A weekend project that converts video into a colorblind version of the video.
Currently, it only supports protanopia. I am protanopic, so I am able to Q/A the algorithm effectively compared to the other types of colorblindness. In the future, I may add other types of colorblindness, but I don't plan to.
For a deeper description of the algorithm I used, please see a talk I gave on the subject.
This project is licensed under the MIT license.
I haven't done C++ work professionally in 6 years, so I kept it as simple as possible. I have also made no effort to profile or optimize the code. There is room for improvement.
This project is a C++ project. You need three things:
- A local copy of the OpenCV library
- A local copy of Boost with Boost.Test
- Bazel, Google's OSS build tool
Open $project_root/WORKSPACE, and modify the path to OpenCV and Boost.
Use bazel build :all to verify that everything can be found.
bazel run :video_colorblind -- --input_file $input_file --output_file $output_file
Note the double-dash between :video_colorblind and --input_file.
This will process any video file that your OpenCV is compiled to handle.
bazel test :all_tests
Unit coverage is limited to the color functions. Modify carefully.
A video that simulates what a protanope will see
The RMSE in XYZ (between the original and protanope) is printed in the upper-left corner