The future of SMPTE 2110 and beyond
A year ago the broadcast equipment manufacturing industry was at loggerheads in pursuit of incompatible paths to the common goal of transitioning video from SDI over IP. As we head toward NAB, it’s fair to say that there’s been a remarkable sea change. By and large the industry has aligned behind a set of standards which should deliver unprecedented interoperability and therefore accelerate adoption of video over IP.
“There’s no doubt anymore that AIMS – or perhaps it’s better to say SMPTE 2110 – has won the race,” says Felix Krückels, Director of Business Development, Lawo. “Open standards are a must in our niche industry. I would say that NMI stays as long as we have a UHD codec agreed on and ASPEN is more or less dead already today.”
“The industry is coalescing because you need a rock-solid foundation for how next-generation facilities and studios will be built,” says Grass Valley CTO, Production, Chuck Meyer. “A key to that is interoperability since this is the only way the industry can develop and grow at scale.”
Particular attention is trained on SMPTE 2110 which may optimistically be ratified as a standard this summer, though some think it more likely in 2018.
“My feeling is that there’s enough vested interest now that it will happen sooner rather than later,” says Tim Felstead, head of product marketing, SAM. “The longer it drags on the worse it is for everybody.”
His only fear is that one company may scupper it within SMPTE much in the way Russia might use its veto at the UN.
Trade group AIMS has the backing of 70 members, including Evertz and Sony, which are both sidelining their own video over IP variants to back ST 2110.
Evertz developed ASPEN for customers needing to move fast into an IP world than the slower cogs of standards bodies were dictating. Now, not only a member of AIMS but an active participant in the development of 2110, Evertz says ASPEN will morph into 2110 and, over time, its reason to exist will cease.
“I don’t see ASPEN dying away immediately,” says Mo Goyal, Evertz’ director or product marketing. “There still needs to be transition point for those facilities running ASPEN, which has been proven to work for facilities at scale. There is a requirement to have an upgrade path to 2110 and our position on this is via firmware on the device.”
For Sony, Nicolas Moreau, product marketing manager IP live production & workflows, says: “The future of live IP production will be based on ST 2110 plus NMOS and the legacy of our solution.”
These legacy elements of NMI include encryption and the Sony codec LLVC. It will be showing its developments on the AMWA and AIMS stand at NAB.
What is in 2110
Like 2022-6, ST 2110 defines a transport and timing protocol for A/V and metadata, but unlike 2022-6 the key concept is to split the signals into independent essences. This approach, encapsulated as TR-03 devised by VSF, is better suited to a production environment than a composite one as, for example, it makes audio processing much easier since no de-embedding and re-embedding is required.
TR-03 itself is composed of a number of existing standards including AES67 for uncompressed audio, RFC 4175 which defines video and SMPTE 2059 for clock synchronisation.
“ST-2022 is fine for simple OB and studio set up which are self-contained but if you look into more complex broadcast centres, transmission and playout then at that point you are needing to process video and audio separately, as well as handling subtitling information,” says Gearhouse Broadcast, systems integration manager Martin Paskin. “You don’t need all that to be carried in a payload. Moving to active essence-only content we end up with better use of data.”
The compression question
ST 2110 will define uncompressed transport of 3G HD-SDI over IP. The question is at what point compression is necessary. Since last year the industry has witnessed the introduction of 25G interfaces with 40G, 50G and 100G interfaces coming.
“Suddenly the need for compression is less,” says Felstead, “though not gone entirely.”
AIMS adheres to the idea that compression at the heart of production should be minimised if not kept out altogether.
“Any compression causes latency and artefacts when ideally you want picture and audio as clean as possible,” says Felstead.
However, in contribution links or remote production where bandwidth is too costly, then having the option of compression is considered useful.
“Even if it is possible to do 4K uncompressed today you have to consider the cost [of bandwidth] especially when you have a perfectly working visually lossless codec,” says Moreau. “It’s true that 4K uncompressed will save the hassle of having to choose a codec but you have to consider the reality. We may soon have 200GB switches but you have to weigh the cost of using that. Even a 40G interface may be too expensive today.”
One candidate, VC2, is already a published standard in SMPTE. Sony has submitted its LLVC codec, Fraunhofer HHI offers an ultra-low latency video encoder compliant with the H.264 baseline profile. Others back the TICO scheme developed by IntoPix. AIMS does not express a preference although individual members will (Grass Valley is a TICO adherent, for example, while SAM and Lawo favour VC2).
“We’re not far away from 4K 120p which requires 24GB and even 8K, for which we would probably want to use a mezzanine codec,” says Meyer. “The prevailing idea is to use a lightweight codec of 6:1, 4:1, 3:1, for example, as a way for customers to scale and future proof workflows as video formats and framerates evolve.”
A clue can be found at the JPEG committee (a joint working group of the International Standardization Organization (ISO) and the International Electrotechnical Commission (IEC). It has formally agreed to develop a low-latency lightweight image coding system known as JPEG XS for which the baseline is TICO. JPEG XS is being designed to support increasing resolution (such as 8K) and frame rate in a cost-effective manner.
Lawo says it prefers one type of codec, but expects to see variations of it: UHD in low latency mode, all formats in high compression mode for contribution and remote production, plus very high compression in low latency for monitoring.
Registration and discovery
To scale systems it is necessary to have the ability to plug in a device and make it known to the IP network, and then have a common way for that device to describe all of the things it is capable of doing.
The work of AMWA project Networked Media Open Specifications (NMOS) protocol is significant here. Inside that is registration and discovery mechanism IS-04 which, de facto, all AIMS members agree to support and incorporate in their product roadmap. It will not, though, be part of SMPTE 2110.
“It will become an essential part of an IP system, but it’s one of those strongly recommended practices rather than necessary as a standard,” says Meyer.
“IS-04 is an essential part of the IP system,” affirms Felstead. “You don’t want to plug all these devices into a router network or a network of routers and have to type in a bunch of IP addresses in any sort of manual sense to get them registered on your control system.
“What shouldn’t be standardised is the control system for devices or networks,” he argues. “When you get down to device level control it’s a value add on the part of the manufacturer. Its sophistication and how well it works are points of differentiation between suppliers. If we try to put a common control mechanism into a standard like SMPTE 2110 we run the risk of sanitising everything and it smothers innovation.”
AMWA has begun exploring how NMOS will work in practice. Ultimately, this will lead to new specifications which will allow the industry to truly embrace data centre and cloud technologies and feel confident relying on another company’s platform, hardware and servers.
“The industry needs to look at bringing 4K into the conversation and how we carry and leverage higher bandwidth interfaces. How does 2110 map to 25G; [and] how do we move forward to incorporate 4K, 8K,” says Goyal.
Sony notes that some aspect of live – such as the multiviewer – are on the verge of virtualization while vision mixers will take more time purely because of the amount of data they must handle.
However, obtaining true virtualisation and remote production is arguably more of a battle to find bandwidth and processing power than it will be about standards.
“2110 is not a great determining step toward virtualisation and the cloud for production, but it is a great determining step for taking SDI and putting it over IP,” says Felstead.
A quick fix?
While Grass Valley will be among vendors announcing 2110 compatible product at NAB, many manufacturers are waiting for the standards to settle before they put in an IP core.
“While we can build a full 4K IP studio today it will take up to five years for systems integrators to be able to build one with the same ease and effectiveness as they can with SDI,” says Paskin.
The move to IP had been grinding too slowly for an industry pent up with demand, such that last year a number of US-based OB firms, including NEP Group and Game Creek Video, called on manufacturers to deliver an 12G SDI kit as a viable alternative to 4x3G for transporting 4K.
“This was [conceived] as a temporary solution but it’s now actually coming to the fore,” says Paskin. “People are used to SDI and are able to work with it quickly.”
FOR-A is among the vendors to have brought out 12G equipment. Elsewhere, IntoPix has worked with Japanese developer Village Island to launch the VICO-4 encoder, which takes in quad 3G-SDI and converts it into a single 3G-SDI to reduce cabling.