VRT's Video Snackbar

👨🏻‍💻 Blog & updates

🌦 Weather report: NAB 2019

I had the honor and pleasure of attending this year’s NAB show in Las Vegas. In this article I write up some of my impressions and highlights.


NAB is the yearly broadcast and media technology exhibition taking place in April in Las Vegas - the biggest and most important one in the world, with the IBC show in Amsterdam in September as a close second.


Together with our CTO, CFO and a delegation of the team in charge of the new building I wandered around the huge Las Vegas Convention Center from Sunday 04th till Thursday 8th of April visiting booth after booth, soaking up updates, discovering new trends and applications and catching up with colleagues and friends from across the globe (as it’s one of the few places and moments during the year that everybody is on the same square kilometer).

NAB is very extensive. U can find literally everything u need to make media.
This overview is only a small window on what this whole market has to offer. It is also a biased one, as I am looking for compact, mobile, flexible and innovative solutions that can help VRT’s online content creators.
I also made a point of going on a hunt for new players (or at least: new for me) rather than visiting the more established or bigger brands.
I will refer to some brands that I visited during the show, but without a doubt there will always be other vendors that offer similar solutions. Please feel free to let me know what else is out there!


If “cloud solutions” up until now usually meant: let’s run our software solution on a VM-ware machine and put it on AWS, I feel that this NAB some significant steps towards cloud native live production have been taken.

Newtek released a new software only version of it’s top range video switcher and announced a cloud based subscription model to be made available later this year.

At the AWS, Microsoft and Google booths many cloud native brands and solutions were on display, such as the Veset playout system.

With cloud video switchers by Sony and TVU that u can rent by the hour and an MCR in the cloud by Make.TV we meanwhile have cloud native production tools across the whole chain, from live production and contribution to storage, editing, playout and archiving.
Companies like Qvest even offer this complete chain in the cloud, based on established solutions by different brands but “knit together” and enhanced with their own GUI and workflow management system.

The availabilty of AI in the cloud to enrich your productions and the pay as you go model makes going cloud extra attractive.


😳 But “uncompressed”?

Granted, we’re talking about playmobil bitstreams of around 10 mbit/s with these solutions, but the intention and the functionality is there.
With a big pool of new content makers who might favor (or at least balance) affordability, low latency across open internet and interactivity to uncompromised 12 Gbs image quality, these solutions are here to stay and will mature in the coming months.


One thing which comes in handy if you want to go cloud, is getting your signals in and out of the cloud, or move them quickly and reliably inside and between clouds.

That’s where SRT comes in. SRT enables the delivery of high-quality, low-latency video across the public Internet. It was developed and pushed by Haivision, then open-sourced and now supported by the 200+ companies that make up the SRT-alliance.

For me it was one of the most eye-catching and most mentioned technologies this NAB.

🤖 AI & ML

Every product seems to make use of AI these days.
Speech to text algorithms, face and voice recognition, object and context detection, workflow optimization. Everybody is experimenting with it, and it can really enhance your workflows if it’s done well and if the User Interface or API allows you to tweak the algorithms and to leverage these extra metadata.

For example: AI assisted live productions are becoming a real thing.

Automated live sports productions by Pixellot or Mobile Viewpoint, with live cutting and highlights detected and pushed towards your clients seconds after they occur can expand your live sports offering in a lightweight way, ideal for tier 3 sports.
It typically makes use of 4 or more camera’s that create a giant horizontal 16K canvas that films the complete pitch. An AI director follows the ball and zooms in when the action becomes more intense.
It can now also work with additional camera’s around the offside lines (soccer) or behind the baskets (basketball) so the pmg out can consist out of different vantage points. They also offer the possibility to intervene manually.


Automated multicamera live registration of talking heads, based on AI assisted face recognition, body tracking and voice activation which controls the movement, zoom and focus of multiple 6 DoF robotized cameras start creating credible camera cuttings without the intervention of cameramen or directors.
Seervision, Nikon and Mobile Viewpoint were showing this, and the number 1 robot arm that came back on almost every booth that had robotics on display was from “Mark Roberts Motion Control”.



For one of our projects we’re looking for affordable HD transmitters that can send a low latency camera signal one way and intercom, tally and camera control the other over at least a few 100 meters, so we can quickly deploy a multicamera setup, without having the need to pull cables.
This market segment is dominated by the Chinese. I visited Hollyland, Cinegears, Nimbus (Korean) and ST Video. They all offer similar systems to transport low (no) latency high quality HD signals, intercom and control data over distances between 100 meters and 3000 meters, using their own interpretation of 5Ghz WiFi.

A beautiful product from China is the Z cam E2. A 4K 120 fps cinema camera on a shoestring budget, with a very vivid online community that feeds back to the makers of this camera. The company in return releases (free) firmware updates with new features added almost weekly.



A few days before the start of NAB, Vizrt (the very high-end and expensive graphics company) announced it had acquired Newtek (the inventors of NDI and the pro-sumer democratizers of live video production). An unexpected marriage, with alot of synergies, but also a lot of contradictions.

This made me anticipate Newteks already hyped pre-NAB event on Sunday evening even more. Unfortunately, the shoes were a bit too big to fill.
Where last years event took place in a smaller hotel conference room and had something of a cult gathering of NDI enthusiasts, this year the “NAB main stage” (800 seats) was too big, the debates too unprepared and the keynote by Dr. Andrew Cross not compelling enough. You could feel that both mr Cross (Newtek’s President and CTO) as mr Petter Ole Jakobsen (Founder and former CTO of Vizrt) were twisting their tongues or that they themselves did not know yet what this acquisition will lead to.


The future will show what direction the new combined company will take, but it shows how vulnerable a proprietary protocol is: if “VizrTek” decides to take down NDI, or to “put it behind a paywall”, that could be most inconvenient. Of course the minute this would happen, somebody will develop an open source version of NDI (suggestion: NDX, or ONDI?).

On the other hand, they could just keep things the way they are -which succesfully combines a high degree of openness (with a free license and SDK) with the speed of development of proprietary- or even “open source” it at one point.

I’m also convinced that if they would release a “pro version” of NDI with some extra features required in the broadcast facility, broadcasters would gladly pay up for this.

On the bright side, they announced NDI 4.0 with lots of new features (16 bit, direct to to disk recording of the NDI streams, synchronization and more) and the ecosystem of companies that offer NDI connectivity or that create new tools built on NDI keeps on growing. Sony being one of the most recent members to join the NDI train.


While the joint effort of VSF, SMPTE, EBU, AMWA and AIMS to push an agreed upon set of open standards and specifications remains in some ways exemplary, it feels as if we are still not there.
“There” would be: a future-oriented, reasonably complete, easy and affordable way to implement facility wide, software ready and cloud-able high-end video over IP systems that go beyond “let’s connect FPGA boxes by fiber cable and an SDN”.
A nice illustration to this is EBU’s Video over IP Pyramid, showing where we currently are in the 2110 adventure.

As I’m a consumer with a limited budget, I was relieved that companies like AJA offer USB-C video capture cards that can take in and play out 2110 video so i can hook up my laptop to the video network.

Embrionix (and their booth) keeps on growing show after show.
They make 2110 gateways to and from sdi and hdmi in the form factor of an SFP that can go straight into your network switch. This offers a very granular road towards “going IP” as u can buy gateways per stream in stead of 8 in one go. New this show was an SFP without inputs or outputs on the front of the SFP, but with FPGA power inside that you can address from within the network. An efficient way to plug those empty holes in your network switch!

The same can be said about Aperi, four years ago one of the first companies to sell a 1U box rich with FPGA power inside, on which you can spin up and down broadcast and telco apps in a matter of seconds.
The front and backplane of the box are connected to your network via 2110 signals (or 2022-6, TS, J2K, …).
They are flourishing in the Telco space, maybe one day they will re-enter the broadcast market.

I didn’t visit much of the bigger brands that push 2110, but i did visit the IP Showcase.
A professor looking type was explaining away about the many subsets of standards that belong to the 2110 family to a crowd of 30 middle aged engineers.
Across the hallway, about 100 20- to 40-somethings where listening to a guy showing how to colorgrade on an IPad, in the Mobile Video Editing pavilion.
Not that one should be valued as better then the other, but it illustrated a certain feeling perfectly: the market for dedicated hardware to produce good old high end tv entertainment is under pressure, while new markets of mobile video, esports & gaming have opened up. That means new (software) opportunities but also new expectations.
By staying stationary too long, the 2110 effort risks to become something very niche-y, in favor of something that better catches the sense of time.

That being said, we do support the open standards approach and we will continue running our own projects making use of the 2110 technologies. We’ll see where it leads.


To round up this article I would like to give a shout out to the Belgian companies that made the trip to NAB.

VRT Sandbox alumni Tinkerlist, Azilpix, Limecraft, Cheqroom, Cyanview, On-Hertz (first time, congrats!), Theoplayer, EVS, SDNsquare, but also Skyline, Mediagenix, Intopix and some that i’m forgetting now hopefully came home with alot of new leads and opportunities.


Thanks for reading, please contact me if you have suggestions, ideas, remarks.

All the best,

👨🏻‍💻 Karel De Bondt

Karel De Bondt