Video and audio production using IP networking is on the threshold of widespread adoption and may soon become the default way of working for live production. But nobody expects a total switchover to IP production anytime soon, even though, on the face of it, that would seem to be the ideal outcome. Practical considerations such as backwards compatibility and failover to more traditional methods are likely to require hybrid solutions for some time to come.
So, the transition to IP in live production is likely to be a blend of cutting-edge IP architectures, increasing virtualization, and retaining the ability to work with “traditional” media transport and compatible devices. If and when a total transition finally comes, it will be the result of a measured approach, where reliability and robustness are critical attributes.
In the early days of IP media transport, while standards were still being devised, IP video and audio was typically a straightforward SDI substitution. In other words, IP networking was used to replace sections of point-to-point coax cabling. It could be thought of as a virtual SDI cable because SDI went in at one end and came out of the other. This was a valuable use case because video production facilities are always likely to have a hybrid element, and it’s important to be able to move between the SDI and IP domains easily.
At the risk of stating the obvious, networks are different from SDI cables. Some of these differences are what give IP its superpowers, and it’s very easy to gloss over them, missing some quite remarkable capabilities.
Network Cables Are Not Unidirectional
SDI cables don’t have a polarity: it doesn’t matter which way round you install them. But once a signal flow is established, it is exactly that: a flow in a single direction. Outputs are outputs, and inputs are inputs. It’s conceptually simple, but it’s also limiting and potentially wasteful because it leads to much more cable than might be needed if the signal could flow both ways. We will explore this in much more detail later.
You Can Put Multiple Signals Down A Single Cable And Access Them Individually At The Other End
While it is possible to put multiple digital signals down a single cable (as in the case of MADI for audio), this normally involves Time Division Multiplexing and requires relatively complex de-embedding to extract individual signals.
The Concept Of Signal Flow Is Virtual, Not Physical
On a network, to oversimplify but also to explain, everything is “visible” to everything else. It’s as if data is put on a noticeboard in an envelope with the recipient’s name on it. Any authorised person who can see that noticeboard is able to peek inside the envelope. Network switches, routers and IP management software segment the topology, but it is broadly accurate to say that everything on the network is available to everything else.
In a facility, any network port becomes an input or output for any signal on the network. Need to connect a camera to a switcher? Plug it into the wall. Need to view a camera channel on a monitor? Plug it into the wall: any socket on any wall. Need to move an entire production into another studio? No problem. Just take your stuff, plug it into the sockets on the wall and re-route. And that’s just within a facility. But once media is in the IP domain, it can be routed worldwide. This opens up brand-new topologies and production paradigms. We will look at this later in “Virtualization”.
Key Advantages Of IP Production
You Don’t Need A Physical Cable For Each Signal
This, perhaps, is the ultimate advantage of networks. Physical network cables (and the cloud) provide a virtual route for video and audio that is easily configured and re-configured. Crucially, you can fit as many IP video and audio signals down a single physical cable as bandwidth will allow. This jump from the physical to the virtual results in orders of magnitude more flexibility. A
We’ve already hinted at this with our “noticeboard” analogy. With networks, the signal flow is determined not by the physical topology of the cables but by software. At a low level (e.g. the Ethernet level), on a simple network, all signals are present at all points, at all times. On IP networks, data packets find their own way to destinations thanks to routers along the way. But the effect is the same: with software management, any point on any network can be the start point or end point (think of them as inputs and outputs) of any video or audio signal. This immense flexibility means that it’s possible to design and build a live production facility well in advance of the final configuration. Once the IP infrastructure is in place, within reason, it’s possible to plug anything in anywhere.
A quick note about media essences in the IP domain: We are now no longer tied to combined video and audio signals needing specialist embedding and de-embedding equipment for audio processing tasks, as IP allows transport of multiple video and audio essences separately, and to or from any endpoint.
Networks are incredibly flexible when it comes to routing. Within reason and bandwidth permitting, it is possible to route anything anywhere. In practice, software routing is inevitably more complex than this, but that complexity can be hidden from users, who interact with it via intelligent control systems, some of which could be familiar-looking router control panels that instruct the software.
Traditionally, live productions have been based around a centralized architecture. That makes sense on a local scale. But with IP video, centralization is not a necessary default, and moving away from centralization opens up a multitude of new topologies that can enable new types of production, both in content and scale. This has huge implications for the future of live production, where scalability is built into every IP media installation.
Networks take the “physical” out of managing and transporting audio and video. You still need physical network wiring, but what passes over those cables can be determined by software and is, therefore, far more flexible. One important consequence of using networks to transport media in real-time is that inputs and outputs are completely virtualized: there is no longer a one-to-one relationship between signals and I/O ports. Given enough bandwidth, any number of “channels” can enter and exit through a single network physical port. There’s no longer any need for the one-cable-per-channel, which means fewer physical cables and greater routing flexibility.
Virtualization In Facilities
Whole live production workflows can be virtualized. As long as there are enough network ports and the necessary supporting network infrastructure, the physical parts of a broadcast facility can be added, subtracted, and moved at will. Any network port can offer a range of services: inputs, outputs, for cameras, mixers, audio and video monitors, and so on.
Remote Live Production
Because IP media systems are intrinsically non-centralized and distributed, they lend themselves perfectly to remote production. This can take many forms, but one of the most effective is where live events can be remotely controlled from a central facility. Local teams set up and tend cameras at a venue, but the production team and their associated equipment can be at a remote facility. It reduces the need for travel and gets the maximum return on production equipment that can be in use twenty-four hours per day instead of the approximately two hours length of a sports event, with many “dead” hours transferring the equipment from one venue to another. Reducing travel also helps with staff efficiency and in retaining skilled production staff who might rather work centrally than travel to multiple venues.
Standards are things that aren’t supposed to change, but ironically, their existence as a stable foundation enables change. That’s because if you can assume the basics, it makes it much easier to innovate at a higher level. This is especially so with IP networks, where interoperability means so much.
ST 2110 is largely thought of as enabling uncompressed IP media, although the standard is rich and complex and does allow for compressed video and audio too. It had a relatively long gestation, first proposed in 2012 and ratified in 2017, and will likely continue to evolve while assuring backward compatibility. It can essentially be thought of as a media essence transport with network superpowers. The standard also includes provisions for synchronization and security, essential for robust IP media transport.
The adoption of IP embraces not only open standards but also some proprietary standards such as NDI. NDI has a different approach. Its goal is to use - as far as possible - traditional low bandwidth 1GigE networks and to achieve this through compression leading to a low bandwidth overhead. It comes in several flavors, one offering high quality and moderate compression with very low latency, almost analogous to (but specifically not) Apple ProRes compression - easy to encode and decode. Another variant uses long GOP compression where latency is less important than very low bandwidth requirements.
Similarly, in audio, both proprietary and open standards are deployed, with the most popular being Dante and AES 67, both audio network protocols that are widely used. Both can handle very large numbers of full-resolution audio channels. These standards are easily deployed as redundant networks allowing full protection of the media assets, this is a popular way of configuring modern production networks.
IP production is rapidly evolving, but it is now a relatively mature, reliable, and versatile method to transport media locally and over larger distances. It breaks the one-to-one link between video and audio channels and cables. It brings the multiple advantages of network topology to live production. It is intrinsically scalable, from a local facility to a distributed, global network of facilities. Software control, routing and configuration provide incredible flexibility within a facility and for co-producing live broadcasts with remote facilities.
With standards like ST 2110, NDI and Dante, it is increasingly feasible to interconnect video and audio equipment - and entire facilities - with networks. Interoperability is only likely to increase. And with standards in place, existing IP media facilities can grow alongside network and computing technology as they inevitably become more advanced. New facilities become more cost-effective as they require less specialist equipment and use high-performance but off-the-shelf IT products like network switches and routers, which benefit from economies of scale to lower costs still further. Change was already endemic in the technology-dependent media production world. Now, with the rapid rise in the requirement for flexible production capability, the future is becoming less predictable. It is anybody’s guess what the media production landscape will look like in ten years' time. But video and audio transport over IP is so intrinsically flexible that it is likely to be the only dependable way forward. The future will call for previously unseen degrees of flexibility, and IP media has arrived on the scene just in time to provide this.