During the winter of 2021, I've started looking around a bit at domains where my skills in expressive rendering could be useful. One of them was V-Tubing, as a large portion of that community uses anime-style avatars. After reading a bit on the subjet, I met Tanchyuu and Ascalon, and started working on porting one shader from Blender to VSeeFace for better anime face shading.
The algorithm itself was pretty simple, as it bends normals from a sphere projection to better match a face's shading. It can be seen from Ascalon's videos.
In order to implement that, I contributed some code to VSeeFace to set shader params dynamically.