Replies (51)
Now we just need a solid video client.
👀
nevent1qqszevvhychldtuhu35atyql47ktcpdye778z4xkdf28w7n7j3e6tfcpramhxue69uhkummnw3ezuetfde6kuer6wasku7nfvuh8xurpvdjs9uzg7m
This was very mindblowing, for sure.
I pasted that command on my terminal and it played some music.
😂 you were rickrolled
You're making me nervous ( a good nervous )
Yep we will probably need to pay someone for hosting if we want to compete with Usenet? I get like blistering fast downloads and almost anything I can imagine for like $20-30 a year.
This is.some amazing stuff,
@franzap is the goat
View quoted note →💪💪💪💜
The missing part I see is a standardized way to do discoverability of the blobs. Something like DHT, but its just an event written in a particular nip that link users to the various pieces.
Nip-01 + blossom blob + Nip-?? to have a decentralizable, replicable, indipendently updatable (clients should decide who trust in updating nip-?? note)
You're gonna make me work on Video/Audio widgets asap aren't you? :haha:
Need a #5yo version on that last line :prayinghands:
Yeah, we were talking about this a while back, when we started splitting up and indexing publications that contain non-text data, but I don't know if the plausible deniability thing actually works in a court of law because everyone low-key knows you've implemented it to help traffic kiddie porn.
I didn't even realize that nobody was doing this, already. Sort of funny. We do this with PDFs and stuff, too. Break it up into a stream, including the pictures and videos and games (people put computer games into PDFs because of course they do). But the whole thing makes me nervous, so we only store Asciidoc and export the embedded media to a media server, and call it from there.
We don't work with anything encypted, but you do you. The police know where we live. LOL
Breaking it down into a stream also means you can set anchors into the stream, and refer to those anchors in a highlight, which is seriously awesome.
Oh :eyes: so you mean when the publisher of the video/podcast/song/... defines sections, you could just chop up along those marks?
It’s easier to use time stamps for anchors. Subtitles for example are just text with a timestamp.
Breaking up a video in segments is mostly important for handling uploads/downloads of large files (>100MB) especially over slow connections. It also has benefits for CDN caching.
We will have paid blossom servers.
- Uploaders will have to pay a certain amount of sats/GB/month for storage.
- with automatic cashu payments from a NIP-60 wallet
- Users / Downloaders might need to pay a few sats for bandwidth to get content from the blossom servers.
- Only payments will incentivize people to run blossom servers with Terabytes of storage and Petabytes of bandwidth.
I am seeding 750+ torrents. Only placing ~ 2 TB/month. Incentive: fuck all of the streaming services.
uploading ~ 2 TB/month*
Yeah, sure. And you can add arbitrary anchors to a a video stream that you're converting. Not just a point in the movie, but a face on the screen at the point in a movie.
Like, if someone is giving a Powerpoint presentation and presents a chart, you could highlight a bar in that chart and add some text, or cause the video to pause and add an explanatory slide.
Epic options we have 🌸
We're using the anchors on the images themselves.
We also alter the text inside of images by breaking it up into layers and then replacing the text layer. There's been a trend to put table of contents directly into images because it looks pretty, including page numbers, on PDFs, but if you convert PDF into something pageless (like Asciidoc), then you need to remove those references and replace them with something like relative placements or a hyperlink to a new anchor.
I'm just sort of surprised by everyone getting so excited about this, as it seems sort of obvious to me. Everything is numbers, to an operating system, and a matrix of numbers can always be broken down into sets or strings, transmitted in parts, and then pasted back together. The computer isn't actually saving pictures or sounds, it's saving text instructions for recreating pictures or sounds.
This is just signal processing.
I don't know what CDN caching is, but it's probably easier and cheaper to store lots of little chunks, than one large one, as it's easier to find a spot in the memory that size. Little chunks are probably generally better than large chunks, now that processors and networks are so fast.
Chaotische Lagerhaltung, but for computers.
I love your incentive.
Streaming services still have superior UX, we need to work on that.
I want to: find video, watch trailer, sounds good, play and starts streaming and I'm certain it won't stall. Will pay many sats for that.
Well, people will do it for free, but free stuff often has dodgy quality or unreliable service.
That is already happening, in my example nak is pulling an event of kind 1663. Not sure where the nip01 is.
Kind 1663 is a placeholder and draft proof of concept for now. It describes the segments and the servers. As segments are hashes they can be djscovered and updated by other mechanisms too
I think thats cool and hopefully we can do zap splits to send a (major) portion to the creators who publish directly to nostr. Fuck the middle men. Bitcoin disintermediated payments. Nostr will do the same for comms and content distribution.
I don't pretend I discovered gunpowder. So far, all videos are entire files and that's it. I wanted to showcase and start defining a NIP for HLS because of the insane amount of advantages it unlocks
I did not implement it to help traffic kiddie porn, wtf. Those people can rot in jail.
By that same token Signal should have a backdoor because its encryption is helping that traffic too. We've seen this argument countless times, it's tiring.
I'm extremely happy to pay sats for youtube videos I know I can access fast, without ads, and that don't prompt me to "sign in to confirm you're not a bot"
I’ve been experimenting with HLS on blossom but I had put the m3u8 in blossom as well.
Generating the m3u8 playlist dynamically makes more sense and enables nice features like the aes encryption.
I was solving the server failover on the fly in the browser while playing but doing it before playing with HEAD requests is also reasonable (short delay).
We could have a kind 21 video event pointing to multiple 1663 events and generate a master playlist dynamically to support multiple resolutions.
I will build a browser version of your demo if no one else beats me to it.
Failover on the fly sounds even better!
Mine was a shitty proof of concept, you are much more familiar with all this so happy to keep chatting about a potential NIP that makes sense.
Just struck me as someone "inventing" something we have been doing since the telegraph, and collecting hysterical praise from people who don't understand the topic.
That said, I like the idea of a chaotic logistics system for media, analog to the one we built for text. Of course, it opens the door to remixing the media or referring to the media piecemeal, but that might not work, for third parties, if its stored in an encrypted state.
We were also talking about doing this with code on Blossom servers, a while back. Any large data, really.
It's already the case that people can transmit images in encrypted format, but who wants terrabytes of that sitting on the server in their house or the server registered in their name?
Personal choice, I think.
I'll build another tool to scour nostr for these events and detect if a segment ever is CP or illegal, so it can be removed.
So, it would rebuild them, decrypt them, scan them, and then issue a 1974 report?
Segments are usually fixed byte size or duration to optimize for storage and delivery, anchors/chapters are not. They need to be defined separately.
Yeah, we let the person doing the conversion set the anchors and sections. It's primarily meant for images. PDFs are full of image.
If that works, it would solve the problem, but it would also remove the plausible deniability argument, as it proves that you could find out what you were storing, with a reasonable effort.
Reasonable efforts are reasonable.
I feel like the encryption damages the potential without adding anything, since it can't be securely encrypted without making it impossible for the public to rebuild it. Encryption adds value if only the person who receives the package can open it, or you are trying to circumvent censorship in-transit. We already automatically encrypt in-transit relay communations, tho, and these aren't only meant to be pay-per-view, or something.
But maybe I'm just confused and missing the point of encryption, in this use case.
Yeah, makes sense. Little benefit anyway I think.
For clips (etc), is it easy to just only load the blobs needed for the clip? #noobquestion
Is each segment encrypted separately? That would be better, at least, as you could then decrypt and utilize one in isolation.
Otherwise, wouldn't missing one segment make the whole thing undecryptable?
I'm not sure about that, tho. I'm not a cryptography expert.
The player does that if it supports HLS
Segments are encrypted separately, HLS standard
Perfecto! :Check:
Thanks by advence for this tool, it is nice to think about it now and not when it is too late !
Thank you
@Laeserin 🇻🇦 for asking the good questions too !
Since I started following my manager's new trading plan, I've seen a significant improvement in my financial situation. I'm truly grateful for the support I'm receiving—earning $150,000 as a beginner in crypto trading is a huge milestone. If you're looking to get started, I highly recommend reaching out to Mrs. Susan. You can contact her via email at susandemorirs@gmail.com or on WhatsApp at +1 (472) 218-4301
Since I started following my manager's new trading plan, I've seen a significant improvement in my financial situation. I'm truly grateful for the support I'm receiving—earning $150,000 as a beginner in crypto trading is a huge milestone. If you're looking to get started, I highly recommend reaching out to Mrs. Susan. You can contact her via email at susandemorirs@gmail.com or on WhatsApp at +1 (472) 218-4301
Since I started following my manager's new trading plan, I've seen a significant improvement in my financial situation. I'm truly grateful for the support I'm receiving—earning $150,000 as a beginner in crypto trading is a huge milestone. If you're looking to get started, I highly recommend reaching out to Mrs. Susan. You can contact her via email at susandemorirs@gmail.com or on WhatsApp at +1 (472) 218-4301
