Voices of Video
Explore the inner workings of video technology with Voices of Video: Inside the Tech. This podcast gathers industry experts and innovators to examine every facet of video technology, from decoding and encoding processes to the latest advancements in hardware versus software processing and codecs. Alongside these technical insights, we dive into practical techniques, emerging trends, and industry-shaping facts that define the future of video.
Ideal for engineers, developers, and tech enthusiasts, each episode offers hands-on advice and the in-depth knowledge you need to excel in today’s fast-evolving video landscape. Join us to master the tools, technologies, and trends driving the future of digital video.
Voices of Video
VLC to VPU: Evolution of Video Processing retold by Easy Tools
From the dorm rooms of a French engineering school to the control rooms of major broadcasters, Christophe Massiot's journey represents the evolution of video technology itself. In this illuminating conversation, the co-founder of VLC Media Player and creator of EasyTools reveals how a student project designed to upgrade a Token Ring network became one of the world's most downloaded applications.
Massiot takes us behind the scenes of his transition from open-source pioneer to broadcast technology innovator, sharing the moment of realization that led to EasyTools' creation: discovering that €30,000 commercial encoders were performing tasks achievable with properly configured open-source tools on hardware costing under €3,000. This revelation sparked a mission to make broadcast-quality streaming affordable and accessible through user-friendly interfaces built atop powerful open-source foundations.
What makes EasyTools unique in today's market is its modular approach - one product that can function as SRT gateway, encoder, transcoder, multiviewer, or recorder through combinable modules that eliminate the need for multiple separate appliances. Now deployed at nearly all French broadcasters and expanding internationally, EasyTools represents a paradigm shift in broadcast economics.
The conversation pivots to a game-changing integration with NETINT's Video Processing Units (VPUs) that's transforming the efficiency of professional encoding. When faced with a project requiring 100 channels with multiple renditions, traditional CPU encoding would demand ten expensive 64-core servers. With NETINT's technology, the same workload runs on a single server equipped with multiple NVMe-format VPU modules - dramatically reducing costs, power consumption, and complexity.
Don't miss the opportunity to see Easy Tools and this groundbreaking technology in action at NAB booth W3531, where Christophe and the NETINT team will demonstrate how the marriage of flexible software frameworks with powerful hardware acceleration is opening new possibilities for broadcasters worldwide.
Stay tuned for more in-depth insights on video technology, trends, and practical applications. Subscribe to Voices of Video: Inside the Tech for exclusive, hands-on knowledge from the experts. For more resources, visit Voices of Video.
voices of video. Voices of video.
Speaker 1:The voices of video voices of video well, we are back with yet another special edition of voices of video. So this is our buildup to NAB. If you've been watching all of the episodes we've been dropping recently, you'll see that there's been a flurry of partners that we are talking to and we are just really excited for this year's NAB. We are bringing together the ecosystem of VPU. There are many, many, many deployments now of NetEnt VPUs in the market. Nab, you might say, is a little bit of a coming out party. As an ecosystem, we're coming together to showcase some really amazing use cases, location and deployments of VPUs. So with me today is Christophe Massieu, who is the founder of EasyTools, and first of all, christophe, I want to welcome you to Voices of Video. Thank you for joining us.
Speaker 2:Thank you, Mark. It's a pleasure to be here.
Speaker 1:Yeah, well, it's really great to have you, and of course, we're going to get in and talk some more detail about what you are doing at Easy Tools with BPUs, how you're deploying them and the value they bring. But I think not all of our listeners probably know who you are, and so we should start there, because you've done some very interesting things in the industry. Why don't you introduce yourself?
Speaker 2:I am some kind of a veteran in the industry because in the days when I was still a student, actually in my engineering school in France, I co-founded a little software called VLC VLC Media Player. You may have heard of it I think I've heard of that VLC.
Speaker 1:Let's see.
Speaker 2:Yeah, so I was one of the first students. It was a student project originally, so I was one of the first students to write the code of that. I stayed on the project for quite some time afterwards. I'm not so active lately, but still it's a pretty good reference in the industry, and I'm also still active in many open source projects that are not as well known as VLC, of course, but still.
Speaker 1:Amazing. Wow. So you know. Maybe you can give a little insight about the early days of VLC and how it even came to exist. What was the genesis, you know, to that project?
Speaker 2:Well, there are two versions. The regular version is that it's a student project. You know, when you're a student in an engineering school, you have to build something. So some people build a car, some people build a rocket, and we decided to build software to do video over a network. So that was the origin of VLC. Vlc means Video Land Client, because there was also VLS Video Land Server at the time and there was a server and a client and we were doing video over a network. The real reason was that we needed an upgrade for our network. That was at the time token ring for people who are old enough to remember that.
Speaker 1:So I am old enough to remember that yeah, I know yeah.
Speaker 2:Not so many people remember that, but yeah, it was old technology and we wanted to upgrade to Ethernet and one way to get subsidies was to create a project that would involve video over network. So it would require upgrading the network because it wouldn't work on Token Ring. So that's one of the motivations of the students, but still it has led to a great project that is still in use billions of downloads today, so it's quite a big success actually.
Speaker 1:It's amazing. Yeah, yeah, that's incredible. Well, so VLC, and then you build a company. Now did you start building EasyTools shortly thereafter? Did you do some other things in the interim? Tell us the story of EasyTools?
Speaker 2:No, I've had some real jobs before Real jobs. I've worked for some time as a major broadcaster in France called Canal+, which is very well known.
Speaker 2:And after that I spent 10 years at a telcos and I actually created, designed, installed, operated the entire television head end of that telco. So the telco was a little bit special because it was a challenger, not a major telco some telcos that just started and so we had a very low budget and, in order to provide the same quality of service as other telcos, I created my own head end actually. So I used VLC, of course, of course, and other open source projects that I founded or co-founded during that time and other open source projects that are also very well known, such as FFmpeg, x264, that kind of projects. And in the end I spent 10 years at this telco and we had a very efficient and very affordable head end, gathering channels from all around the world through satellites and other means, transcoding, delivering to clients and so on. So it's been a wonderful experience and during that I also had the pleasure to meet most of the CTOs of the French channels, french webcasters. I knew all of them because I was the point of contact for them, and one day I became friendly with most of them and one day one of them asked me the encoder you just installed for me SGSDI encoder.
Speaker 2:At the time it was around 2010 or so. He asked me how much does it cost? Well, I thought about it for a little while and the PC cost maybe 1,500 euros at the time, maybe 2,000. There was a Blackmagic card that cost maybe 500 euros. Let's say so, yeah, we had a complete setup for under 3,000 euros probably. And they said because the encoders I'm buying, I pay them 30,000 euros at the time.
Speaker 2:It's true, but in addition to the hardware I installed, there is software and software that I developed Some of it is open source, some of it is not and the work of many people, and it was very complex, very difficult to set up. So I had command lines like that, very big command lines, to just start a process of encoding, and it was not possible for a regular CTO of a broadcaster to do that, but still I got the idea that maybe there is something to do that to make broadcast and make streaming affordable for broadcasters and for telcos, and so that's when I started the first company that later became EasyTools. In the objective, we use an open source basis, but we try to make it affordable and usable for regular people, people who are not developers like me I'm originally a developer but for people who are not developers, they can set up very complex workflows and do broadcast quality things.
Speaker 1:Wow, that's a great origin story and it's unique because so you mentioned the year 2010. I think in 2010, nobody was thinking about bringing like FFmpeg and X264 into a broadcast production environment. Now that's, you know, not so uncommon. And you know, of course, you're out there with the product now many, many years, and there's others who are building their own. But 2010, the conventional wisdom was you needed dedicated encoders, black boxes.
Speaker 2:That was the only way to do to build a head end right, absolutely, and also one of the very early decisions that we made was to base everything on IP, autocast, ip Also in 2010,. That was new. Many people still relied on SDI, dvb-asi and so on. Yes, the company started formally in 2011. And, yeah, we've had very quickly many French broadcasters coming to us. So now I think our software is installed, almost all French broadcasters have it, and we're slowly expanding to other countries in Europe and also Latin America.
Speaker 1:Yeah, amazing, amazing. Well, let's talk about the product, where the product is today. So it's a tremendous origin story. Obviously, the cost advantage is very easy to get excited about, but you still have to meet all the performance requirements you still, you know, functionally it has to do everything that kind of a traditional head-end would do. So tell us the story of how did you overcome some of the challenges of converting, you know, these open source projects which definitely did not have all the features needed by a broadcaster. How did you approach?
Speaker 2:that we have a very modular approach. That's really the foundation of what we do. We have a unique product. That's very rare for a company to just have one product, not two, not three, just one. We sell one product, but it's a collection of modules that you can combine together to do things in very various areas. For instance, we can do SRT gateways, encoders, transcoders, multiviewers, recorders All of these we've sold a lot of them and you can combine all those features. You can have SRT coming to your gateway and recording it, for instance. It is something that is very unique in the industry because you would normally need several appliances to do the same thing One to do SRT, one to do the recording, one to do the material and so on. So that's very unique. And the way we do it is by using a very modular approach, even at the development level. And we created our own open source project called UPipe, ipe. It's open source, upipeorg. You can go check it out. It's also used by other companies in the industry and this uPipe.
Speaker 2:Of course we do not reinvent FFmpeg, we do not reinvent X2C4. They're products that exist, they're working well. But what the industry was missing mostly is the transport stream part. Transport stream is the foundation of the industry we are in, because the dcts is transport stream, satellite is transport stream. Even when you do srt contribution and so on, it's transport stream. So transport team is the basis of what we do and transport stream has very strong requirements, especially on constant bit rate and that kind of things and the way you have to multiplex all the elementary stream streams together. It's a very complex way and very difficult way and all these projects FFmpeg, vlc and so on they do not do it compliant enough. They're compliant to some extent, but not fully compliant, and so it may work in some circumstances, but generally it doesn't. Let's say so. We wrote it from scratch With this open source project, u-pipe.
Speaker 2:It's a project that is designed to work with transport stream. For instance, one of the characteristics of transport stream is that you can add or remove a new component at any time. You can have a new audio coming at any time. You just have a new PMT, new audio, new subtitles and so on. All of the workflow management tools like SFMPEG or GStreamer. They don't allow that. It doesn't work. So U-Pipe has been designed to work with that and to make it possible. So that's the foundation of what we do.
Speaker 1:Interesting, interesting. Now you mentioned that you have deployments in, I think you said nearly all, or even all, of the French broadcasters. So is this the primary distribution path to set-top boxes, or is this like sort of the TV everywhere, the streaming side of the business? Or maybe you can explain what part of their business is that you are powering?
Speaker 2:It depends. We can deal with many parts of the businesses. Usually we are a small company. Let's face it. We're a small French company, not as well known as many other major brands. So generally when the CTO comes and sees us, he doesn't know if we will be there in 10 years or how strong we are. It's normal to ask those questions. Generally he just tests us on the features that are not essential to his business. For instance, we sell a lot of multiviewers. Multiviewers well, if it fails it's bad because you don't have any visibility on what's happening, but still your channel is still working. Multiviewers, recorders that's the kind of things we use generally to get into new clients. So we have many multiviewers, in Control Room, for instance, that are made by us, because we have a unique position in multiviewers where we are lower end, let's say lower cost, than many other multiviewers but we do less features. But generally some broadcasters, they don't need all these features.
Speaker 1:They don't need all the features of the more expensive product.
Speaker 2:And after that we have also major broadcasters in France who like the product so much, the ability to combine all these features together. The engineers generally like it very much. And we have broadcasters that are doing everything with us. So they're doing the mezzanine streams, encoding from SDI, they do it with us, they do the other OTT streams with us, do the HLS packaging. All of this we can do.
Speaker 1:Interesting, yeah, interesting. Now, how are you deployed? So, is this going in public cloud? Is it going in private clouds? You know, is it machines, you know, in a data center, in the broadcast? You know, broadcast facility originally.
Speaker 2:We're a software provider, so we we we sell our software and they just install it on on regular pcs from dell hp, super micro, whatever sure x86 right or do you support? Our x86 at the moment. We we used to have an arm version in the past past. We may start it again eventually.
Speaker 2:But, at the moment. X86, yes, and so Intel and AMD, basically yeah, and so originally people run it in their own local data center. Usually All broadcasters have some kind of room that they call their data center and they run it there. But more and more we're shifting towards the services model. I think that's the case of most software editors. Actually, many people just start their own cloud. You know that represents now a majority of our revenue, actually the services part, and so we just run our software for our customers, the kind of things we do we downlink channels from the satellite, we transcode them, we send it via SRT to clients at the other end of the world or to French telcos and so on. So that's now a major part of our revenue. Interesting, yeah, that's great. And of course it also runs on the cloud. We have many customers run it on AWS or GCP, for instance, and ourselves we have some servers. We probably have 30 servers in the cloud for our own customers as well.
Speaker 1:Yeah, interesting. Okay, well, somewhere along the line you came across this new category of encoder called BPU and this company called NetIn. So how did that happen? And you know, I think a lot of listeners would be interested to hear your journey from software encoding on CPU. You know, curious if you tried, or if you deployed GPU or if you just jumped straight to VPU. But why don't you share your experiences there?
Speaker 2:Yes, I have been in the past. I have been a strong advocate, actually, of software encoding. In the 2000s I was one of the first to put into production 6.4 on channels that were previously done by FPGA or that kind of technology.
Speaker 1:And, by the way, in 2000, you were going very much against the conventional wisdom. You know, everything was hardware encoders, so you know you were really. You were on the cutting edge.
Speaker 2:Absolutely, because in 2010,. It made sense because the new generation of Intel CPUs they were so powerful that you couldn't do several AG channels per servers, so it was possible. Now we're facing different challenges. A couple of months ago I won't say the name of the customer, but we had a customer coming with us with 100 channels to encode 100. So it's a big group, big group doing news, and they have different languages. That's why they have so many channels and each of these channels had to be sent over OTT, so they had to do multiple renditions for each of the channels and we calculated what would be needed with the CPU to do all of that and we calculated it would be something like 10 servers, 10 servers with 64-core CPU, not your regular server.
Speaker 1:Those are expensive servers.
Speaker 2:Yeah, yeah, I think $15,000 or so, probably, and I'm just talking about main servers, but you have also backup servers, so yeah, so at this point it was not reasonable to do the project with that number of CPU. It's too high. Well, approximately a year ago, I think, we came into NetSense. We were neighbors.
Speaker 2:Actually, at NAB last year we were we were neighbors so we talked, but I think we knew you from previous IBC as well. So this is a technology we've been looking into because with OTT, you need more and more encoding, actually because you need four or five renditions per stream. Some groups they have multiple channels, so you need to lower the cost of encoding these channels. Also, another reason is that you have new codecs coming. So H.264 is one thing. H.264 works very well in software, but more and more people are asking for HEVC, they're asking for 4K, and what's next One? Maybe Everyone in software encoding in software, everyone. Good luck.
Speaker 2:I wish it would be possible, but I don't think so at the moment, not in real time at least. So you have to look at the other choices that you have on the table. We've had a look at GPUs. To be honest, the GPUs, they have a limited number of streams they can do concurrently and also they just suck up a lot of electricity. Your regular GPU it's 300 watts or 500 watts or something like that. It's big.
Speaker 2:So and when we read NetInt fact sheets, basically we saw that theoretically we could do the project I was talking about. That required 10 servers. We can do it with one, one server, you just need to put as many NetInt modules as possible in the front panel. But NetInt made a very clever choice to have a form factor that is based on NVMe. So, like a disk, that means on a regular one rack unit server, you can put up to eight or ten modules. So that's a game changer.
Speaker 2:Actually, if you were just PCI Express like a GPU, like a regular GPU, you could put one, maybe two, per one rack unit per server. Otherwise you will need very big servers that are very, very costly. So no, the form factor is also a game changer. So we've been working with you for six months one year now and trying to make it work for our broadcast use cases, which is a little bit different to what you've been used to doing up to now. As I said, we have very strong requirements on constant bitrate, the use cases, which is a little bit different to what you've been used to doing up to now. As I said, we have very strong requirements on constant bitrate, the TS compliance and so on, which required a little bit of modification on both sides. But I think now we've nailed it, mark, I think.
Speaker 1:Yeah, I think it's worth pointing out to the listeners, out to the listeners. You know there's a perception about hardware silicon encoders that actually is largely correct At least it was and it is for other products, not ours, but here's the perception. The perception is that that encoder, no matter how good or not so good it is, is like frozen in time At the moment that that chip is taped out. It can never be changed, it's just kind of fixed. So if it works for you, great. If it doesn't, well, you have to wait for the next generation.
Speaker 1:One of the things that also is very smart about how we approached it is certainly there's elements of, for example, the rate control, which is lockdown. You know choices had to be made. It's hardwired in silicon, the gates are all you know that can't be changed. But so much around. The encoder can be modified with a firmware update and so we're actually able to take what normally is a piece of silicon that's pretty, you know, kind of fixed function, and even years after that product has been released to market we can make it do things that normally you wouldn't be able to. And so the work that you're referring to is exactly, you know, as a result of those design and engineering choices that we made, which I think is just important to point out, because I know a lot of people are quite shocked when they realize whoa, you mean that some of these requirements that maybe you don't support today you could support, and a lot of times the answer is yes, that's right.
Speaker 2:Yes, it just required a bit of work on both sides on your side and our side as well, that's right.
Speaker 1:That's right.
Speaker 2:But we finally managed it and we will show it at NAB together.
Speaker 1:At NAB. Yeah, yeah, that's great. Well, you know, christophe, we're very excited to have you and have EasyTools in our booth. So, as we've been promoting quite heavily, this NEB is going to be really exciting because we have a small number of partners of which EasyTools is going to definitely be featured, and it's just an opportunity for the industry to come learn about what VPUs can do. But, more importantly, I often say we sell engines. The VPU is like an engine, but, at the end of the day, almost everyone needs to buy a car. So, though, an engine is very important, and most people, when they buy a car, want to know that, hey, it's an engine that's fuel efficient, it can produce a lot of power, it's reliable, all of those things, but at the end of the day, you have to buy a car. So we're excited because you bring the whole software framework, the media processing framework, for a broadcaster that they can deploy a headend much more affordably, much more flexible, a lot more flexibility, really get a modern tool chain, if you will. So we're excited about that.
Speaker 2:Exactly, and we'll be able to expand into use cases you've never dreamed of. For instance, do you know that we can use your?
Speaker 1:NetInt module to build a multiviewer and the module will just's a use case that you know. Obviously we know about, but not too many folks have been out talking about it and building a product around it, so we're excited about that yeah us too. What else will you be showing off at NAB? Can you give some insights into what visitors will see when they come to your kiosk?
Speaker 2:We will be presenting demonstrations. The main focus of the year for us is really this NetInt integration. So we will show the various ways where we can integrate NetInt products so encoding, transcoding, multiviewers. That's basically it. But by doing that we'll also be able to see all of the features that we have in the motivator, for instance, which some of the features are very interesting. But yeah, the focus will be primarily on our integration with NetInt and how well it works.
Speaker 1:Sure, yeah, that's great. Well, we are in the West Hall. It's on booth W3531. The West Hall, it's booth W3531. And I just want to encourage everyone if you're at NAB, make sure you come meet Christoph, come talk to the Easy Tools team, say hi to me, say hi to the rest of the gang at NetIn. Really, we are coming to NAB to show off our partners, to show off partners like Easy Tools coming to NAB to show off our partners, show off partners like EasyTools, show off the ecosystem that is now really quickly emerging, being built out, and so I think that you definitely want to make sure you don't miss the NetEnt booth. So, christophe, you know, thank you so much for joining us. It was really great talking with you. Wonderful to hear the story. You know, I have a feeling we should do a follow-up interview. You know, it'd be kind of cool just to talk about the early days of developing VLC as a student project. That might be kind of cool. Maybe we should bring some of the original, you know, team members and contributors to the project.
Speaker 2:The original one. It would be difficult, because I think I'm the only one that is still working in the industry. Really, everybody else is doing other things you have younger ones that are very well known, like Jean-Baptiste Kempf, who is now the president. He says he's a lot younger than me, so he cannot possibly be the founder of PLC, but originally we were four or five and, to be honest, I have lost track of the other four.
Speaker 1:Yeah, wow, well, cool. Well, it would be awesome to do a session where we talk with you, you know, and we just talk about those early days. You know, there's so much technology that the industry really owes, you know, I would say, a debt of gratitude toward you know too, because if some of these early projects VLC being one of them, you know, ffm, peg, there's, you know, there's others right, but if those didn't come to be, you know, obviously solutions would have been created. But, you know, I think the industry would look different. So, absolutely, absolutely.
Speaker 1:Yeah, yeah, that's wonderful, all right. Well, christoph, thank you again for joining us on Voices of Video. Can't wait to see you in just, oh, three, three and a half short weeks, or three weeks, now it's. It's fast approaching, so I look forward to spending time with you and, for all the listeners, don't forget Mark mark W3531 in your diaries, in your calendars. Make sure you stop by the booth, say hi to Christoph, get a demo of what Easy Tools is doing, and we will look forward to saying hi to all of you. All right, thank you again for listening and watching Voices of Video. This episode of Voices of Video is brought to you by NetInt Technologies If you are looking for cutting-edge video encoding solutions.
Speaker 2:check out NetInt's products at netintcom.
