Skip to content

singer-yang/AutoLens

Repository files navigation

AutoLens

Automated lens design from scratch using gradient backpropagation + deep learning. This project is built on the top of DeepLens framework.

We are planning to to build AutoLens as an open-source lens design software, imaging an open-source Zemax. Other algorithms (for example, end-to-end lens design and implicit representation) will be updated in DeepLens. Welcome to join us if you are interested in optical design! Contact Xinge Yang ([email protected])

How to run

Method 1

  1. git clone or download this repo
  2. run python autolens.py

Method 2

Run it in Google Colab

Open In Colab

Method 3

Use our packaged .exe file (upcoming)

Lens Design Examples

  1. Example automated lens design of (left) FoV 80deg, F/2.0, 4.55mm focal length, and (right) Full-frame, F/3.0, 50mm focal length.
AutoLens AutoLens
  1. 20 random automated lens design results for FoV 80deg, F/2.0, 4.55mm focal length.
AutoLens
  1. An aspherical lens with outstanding optical performance.
AutoLens

If you find this repo helpful, please cite our paper:

@article{yang2023curriculum,
  title={Curriculum learning for ab initio deep learned refractive optics},
  author={Yang, Xinge and Fu, Qiang and Heidrich, Wolfgang},
  journal={arXiv preprint arXiv:2302.01089},
  year={2023}
}