Hacker News Reader

Comment by saurik | Hacker News Reader

Sadly, it requires RTMP to push input, so the sources either need totally custom software such as OBS or Flash installed in the browser; in 2017 I was sort of hoping to see WebRTC as one of the ingestion protocols (with an API that let you just get an offer plus candidate set for a new stream and another couple APIs for setting the answer and adding candidates as they stream back).

View on HN

I think this is designed for OTT video broadcasting and not so much video conferencing. RTMP remains the defacto standard for push ingest, although that it changing with the wider adoption of HEVC, which RTMP can't carry.

FWIW, maybe I am just on the leading edge of the "changing", but I am specifically excited about WebRTC for video broadcasting due to not only how devices now have hardware support for HEVC but also for the lower latency and lower barrier to entry afforded by WebRTC. WebRTC is a protocol with a widely deployed implementation that is already tied to people's cameras, not a use case, and has absolutely no ties to video conferencing.

Is there any technical reason you couldn't have HEVC over RTMP?

Also, what hardware can do real time HEVC encoding?

I encourage you to request support for WebRTC.

Is it possible to apply a dynamic filter graph a la FFMPEG? For instance, ingest two RTMP streams and apply a filter to overlay one on top of the there in a PIP configuration while preserving both audio tracks? That would be very useful

@simook do you have a direct email to request features? Thanks

You can request features over at https://forums.aws.amazon.com/forum.jspa?forumID=269

Reply on news.ycombinator.com