Taking IP to the Wire: NewTek's Boss Talks Tough on Transport
Standardizing an industry move to video-over-IP is generating considerable confusion, but NewTek claims its approach is the only one focused on IP's real-world capabilities.
Much of the debate surrounding the move from video-over-SDI to IP has been framed around 4K/UHD—that for 4K to work economically, IP infrastructure is a prerequisite. NewTek has a different take: It wants to move the industry from what it sees as the relatively narrow focus on 4K and other technical specs toward applications which genuinely take advantage of internet connectivity.
“Contrary to popular belief, IP is not a cheap way of communicating digital data,” says NewTek president and CTO Dr. Andrew Cross (right). “The major benefit of IP is that it can connect any device with every other in the world. You can pipe it around, you can write applications and do things that were never thought of before. Crucially,the link is bi-directional. Where SDI signals flow downstream from a camera to, say, a monitor, with a true bidirectional IP link you have intelligence about the device it is connecting to. This is a rarely acknowledged but fundamental part of the change the industry is going through.”
The concept has been familiarized at the BBC as object-based broadcasting, and Cross hints at working with the broadcaster on tests in this area. The idea is that any element of the signal—video, graphics, audio—can be split into its constituent elements and reconfigured at the receiver according to individual preference and environmental context.
“4K or HDR, HFR, 8K or virtual Reality streaming... these are just technical specs, compelling and important certainly, but not in themselves anywhere near as revolutionary as the pipes they run on,” Cross says. “IP changes everything for the industry, and we think the focus should be on this bigger picture.”
There are a number of competing protocols for packaging video and audio over IP networks of which NewTek's is one. Most are vendor-specific, yet have the backing of multiple parties. Some vendors support more than one candidate. All claim to be standards-based and all tout their open credentials. But no scheme has has been fully ratified by any international standards body and the degree of openness is frankly "open" to interpretation.
Last September, NewTek launched NDI (Network Device Interface). While all the protocols are attempting to tackle video-over-IP in live production, Cross believes that NewTek's approach differs markedly.
“We are not against any of the other [protocols],” he says. “We work closely with them and we've reached out to them and we're working on collaboration to the degree that we can, but we do believe our approach and mindset is different.”
The Protocol Soup
Before exploring Cross's argument, it's worth recapping the other protocols for video-over-IP.
Most take the unidirectional SMPTE standard 2022-6 as their starting point. They include Sony's Networked Media Interface (NMI) and The TICO Alliance, led by IntoPix, a chief difference being the promotion of their own licensed codecs at the heart of an IP live production system.
AIMS (Alliance for IP Media Solutions) is an industry grouping of which Grass Valley, Imagine Communications, Lawo, SAM, and Nevion are founder members, lobbying to use the AES67 and Video Services Forum-devised TR03 which splits the video, audio, and metadata into separate paths.
A different tack comes from Evertz, which promotes ASPEN (Adaptive Sample Picture Encapsulation), a protocol that works with media over MPEG transport streams.
“These are all being created by companies at the higher end of broadcast using more traditional equipment,” Cross contends. “What they are looking to do is take workflows they have today and make them work with IP. They look at IP and see it as SDI, as a more traditional video transport, and within certain markets that makes sense. Yet, this form of video-over-IP requires strong timing requirements that have to align packets on a network with raster scans of an image extremely accurately.
“If you believe, as we do, that the world is headed toward the use of general purpose computer systems, then you must realize that you can't achieve nanosecond or millisecond accuracy for timing on the network. For instance, in TR03, you need ten microsecond timing accuracy for each scan line of video. There's just no way that a computer system without customized hardware is ever going to be able to achieve that.
“One of the promises of IP technology is that we can do away with hardware, like capture cards, yet with all these other video-over-IP concepts you will end up hanging a lot of these off specialized hardware off the network.”
The approach is valid, Cross says, but only for the big beasts of broadcasting. In contrast, NewTek's target is the up-and-coming video generation of niche live sports streamers, corporate communicators, and start-up video innovators.
“If you look at almost any other industry it's going to be the small companies with cool ideas; they're going to revolutionize this industry,” Cross says. “We want to enable the cool ideas by removing any license fee and expensive infrastructure which will inhibit the industry and stifle innovation.”
An Interoperable Plea
NewTek wants to make its technology work alongside others in the industry so it can implement an NDI interface with SMPTE 2022-6. It is offered as a royalty-free SDK. NewTek is part of AMWA, and Cross says the company talks regularly with members of AIMS and SMPTE. Nonetheless, the company has not gone so far as to back another initiative.
“We are fully prepared to do our part to enable interoperability,” Cross insists. “We are the first [of the rival standards] to state that we will put efforts into interoperability. Neither AIMS or Evertz have said that.
“Our goal is not to build an insulated garden to keep others out. In fact, quite the reverse,” he adds. “We're making it very easy for people to build products that work over IP. In many ways, I don't see what we're doing is counter in any way to the existing standards.”
There are two ways to become an industry standard—either work through a standards body or work to achieve mass adoption. In the absence of standards—which are grinding their way too slowly for many through bodies like SMPTE—the competing protocols are trying to amass as many backers as they can.
NewTek can call on PlayBox Technology, Vizrt, Teradek, Wasp3D, AJA, Deltacast, ChyronHego, and Matrox, with solutions from JVC, LiveU, Panasonic, Sienna, and Wowza forthcoming. Several of these—AJA, ChryonHego, and Matrox among them—support other initiatives, as well.
More than One Standard
“IP is already standardized on transport,” Cross says. “This is, in fact, allowing a lot of different companies to innovate protocols on top of it, and we will see a small number of them become the de facto standard.”
Cross argues that the history of the internet has shown that that more than one standard is possible. In data storage, for example, there is JSON and XML. Another example is TCP/IP and UDP.
“You do need international standards, but for that to work you need mass adoption,” Cross contends. “Thousands of standards have been taken to bodies like SMPTE; some survive the evolutionary process, most wither away. We are focused on making a technology for users that solves a real world problem.”
Cross continues: “In the past, the industry has been burned because if a facility made a decision about a standard and bet on the wrong one you would end up with huge sunk cost with no way out. To a degree IP has overcome that since IP cabling or network doesn't need to change. So I don't believe [the multiplicity or lack of standards] will be as much of an issue this time around.”
The Achilles Heel: Compression
A key issue which puts the various standards at odds is their view on compression. Both AIMS and ASPEN are notionally uncompressed while IntoPix and Sony NMI promote their own codec, especially for carrying 4K. Pretty much everyone acknowledges that a form of mezzanine compression is needed to put 4K (and certainly UHD HDR) over a 10GigE pipe with different schemes, including the royalty-free VC2, being preferred.
NewTek NDI is compressed at what Cross calls baseband quality, similar to ProRes or DNXHD. “Most everyone will come to the same conclusion,” Cross says. “If you want to transport uncompressed around then SDI is going to be better than IP.”
The company describes its encoding algorithm as resolution and frame-rate independent, supporting 4K (and beyond) along with 16 channels (and more) of floating-point audio.
“When corporate enterprises want to broadcast to their employees you can't assume they are going to be ripping out their wiring. You have to assume it will be based on 1 GigE, so you need a system to work with that. This gives you huge gains; for example, the ability to share multiple streams of video over a GigE LAN.”
This may not be sufficient for premium pay TV sportscasters like ESPN or Sky as they look to revamp facilities for UHD, but it's compelling for a large portion of the future video industry.
NewTek can back its arguments up by pointing to over a decade of its technology shipped with video-over-IP capability.
“In that time we've learned and improved what we've done. For years, we’ve had the ability to take video from a Vizrt virtual set system, for example, and send it over IP into a NewTek TriCaster. We have probably about 100,000-plus systems in the market today that can already talk these protocols.”
TriCaster operators will be able to access Mushroom Networks Streamer control and feeback via the TriCaster user interface
Broadcast is moving to an IP-first, software-defined video infrastructure, and in the process reducing costs and creating new editorial opportunities.