I recently discovered Super SloMo, a deep learning model that generates HQ intermediate frames between video frames to produce smooth slow-motion interpolation. While exploring the paper by Jiang et al. (2018), I came across an older implementation of this model that was no longer fully functional.

Seeing an opportunity to try in-practice some High Performance video processing technologies and to provide a working tool for others interested in this domain, I made several improvements: simplified its usage with an all-in-one Bash script, updated dependencies, and enabled Apple Metal (MPS) to support GPU acceleration on macOS systems.

Additionally, I integrated CodeScene to continuously analyze and improve code quality throughout development.