Consider also Tube Archivist. If you just want to download a few videos it's overkill, but I use it to archive and index technical channels I like. It can do advanced full text and metadata searches on the transcription/subtitles as well as comments (and title and description). Much better than what alphabet provides, annoyingly.
I was looking at TubeArchivist. But didn’t like that it required 3 separate containers. Then I found Pinchflat on the selfhosted sub. It works as a single container and was easy deployment. Has been working well.
The file naming is predicated on having the database. For ephemeral storage this is understandable but any software that describes itself as "archival" should assume that it will not outlive the data it is preserving.
I was excited to have a PeerTube clone with download capabilities until I looked at the result. `yt-dlp` and clever configs and scripts will do the archival job properly.
I like how you're willing to archive videos but refuse to archive the playback mechanism for it. Is backing up a docker image really a deal breaker for you?
Google will never make it comfortable to use their products in any structured, advanced way as it is super beneficial to the user and gives away way too much control. Google hates that.
I've been meaning to migrate to tube archivist forever, but I have several TB of youtube videos I don't want to re-download which I gathered using tubesync. Luckily I have the format string used to save the files by name and since tubesync uses django and a postgres backend I might be able to tie them to the youtube ID. But migration still seems like it might be a bit of a nightmare
https://www.tubearchivist.com/