← Back to context

Comment by imiric

2 years ago

Most of those features are available in OSS tools as well. And for those that are not, there are alternative solutions that might take a bit of work to implement.

I'm not claiming that the OSS tools have feature parity with 1st party frontends, or that they won't require some sacrifices, or effort adjusting. I just think that the trade-off of losing some of the convenience in return for not being tracked and manipulated is well worth it to me, though I can see how it might not be worth it for others.

I do actually think that OSS tools provide a better UX. I can download the media and consume it offline, using any player of choice, on any device, at any time. I find YouTube's recommendations a nuisance, and I can turn those off in Invidious and Piped. Scrubbing in mpv is instantaneous for me for local files and even those served on the LAN, though there is a slight delay when playing directly from YT. There is also a solution for generating thumbnails[1], though I had some issues with it, and didn't end up using it.

At the end of the day, it's a personal choice depending on what you value most, and I'm not trying to convince anyone my choice is inherently better. Thanks for providing your perspective.

[1]: https://github.com/tomasklaen/uosc

>Scrubbing in mpv is instantaneous for me for local files

Yes, I agree that scrubbing in mpv or vlc is "instantaneous" but Youtube's web ui is even more hyperfast "instaneous" than mpv.

>There is also a solution for generating thumbnails[1], though I had some issues with it, and didn't end up using it.

For me, using an offline tool like thumbfast to generate timeline previews defeats the purpose of using Youtube's pre-existing timeline thumbnails that Google's datacenter already generated. Let me explain...

>I do actually think that OSS tools provide a better UX. I can download the media and consume it offline, using any player of choice, on any device, at any time. I find YouTube's recommendations a nuisance,

I'm guessing it's a difference in usage pattern. I'm often browsing a bunch of Youtube videos as a research tool. Like a "visual wikipedia" for various topics (especially DIY tutorials and products research). I want to jump in and out of videos fast. Downloading videos with yt-dlp to play in mpv isn't the workflow here. That's too slow and cumbersome. Instead, I'm sampling a bunch of videos and maybe a few of those will be ultimately be downloaded. E.g. Preview/scrub fragments of 10 related videos, read some viewer comments, scan some transcripts, etc... and eventually only yt-dlp 2 of them. This is why "mpv yt-dlp with workarounds" is not an acceptable substitute for using Youtube's web ui.

  • That's fair. It's indeed a difference in usage.

    My only usage of YT is queing up videos for short-term playback. So I browse a feed of my subscriptions in Piped, drag links of videos I'm interested in to a text file, and run a small script on my HTPC to download them with yt-dlp in parallel, and add them to a playlist. With a fast connection, it only takes a few minutes to download even dozens of videos at a time. Then I serve the videos on my LAN over HTTP with nginx, and watch them on any of my devices using any media player that can stream HTTP, which is usually mpv.

    I started a project some time ago to make this fancier, but honestly, this workflow does 90% of what I need, and I'm too lazy to change it.

    To each their own :)