A user just tried this recently! Right now I'd just feed a huggingface model url and Kino will try it out on your footage in a way that can be undone / modified.

We are supporting more and more models natively (voice healing, scene edit detection, rotoscoping) given how annoying it can be to manually download local models or set up your own cloud inference. There isn't a great open source Mac option for this one so we're prioritizing!