> notamy: I have an application on OVH (on the USD $3.50/month plan) that pushes/pulls >10TB/month
That entire discussion is recommended for anyone looking for a cheap VPS.
I seem to recall a friend having stability issues with their VPSes several years ago, so we stuck to the dedicated stuff from them, but it's been extremely good especially considering the price. Have you had good stability with their VPS services?
I am debating starting a Twitch -> YouTube video stream duplicator/archiver that would initially make money by auctioning available capacity with the long-term goal of being aquired by Twitch since their integration is so unreliable.
[1] OVH 10TB traffic throttling https://github.com/joedicastro/vps-comparison/pull/25
[2] Scaleway truly unlimited https://github.com/joedicastro/vps-comparison/issues/9#issue...
I actually tried the Hearthstone one and the very first clip in the current example 'Greatest Clips' (BJwDyxrplpo) appeared to miss the actual action (clicking "Disenchant" - which may actually have been the point since it could have been just a tease) but the rest of the clips seemed complete (and interesting).
I've thought about stream-jumping/recording based on simple indicators like increases in chat comments, viewers, followers, etc. How much of this could be built off Twitch's own 'clip' functionality (whether initiating them yourself or aggregating the manual curation of others -- neither of which AFAIK has an API right now) and collecting them later? Separate note I'm trying to hide in this pararagraph: don't overfit if you want to apply this tech to other streaming sites where real money is flowing (aka NSFW).
Personally I don't care so much about specific games on Twitch (except Street Fighter, which gets relatively little love but your videos are a real time-saver) instead of personalities. It might be worth offering this service to them, focused solely on collecting their highlights. had a tough time with the non-English streams but not sure what options you have there. I'm also interested to see how this will turn out for you using Twitch content if they notice that what you're doing is catching on. Twitch seems to be leaving a lot of low-hanging fruit behind for othes to capitalize on.
Feature wise: more playlists, maybe monthly and/or collecting the highlights of the highlights, with most comments/views/thumbs-up on previous YouTube videos. If there was a way to incorporate chat since most streamers don't in their videos, you should. https://github.com/PetterKraabol/Twitch-Chat-Downloader
Bug-wise, it seems like something is going wrong with the links as the end of this video, appx. 30 seconds of moving images but no links in Firefox with ad block [disabled as legacy]. (8ql3id1lJoM, ilkKuvuna10)
You're right about offering the service to the streamers - that's definitely the way to go to make a business out of it and it's something I've considered. However, I was mostly interested in doing the project for fun, and for some passive income, and making it a service would definitely not be passive.
The clips API returns language information about the clips, which you can use to filter them. Before that, I had to manually maintain a blacklist of non-English streamers.
I do monthly highlight videos, but they're solely based on clip views on Twitch - it doesn't use YouTube analytics, which I'm sure would improve the videos.
It is a cool idea to include chat - another thing I've considered but haven't implemented, though I've noticed some Twitch highlight channels (that do manually edited videos) do it. Thanks for the link to the downloader.
The links at the end of the videos are tricky - there's no api for that, so currently they're populated by a Firefox macro on a desktop that's supposed to be run every day - looks like there's an issue with it running! The better version would be to use a webscraper or headless browser to automate those clicks via the render server. That's what I'm supposed to be working on next, in fact...