Hacker News Reader

Comment by dfee | Hacker News Reader

There’s one feature of OBS that’s missing for me, and that’s recording sources to separate files. I believe this is called “multicording”, and it something I’ve only found in paid software, like screen flow.

Here is the multi year discussion in the OBS forums: https://obsproject.com/forum/threads/recording-sources-to-se...

View on HN

This is effectively our single biggest requested feature right now:

https://ideas.obsproject.com/posts/41/multiple-video-outputs https://ideas.obsproject.com/posts/7/multi-service-streaming... https://github.com/obsproject/rfcs/pull/8

Much more details in those links than the forum thread you posted that explains why we haven't done it already. Most of the issue is UI, which is most of what is discussed in that RFC. We need to change the paradigm on how outputs are handled in the UI to support things, and UI is really, really hard. We basically have to throw out a large portion of the application UI and start over to accomplish this, which is a monumental undertaking, and we just haven't had the time or resources to devote to it with all the other higher priority items that keep coming up.

We want it. Users want it. It's definitely not an if, it's a when. Unfortunately, I don't have any timeframe I can give you. OBS still only has one single full-time developer working on it, so resources are limited.

If you are using any of the many hardware-accelerated encoders available on desktop today the CPU is not at all limiting. Quicksync can run 9 HD h264 encodes, NVENC can run 4 to 17 depending on the card, etc.

Most OBS users don't configure software encode because of the resource issues.

This is very true, but it's worth noting that NVENC requires a Quadro card for more than one or two streams (depending on the model).

There are patches for the drivers to remove the limitation. No idea how that affects your license to use the drivers and such, though.

https://github.com/keylase/nvidia-patch

Windows version available in a directory of the main one, too.

I wasted a good chunk of a day looking into this. If you start multiple instances of OBS (you need to do this on the command line otherwise clicking will just foreground the current instance) then "multicording" should work fine, though perhaps using lots of system resources (the fan goes nuts). This was on macos.

Overall, I realized it's better to just use OBS and have it mixed into one file/stream. You trade off a bit of flexibility to do after-the-fact editing but save a lot of production/editing time.

It's not better--you're just not paying for tools that do it. When it works it solves entire categories of problem and it's one of the reasons I pay a lot of money for vMix. (In fairness, the OBS developers I know are super sharp and I'm pretty sure that this is somewhere on their to-do list.)

The main big one for me is similar - not recording but being able to stream to multiple servers.

We can do things like run nginx with nginx-rtmp-module as a proxy, but then you have to put stunnel or something in front of it for Facebook because nginx-rtmp-module doesn't support rtmps (and development seems completely stalled).

I've been looking into building a containerized FFMPEG stream routing solution but haven't been able to think of a good use-case. What is your alternative destination after Facebook RTMPS?

We’re just trying to stream to YouTube and Facebook at the same time.

Actually, ffmpeg is the other solution I was looking at playing with on the weekend.

FFMPEG is a great tool for any IP video streaming workflow. You can easily define multiple outputs. Check out the live streaming guide: https://trac.ffmpeg.org/wiki/StreamingGuide

I just haven't yet found a good way to export active stream status from the CLI out to other interfaces for the tool I'm working on.

The protocol is one thing, the other nearly more important one is the bandwidth/encoding for each platform. On YouTube you may want to put out a high-res/high bitrate version and let YouTube handle the churning down to lower resolutions. On Twitch, unless you're a partner, you want to pick a low enough bitrate, otherwise your viewers will be hit with a high bitrate and some end up unable to watch the stream.

So lots of reencoding happening on the fly, which is why I know of some streamers using dedicated streaming computers, that take the OBS stream as input, but I don't know what software they use to distribute it.

[0]There are services that do this. Of course, this goes against TOS for affiliate and partnered streamers on twitch. As of my last check, it's not an issue for other platforms yet.

Being able to do this is actually the number one reason why I strongly believe nobody should ever become a Twitch affiliate.

[0]https://restream.io/

What's the use case?

Exporting to something like iMovie to manage chroma keys (video alpha channels), or manage the two videos separately: zooming in, highlighting in the background recording, effects for the foreground recording etc.

For editing together screen/face/camera streams later probably.

Reply on news.ycombinator.com