AutoColor: Learned Light Power Control for Multi-Color Holograms

Yicheng Zhan, Koray Kavaklı, Hakan Ürey, Qi Sun, and Kaan Akşit

Description

AutoColor introduces a light-weight neural network that is able to predict required laser powers for a target scene when reconstructed using multi-color holograms (See Multi-color Holograms for more). The predicted laser powers could be used as the starting point for Multi-color Holograms optimization pipeline. This way, Multi-color Holograms pipeline is able to calculate multi-color hologram in just 70 steps rather than 1000 steps like the original Multi-color Holograms pipeline. Obviously, this improvement with AutoColor saves time and computational resources, bringing multi-color optimization times from minutes to a few tens of seconds.

Citation

If you find this repository useful for your research, please consider citing our work using the below BibTeX entry.

@misc{zhan2023autocolor,
    doi = {XXXX},
    url = {https://arxiv.org/abs/YYYY},
    author = {Zhan, Yicheng and Kavaklı, Koray and Urey, Hakan and Sun, Qi and Akşit, Kaan},
    keywords = {ZZZZ},
    title = {*Autocolor:* Learned Light Power Control for Multi-Color Holograms},
    publisher = {arXiv},
    year = {2023},
    copyright = {Creative Commons Attribution Non Commercial No Derivatives 4.0 International}
}

Getting started

This repository contains a code base for estimating laser powers required for a target scne when reconstructed using multi-color holograms.

(0) Requirements

Before using this code in this repository, please make sure to have the right dependencies installed. In order to install the main dependency used in this project, please make sure to use the below syntax in a Unix/Linux shell:

cd autocolor
pip3 install -r requirements.txt

Note that we often update odak, if this requirements.txt fails, please use the below syntax to install odak:

pip3 install git+https://github.com/kaanaksit/odak

(1) Runtime

Once you have the main dependency installed, you can run the code base using the default settings by providing the below syntax:

git clone git@github.com:complight/autocolor.git

cd autocolor

python3 main.py --settings settings/holoeye.txt --weights weights/weights.pt --input ANIMAGEFILE.png

Please note that ANIMAGEFILE.pngin the above syntax must be replaced with a target image. Target image could be at any resolutions and RGB but we highly recommend to follow the same resolution. If you are looking into finding sample images, consider visiting our images repository. In the above example, holoeye.txt saves the estimation at ~/output/autocolor/ANIMAGEFILE.pt. You can use this pt file that contains laser power estimation with *Multi-color Holograms*. In the *HOLOHDR*, repository, we provide a sample setting for this purpose as settings/autocolor.txt. A sample usage is as follows:

cd ..

git clone git@github.com:complight/multi_color.git

cd multi_color

python3 main.py --settings settings/autocolor.txt

(2) Reconfiguring the code for training purposes

Please consult the settings file found in settings/sample.txt, where you will find a list of self descriptive variables that you can modify according to your needs. This way, you can create a new settings file or modify the existing one.

If you are willing to use the code with another settings file, please use the following syntax:

python3 main.py --settings settings/sample.txt

Support

For more support regarding the code base, please use the issues section of this repository to raise issues and questions.

complight / autocolor