4 ms·
Short-sighted move, super long video data could be quite useful in the near future
by pr337h4m 2y ago
Short-sighted move, super long video data could be quite useful in the near future
- washadjeffmad 2y agoThe infographic says "New 100-Hour Highlights & Uploads Storage Limit". Past Broadcasts (VODs) and Clips are unaffected.
- moefh 2y agoNormal VODs already automatically "expire" (i.e., are automatically deleted) after a certain time. IIRC the time limit is between 7 and 60 days depending on your account type (e.g. whether you're a Twitch partner, whether you have Twitch Prime, etc.). Making a VOD a highlight was a way around that -- Twitch would never delete those.
- washadjeffmad 2y ago>Making a VOD a highlight was a way around that -- Twitch would never delete those. I'm not disagreeing that it was common knowledge that this was a way for non-partners to circumvent the regular retention policies, which is why this 100-hour limit seems like a pretty generous compromise. Clicking through random channels just now, I didn't see a single account with any Uploads, and most of the channels who had any content in Highlights seemed to use it pretty sparingly (<100 hours). It doesn't seem like a common practice, and Twitch doesn't seem like it's trying to eradicate history, just reign in some behavior that the platform didn't intend to support. If you're aware of certain communities who've made a practice of highlighting their entire streams (beyond 100 hours) without being partnered, maybe you could promote them here so people could help archive them?
- mistercheph 2y agoJeff?
- moefh 2y agoI remember when Twitch implemented the VOD expiration with the "highlight" system; the discourse at the time was that some people enabled VOD retention and then forgot about it, uselessly clogging the servers. So if you really cared about retaining VODs, you just had to highlight them. I think they just counted on no one caring enough, since highlighting every single VOD is a pain. It turns out that most people who cared enough about retaining VODs just upload them to Youtube. Youtube is simply a better viewing experience for non-live videos, and it can generate some revenue (though usually very small, unless you have a huge amount of views). One problem with Youtube is that it's more strict about (what it thinks is) copyright content -- for example, some otherwise "free"[1] videogame music is regularly claimed by on Youtube by someone who sampled the original song, so it registers as someone else's content to Youtube's content ID. [1] By "free" I mean that original videogame music is not usually actively protected (even though it's under copyright), because publishers love when people promote their games.
- madshougesen 2y agoAre you sure about the time limit? This VOD is over 2 years old [https://www.twitch.tv/videos/1891768073 https://www.twitch.tv/videos/1891768073](https://www.twitch.tv/videos/1891768073 https://www.twitch.tv/videos/1891768073)
- moefh 2y agoWell, the settings page on Twitch has this help text on the "store past broadcasts" toggle: Automatically save broadcasts for up to 7 days (14 days for Affiliates, 60 days for Partners, Turbo and Prime users) Maybe the video you're seeing is a highlight? Either way, I can't see it because I don't subscribe to Khaldor (I used to love watching his Starcraft 2 casts back in the day, though).
- EwanToo 2y agoDeleting it from the public site doesn't mean they're not keeping it for internal use...
- anticensor 2y agoWhat if they intend to rotate storage, replacing old content by newer content over a period of 2 months, plus 100h/user of non-replaceable content?
- phyzome 2y agoWhat ever for?
- DecoySalamander 2y agoSupposedly for training models to generate gameplay videos, like what Microsoft (and others) have presented [0] recently. [0] https://www.microsoft.com/en-us/research/blog/introducing-muse-our-first-generative-ai-model-designed-for-gameplay-ideation/ https://www.microsoft.com/en-us/research/blog/introducing-mu...
- wongarsu 2y agoAlso virtual environment creation, agent training, etc. For any given game you can create a small dataset of recordings of both player inputs and gameplay footage, use that to create a model that can derive inputs from looking at footage, and then create input sequences for your huge backlog of gameplay footage. From there you can use the backlog to train AI that either recreates realistic player actions from screen inputs, or AI that recreates the entire game (like the AI minecraft) Not to mention the huge amount of voice samples and webcam footage you could use for more typical voice cloning, text to speech, human avatar creation, etc
- whywhywhywhy 2y agoIf people watch it currently then it has value in a training set.