github.com/electricalgorithm/hologen
cd hologen
pip install -e . 

 License

MIT License - See LICENSE file for details.

Citation

If you use HoloGen in your research, please cite:

@software{hologen2025,
title={HoloGen: Synthetic Hologram Dataset Generator},
author={Gökhan Koçmarlı},
year={2025},
url={https://github.com/electricalgorithm/hologen}
}

Contributing

Contributions welcome! Please open an issue before submitting major changes.