The technical side of the organization of broadcasts and the formation of video streams on the Internet?

Question

IT education

They were asked to find material on the topic "formation of a video stream and organization of broadcasts on the Internet." On the theory of signal transmission. I rummaged through Google, I couldn’t find anything full-fledged, I thought, maybe Habr will save?)

Are there any experts on the technical part of organizing broadcasts. I found one PDF, which describes the process of working with video broadcasting, but it's more Kom. Offer than description of those. parts.

Answer 1

There are two options - for browsers - these are WebRTC , nginx-rtmp and their analogues.
Not for browsers - these are H323 and SIP protocols + special software or special hardware.
Here you can paint for a very long time and tell how it works, there are a lot of nuances. Since it all grew out of telephony, the technology has a lot in common with it, for example, the division into "signaling" and content:
There is a separate protocol (webRTC, H323, SIP) that is responsible for metadata transfer and connection setup and a separate protocol ), which is responsible for the media stream itself (audio separately, video - separately)
Signaling usually works over TCP, media data usually goes over UDP. The use of UDP is due to the fact that the loss of several packets does not affect the media data stream in any way (well, the interlocutor will stutter, or the picture will be smeared).
The architecture for broadcasting to the browser is something like this
1) Media source - a file, a stream from a video camera / microphone, a mix from a screen capture and a camera (obs-studio)
2) Transcoder (s) - a solution that recodes the source stream into several output formats
- with different resolution
- with different codec, for better compatibility with devices. For example, the old iPhone eats only a very specific h264 profile, and it is better to serve a Linux machine with VP8, because h264 needs to be installed manually - it is proprietary.
3) Broadcast server(s) - this could be nginx-rtmp, icecast or something proprietary. They just provide the issuance of the media stream in the right form - HLS (Apple format, old iPhones eat it, WebM is an almost universal format, all modern browsers eat it, WebRTC is an even more universal format)
In order to reach the audience as much as possible, you need to support several formats images (different codecs, different resolutions) and several broadcast formats.

Answer 2

Our site describes several typical scenarios for creating content delivery systems.