Very excited for this, but a waste of energy if everyone is needing to process their video in real time.
As another way to look at this, where does this questioning of energy use end? Should I turn off my laptop when I go to the supermarket? When I go to the toilet? Should I turn off my lights when I go to the toilet?
My point is, we do a lot of inefficient things and there is certainly something to being more efficient. But asking “is it efficient” immediately when something new is presented is completely backwards if you ask me. It focusses our attention on new things even though many old things are WAY more inefficient.
That's what a software engineer would say who views resources as unlimited and free.
from the same guidelines: don't be snarky; don't post thoughtless insubstantive comments; reply to the argument instead of calling names; don't sneer; respond to the strongest interpretation - assume good faith; eschew flamebait; don't trample curiosity.
You're making a litany of unsolicited accusations while subtracting value in a conversation you're not a part of. Please stop with the self-appointed fixation on microaggression policing because it just creates pointless drama. Thanks.
Today we consume twice as much energy as we did in the 80s (and that's mostly coming from an increase in fossil fuels consumption). Datacenters alone consume more than 1% of global energy production, that doesn't include the network, the terminals, and the energy necessary to produce all of the hardware.
Energy-use or electricity-use? Only about 20% of total energy use is electricity [1]. I went through the math for EUV machines in more detail at https://news.ycombinator.com/threads?id=huijzer#42600790.
[1]: https://ourworldindata.org/energy/country/united-states
Who's "we"?
Worldwide we've gone from 4.4 tons/person to 4.7 tons/person (7% increase) from 1980 to 2023: https://ourworldindata.org/grapher/co-emissions-per-capita?t...
Yes.
Because today is today, and if we can project that the energy consumption of doing a task n times on the client side outweighs the complexity of doing it once and then distributing the result somehow to all n clients, we should arguably still do it.
Sometimes it's better to wait; sometimes it's better to ship the improved version now.
> Computation is really negligible in the grand scheme of things.
Tell that to my phone burning my hand and running through a quarter of its battery for some local ML task every once in a while.
The results could be cached, but it's probably unlikely that they would need to be used again later, as I imagine most videos are watched only once.
Another option would be to upload the generated subtitles to some service or p2p, but I believe that would also problem (e.g. privacy, who runs the service, firewalls for p2p, etc).
But yes, that would be quite nice, if there are enough people who don't mind uploading the file names of the media they play (or some other unique media identifier?) along with the subtitles to that service—and with user credentials? It would certainly need to be opt-in, which makes one wonder if it would be much effective at all.
Previously, that was used for mp3 album covers or something?
https://github.com/videolan/vlc/blob/f908ef4981c93a8b76805ad...
and to their own servers:
https://github.com/videolan/vlc/blob/f908ef4981c93a8b76805ad...
should could fetch subtitles as the same time ?
edit: cf, what "a3w" says too.
You can navigate to $foo.exampl.page and it will generate a website on the fly with text and graphics using AI. It will then save and cache the page.
It’s admittedly a useless but cool little demo.
I don't know how many times I've seen subtitles that appear to be based on a script or were half-assed, and don't match the dialogue as spoken at all.
So while I'm excited this feature is now available, having high quality subtitles cached in one place and generated by AI is the answer imo.