scrubbles

joined 2 years ago
[–] scrubbles@poptalk.scrubbles.tech 5 points 1 day ago (4 children)

The problem exists of how do you get files to the server. Again for a big video file you need a place to have the original accessible for the entire duration of the transcode, and you need a drop point to place the file when it's done. And if you're doing that, more than likely you have something like Plex or jellyfin, so why not just hook it up to your existing file locations?

I run jobs all the time from my cluster, but the issue always comes back to persistent storage. Where are the files you need to act on, and where should the finished results be placed? You're trying to skip those important steps here.

[–] scrubbles@poptalk.scrubbles.tech 5 points 2 days ago (18 children)

I'll ask the question why do you want something hosted, if you don't want the files to be on the server too, or at least accessible via a mount. There's a couple projects who do that for a reason, you can't just upload an 80gb video file through a web interface simply. For home use it makes more sense for it to be connected to some sort of nas and you point it to the file on the nas, and then point where on the nas you want the output file.

As written, I don't think you'll find a lot. I think what may serve you better is finding a server solution that watches a directory and then runs transcodes based on pre defined templates, and spits them back out in a known location.

[–] scrubbles@poptalk.scrubbles.tech 4 points 2 days ago* (last edited 2 days ago) (1 children)

Holy yikes batman. That's a lot of trust and open ports right there to the public internet. Not to mention running a tor node is incredibly risky for the hoster personally. Individually these projects are good, in one bundle I'm heckin suspicious.

Edit: it's just a relay node at least, but still, this is targeting inexperienced people to run a bunch of containers, and I don't think the risk is fully explained there.

[–] scrubbles@poptalk.scrubbles.tech 4 points 6 days ago (1 children)

Just be careful, they can be fun but they aren't the real deal, but I also won't say don't use them either. They're meant to have fun with, not be a dependency. That being said, I'd be curious if they can help direct with communicating. I know for me I get frusturated when I say something and it comes off a completely different way. Having your character define when that happens and why would be interestesting

[–] scrubbles@poptalk.scrubbles.tech 34 points 6 days ago (10 children)

Fun project, but don't get lost in it. I'm seeing people start to lose grips on reality by talking with AI too much because they're lonely. Not directly saying that you are, I had fun with it, but... Weird times

This is MY safe space and you can't come in!

[–] scrubbles@poptalk.scrubbles.tech 14 points 2 weeks ago (1 children)

if only it isn't the thing he said he was going to do

[–] scrubbles@poptalk.scrubbles.tech 25 points 2 weeks ago (1 children)

Here we were all begging them to listen what would actually happen a year ago to them. All they had to say was get fucked liberals. Well here we fucking are.

Docker compose in git. Env in 1password or whatever password manager you use. Most support uploading a raw file.

[–] scrubbles@poptalk.scrubbles.tech 13 points 4 weeks ago (2 children)

I mean, you’re right here.

Thank you I thought I was going insane. It's not Reddit levels of activity, but I'd say this is pretty dang active. Stats on the right show almost 600 users a day, ~7k a month, almost 100k comments...

Great callout, thanks for posting

Yeah it'd be a LOT of constant wireshark and reverse engineering to figure out every API it calls. Then probably something in the middle to sit on the host, need to figure out https certs since you'd be spoofing the host, and of course making sure you get the responses absolutely correct.

Not impossible, but it's not trivial anymore either.

view more: next ›