So I love IPFS, I love the idea of it. I think it's great.

But try integrating it into a project and see what happens. Here is a fun example. Development of IPFS is unfortunately a disorganized mess.

I'd love to build an application that loads resources from IPFS, such as a library specifically for books that stores an index of books, like libgen, or uncensorable front ends for smart contracts and things, but IPFS integration is sub par. The situation all round just sucks.

I lost interest when they started with their cryptocurrency bs, otherwise it’s stable and works.

For the exact same reason that nobody is building new centralized video platforms. It's just expensive to create

There is progress on IPFS as we speak though, I'm building a p2p search engine for it

Well, you'd need to make a video-hosting site in the first place. And you need to host all the videos even if you use IPFS if you don't want to provide a bad experience, so you don't escape any of the problems of hosting a video-hosting website. IPFS has its own challenges it adds over regular video-hosting site challenges.

So, it's not really worth it.

In my opinion, the coolest project currently on IPFS is Standard Template Construct. Just no video streaming.

What I read a while ago was that it had too high latency and/or not enough speed to run video streaming for Peertube. It seems like there's been some work done on getting Nix(OS) working over IPFS, and they asked about it on their latest community survey, but idk how much they're focusing on it.

Gave up on ipfs. Used too much memory. Too slow.

Isn't DTube using IPFS or don't they do that anymore?

Also found this cool semi-related OS project

https://github.com/oneloveipfs/ipfsVideoUploader

Years ago Tribler was trying to create streaming via Bit Torrent. Seems like something similar could be done with IPFS. Blockchains and Distributed Hash Tables aren't really that different.

What does D-Tube use?

I have been mapping out a possible way to do this. Pickling and scattering frames across the network, coupled with a client that stitches and buffers into a video. Whereas the manifest for each frame hash is compiled and stored separately.

I feel such a service is definitely possible. As for consumer/product worthiness I feel the other comments kind of covers the variables as to why you probably haven’t found a strong solution yet to streaming.

I definitely feel it is growing or at least it’s still in active development. I feel a lot of people either haven’t grasped it yet (aside from the NFT craze) or are not fond of the process of deleting content stored onto the network. Also a true real-time experience would require some more research on latency.

There's this project: https://watchit.movie/#/ https://github.com/ZorrillosDev/watchit-app

I never got to try it, but they are doing exactly what you want it seems. I don't know how much of their backend they have open-sourced, though. I definitely need to research this more.

They do have a matrix channel and a fediverse presence, so they already look quite alright from the start, and the project is very interesting.

You can see the health of the IPFS network and many other measurements at probelab.io

Because ipfs devs only care about money (crypto shit and nfts)