Sofie: open-source web based system for automating live TV news production
nrkno.github.io378 points by rjmunro 5 days ago
378 points by rjmunro 5 days ago
As a live television newscast director in a major market, I would be very interested to see a feature comparison between this product and its main competitors: Ross OverDrive, Sony ELC, and Grass Valley Ignite.
Due to the substantial complexity of these automation systems, they tend to have a lot of inertia. But if anything could drive a station group to make a change, the "free" part can be effective.
I did take a look at the supported hardware (1). I think that's the pain point for many shops. Free open source production software is great, but being forced to choose form hardware products you don't prefer is a pretty tough tradeoff.
Historically, I suppose that's been one of FOSS' big challenges.
(1) https://nrkno.github.io/sofie-core/docs/user-guide/supported...
It looks like great support (Blackmagicdesign) for building a small broadcast studio from scratch tho.
I could see BMD embracing this. There are lots of studios that are not commercial broadcast that could really use a system like this.
Isn't one of the problems with hardware support is that hardware vendors have agreements with the competitors you listed?
Computers are fast enough now that once you can get the signals into a machine, many of the special functions that previously required dedicated hardware can now be run in software? With proper timing signal distribution of course.
Seems like 12G SDI to SFP+ would enable server class machines to subsume most of the special function hardware.
Disclaimer... I am a director and not an engineer. I can only give you my relatively limited understanding....
> It looks like great support (Blackmagicdesign) for building a small broadcast studio from scratch tho.
Agreed!
> I could see BMD embracing this. There are lots of studios that are not commercial broadcast that could really use a system like this.
Also agreed. Black Magic definitely makes a lot of reasonably-priced and very capable gear. They're not a major player in the TV automation space, but perhaps with the help of Sofie, they could make inroads.
> Isn't one of the problems with hardware support is that hardware vendors have agreements with the competitors you listed?
That's not a topic I'm knowledgeable about. It is my understanding that most shops who have a particular vendor's automation platform will also have that vendor's hardware running at its core. In all the shops I've seen, the switcher that's controlled by the automation system is made by the same company. Or if its another vendor's product, it's sold and provisioned along with the automation system when its purchased. Other stuff like audio mixers, robo-cam products, clip players, and CG/graphics platforms can be from other vendors.
> Computers are fast enough now that once you can get the signals into a machine, many of the special functions that previously required dedicated hardware can now be run in software? With proper timing signal distribution of course.
> Seems like 12G SDI to SFP+ would enable server class machines to subsume most of the special function hardware.
For audio, I think that would be a relatively easy lift with technologies like Dante. However, in most TV stations, you're going to need to literally plug upwards of 100 HDSDI video cables into a piece of hardware so that those sources can be switched to on TV, mixed and keyed on multiple mixed-effects banks, and viewed on multiviewer screens in the control room. I don't know that a regular-ol' PC has what it takes to take in and simultaneously process that amount of video. But just because don't know about it doesn't mean it doesn't exist. ;-) Just haven't seen it yet.
Sofie drives the matrix (black magic videohub for small ones like the 120 squared, but ours are 1000+ which black magic won’t do), the audio mixers, the video mixers, the graphics and by machines (Caspar), etc. your mixers don’t need that many inputs - a typical one might be 24 or 32 inputs with a few ME banks.
All these devices use standard protocols, or it’s just a new plugin for sofie ti drive it.
Of course increasingly the industry is using 2110 on spine and leaf networks rather than SDI. I don’t know if there any COTS mixers aside from the vmix/obs level, I believe some 2110 controllers will provide video matrix style interfaces. Nmos seems challenged in this area from what I hear.
>For audio, I think that would be a relatively easy lift with technologies like Dante. However, in most TV stations, you're going to need to literally plug upwards of 100 HDSDI video cables into a piece of hardware
I don't know the TV stations requirements, but you can maybe have 10 interconnected servers that manage 10 HDSDI flux each (and can send them on another if required for processing) ?
> Computers are fast enough now that once you can get the signals into a machine, many of the special functions that previously required dedicated hardware can now be run in software?
I’m a big fan of the Richard Cartwright view of asynchronous signal processing
https://creativecow.net/matrox-video-announces-nab-2023-line...
But I don’t think it has the traction isn’t deserves. Too many people in the industry are still wedded to ptp timing their packets to arrive in the same 30us windows.
I would love to ask some hard questions of this solution.
Lets say I have a very simple workflow.
Camera and CG in -> conversion to RGB/YCbCr -> compositing -> pass to broadcast encoder
Conversion and compositing can be done at scan line speeds using conventional hardware, so latency is at most a frame. With an asynchronous workflow this is not possible anymore, let’s pretend the network infrastructure isn’t an issue and is operating perfectly with low latency, I don’t need cots hardware on the processor because without some sort of DMI even NIC -> GPU is several frames of latency, the GPU then needs to do the processing, then you again need to GPU-> NIC.
I then need to reorder frames at the receiving device, the stream encoder. Because its async, frame 2 might arrive before frame 1. So now I need a buffer there.
I do not see how this doesn’t add significant latency to the path even in the most simple setup. Add internet services like AWS and your latency shoots of to tens of frames before you even hit the media encoder.
I’ve seen stuff going to and back from AWS with sub second latency, even with h264 encoding.
BT sport have operated remote production channels from AWS for a few years now [0]
I’m no fan of cloud, but the latency to a nearby DC isn’t high, and when your feeds are going from the field anyway it doesn’t make much difference.
I remember OBS (the Olympics, not the open source software) bemoaning the lack of bandwidth available in cities and them having to build data centres in place like Beijing for temporary events, because there simply isn’t the multi-bit links to AWS etc available.
[0] https://www.svgeurope.org/blog/headlines/stratospheric-revol...
This kind of makes my point though.
The hardware AWS would be using, NICs etc do have the level of DMI needed for the latency.
Heres the catch though, the workflow I described can be done with maybe 2 frames of latency for like 400usd, and you own the hardware . I wouldn’t be surprised if some of AWS products are 400usd per hour.
You can setup a very competent broadcast system for not much money and use some actual cots hardware and still have significantly less latency that a second, probably in the low single digit frames.
I don’t the solution proposed by matrox isn’t possible, I just think for most use cases it is very expensive both in actual cash and in latency.
bmd is in a decent position to help with this. making davinci be the nle that ties into this like avid / airspeed or whatever ppl use now, seems pretty cool.
I fall pray to this often. Internal complexity and growth over time lead to great giant feature charts and comparison matrixes. But sometimes you just need a tool that gets a job done.
It's one thing for something simple to not be a drop in replacement. But simplicity and minimalism can also be a virtue. Can this complete the task in an environment designed around using it?
I dabbled around with it a few years ago. It’s a framework and you have to implement all the surrounding parts or copy them from others. I was able to get it to accept and display a rundown from our NRCS. (Had to find some frontend code from a different TV station somewhere to make it show anything IIRC.) But the amount of customisation necessary to make it work with our hardware was more than I was in the mood for, so that experiment ended quite quickly again.
Do you have any recommendations of where I could read more about integrating content on those softwares? I help run a service that provides content for news sites and I would like to make it easy to make that content available on newscasts.
What you're trying to accomplish doesn't really translate into TV production automation systems. The systems being discussed in this thread don't "integrate content" per se.
You may be confusing what we're talking about with an NRCS. (Newsroom Computer System.) The biggest players in this space are ENPS, iNews, Dalet Galaxy, and (to a lesser extent) Ross Inception. As it pertains to making it "easy to make that content available on newscasts," I don't know of any such method. You may be inferring that these systems employ a higher level of sophistication than they actually do. My advice: format your news as plain ASCII text. I really can't think of anything else you would need to do to it.
how are you still a director??? i miss tv fondly but the pain they inflict on everyone and the hours / pay / etc make the best people bounce.
someone on hn surely could use their talents for good elsewhere haha
Thank you for the vote of confidence. :-)
I do ask myself that sometimes. It sounds weird, but I think it's what I was put on this earth to do. Yes, it is a cruel industry at times and pain is indeed inflicted just as you assert. I guess I'm just built for it. And I've been doing it for so long that I've built up a really thick skin and I'm just not that fazed by its unpleasant aspects. I can honestly say that it's a fun job. It's the job I always wanted when I was a kid and I still absolutely love it. (I'm fortunate enough to be compensated at a reasonable level, so that helps.)
People think it's stressful and I suppose it is. But the nice thing about it is that when the newscast is over, I'm completely done. And I have the luxury of knowing, down to the exact second, when that moment will be. I don't take my work home with me. There's nothing to stress out about (until the next day.)
Another thing... unlike an airline pilot or a surgeon, no matter how poor a job I might do, no one dies. That's kind nice too. :-)
> no one dies.
Not counting all the people who watch a particularly shocking news story and then have a heart attack.
That's the thing though, this isn't really "free" software as much as open. NRK is funding it and created it for their use, that's cost money. They spent money on supporting the hardware they clearly had and wanted to for their production. Any other user with their own setups they want supported will have to spend money on developer time as well.
Free software has a well established meaning at this point, and it's not "didn't cost any money to produce".
Looks like this is MIT licensed https://github.com/nrkno/Sofie-TV-automation
Indeed, people usually contribute to FOSS by supporting the authors group or nonprofit directly, and contributing features and bug reports/fixes.
However, from a maintainability standpoint it is important that a project solves their own needs first. The "eat your own dog food" advice is important, or groups end up fragmenting a project into every pet use-case.
Best regards =3
[flagged]
You are trying to establish that the financial meaning of free software is the primary meaning, while the freedom aspect has been the primary meaning. You would do better to change your focus to searching for false claims of "no-cost" software or something rather than trying to undermine an established meaning that is already understood and widespread.
Free software, libre software, libreware sometimes known as freedom-respecting software is computer software distributed under terms that allow users to run the software for any purpose as well as to study, change, distribute it and any adapted versions.
Everyone starts out as a leech, as FOSS offers you freedom to behave in such manners.
However, unless you support the community in your own way, than people are just along for the ride.
Remember to try and be kind... even if you don't agree with your peers =3
It was mostly developed for NRK by Superfly.tv. They are available to extend the system to other hardware or customise it in other ways if the broadcaster doesn't have the expertise to do it themselves. It's already used by several other broadcasters, for example, the BBC use it for their Newsround program: https://www.linkedin.com/posts/ryanwmckenna_great-to-see-new...
"Free software" is a term of art in the software industry: https://en.m.wikipedia.org/wiki/Free_software
The definition on that page is accurate:
Free software "allows users to run the software for any purpose as well as to study, change, distribute it and any adapted versions. Free software is a matter of liberty, not price; all users are legally free to do what they want with their copies of a free software (including profiting from them) regardless of how much is paid to obtain the program."
Very cool they decided to build this and release it to the world instead of just buying an extremely expensive commercial system. The backend play out server is CasparCG which is also open-source [1] (they run their own fork, for stability reasons I expect).
Another similar related automation system (shares some parts and libraries) is SuperConductor [2].
Looks more practical, as when dealing with hardware access C/C++ is the only way to get the latency issues into tolerable playback stream formats.
Cool that it also supports OBS Studio by the way =3
Very cool. You can even control the prompter using a Joycon!
https://nrkno.github.io/sofie-core/docs/user-guide/features/...
How does one break into this industry? I recently started playing the video game Not For Broadcast and have fallen in love. I’m sure it’s a very romanticized and simplified experience but I’ve been having a blast at perfecting the “art.”
Very cool coincidence to see this on the front page right after a sesh with the game.
The answer for anything video production is: start making stuff.
Nobody can see your product if you don't make it. You can't make good product without practice. So get practicing. You need reps.
And when making that stuff, turn your camera the right way. The last thing we need is more door-shaped videos.
May seem a bit niche but man I wished we had this 20 years ago when trying to start a student TV station on a budget. The pro tools at the time were ridiculously expensive.
Any recommendations to what similar FOSS could be used for a radio station?
Has anyone tried it? I tried to set up a simple livestream a year ago and it was far from simple. It felt very hackish and documentation was lacking. I guess putting more time I would succeed in the end. Did ask if I could hire the developers to do setup for me, but didn't really get anywhere with it :/
How does this handle things like replays that are queued dynamically during airing? For instance on a talk show there may be occasions where the host wishes to replay a section of a guest interview or to pull up a clip to play while talking over it. Would the operator override an existing Part or updating a piece in a part? Typically this is handled live with EVS.
From a skim of the docs, it looks like those are handled as "adlib pieces", and can be pulled from what's currently playing or from other buckets.
Would Sofie work for producing live streams a la twitch? I've noticed most bigger streamers end up with a hodge podge of automations and integrations built around OBS which they have grown organically ever since they were small and built everything themselves. I wonder if this would be suitable upgrade.
It could, but it's aimed at news style programmes, ingesting information from commercial news gathering systems and turning that into broadcasts. You can reorder the stories and make edits even while the program is live, and e.g. handle breaking news events interrupting your show.
There's support for ingesting from Google sheets, and from a demo app, so you could use those, or develop your own ingest system to create the stories.
If anyone happen to pass by Hudiksvall in Sweden, you’ll be able to get a live demo of this (ping me!). One of the core developers sits in the same office house as me. :)
Ruby code is always a joi to read.
Can you write real-time shaders in it?
That's not what it does. You would have your real time shaders running on a server somewhere, and Sofie would activate them at the right moment in the show.
It's a tool that lets you drag and drop news stories in to a rundown and it will automatically play them. The news stories may have parts that are read from a teleprompter, parts that are pre-recorded video, live parts from outside broadcasts, interviews, graphics that need to be shown etc. Those are mostly provided by services or hardware that Sofie controls, not that are part of the show. Sofie is an automation tool.
meteorjs' an interesting choice ...
Meteor got a lot of attention and hype on HN a few years ago.
Looks at Wikipedia article
Well, 12 or 13 years ago probably actually.
According to Google Trends meteor js was popular from about 2012 to 2018. Sofie was started in about 2018, so when Meteor was established enough to have all the kinks worked out, but still popular.
Cool, going to create my ai tv station.