Matt Oughton is the EMEA sales manager for Vicon and talks about the Vicon motion capture system that they were demonstrating at the IEEE VR conference. Vicon has been in the motion capture business since 1984, and he talks about some of the specifications and use cases for their system. Vicon cameras are used for Virtual Reality tracking, movies and gaming entertainment as well as in the life sciences and engineering applications and industry design reviews.
He talks about some of the different high-precision systems that can track up to 150,000 markers and a refresh rate that can go up to 2000 Hz. Most of the Vicon camera systems for virtual reality would range from 30 to 250Hz and be able to track up to 50 objects or around 200 individual markers.
The price of their solutions can range as low as 5000 pounds and over a million pounds, and when I asked Matt whether Vicon is considering getting into the consumer market and he said that they’re primarily focused on the high-end and high-precision applications. After hearing about the upper range of some of the specifications for what their systems are able to do in a wireless fashion, then it seems like they’ll continue to serve the needs of their industry customers. However, Matt says that the lowering cost of technology is really unpredictable and so it’s difficult to predict how the technology in the space will continue to evolve. So it’s yet to be seen whether or not Vicon will be disrupted by some of the other consumer-grade motion capture systems that are emerging.
Theme music: “Fatality” by Tigoolio
Subscribe to the Voices of VR podcast.
Rough Transcript
[00:00:05.412] Kent Bye: The Voices of VR Podcast.
[00:00:12.042] Matt Oughton: So I'm Matt Orton, I'm the EMEA Sales Manager for Vicon Motion Systems, and here we've got the Vicon Bonita system running Vicon Tracker software, where we're tracking an Oculus Rift and a user's hand to interact with Eon Reality software. The demonstration here is basically an educational program for small children to pick up body organs and put them in the right place so they learn that part. And the more immersive it is, the better they learn, the more fun they're having, the less it is looking at a textbook.
[00:00:40.808] Kent Bye: Yeah, and so maybe talk a bit about how long Vicon has been around and the part that it's played in the virtual reality research community and industry over the years.
[00:00:49.994] Matt Oughton: Okay, so Vicon has been around for 31 years now. It started as a spin-out out of Oxford University and then a management buyout out of Oxford Instruments. We have developed and led the use of motion capture in a variety of fields. Clinically, primarily, and then moving into the animation worlds, things like films and computer games, and more recently, as technology has caught up, into virtual reality and immersive augmented reality as well. It's used in a wide variety of things, so education is one obviously as we're showing here, but it's also used in things like design environments, virtual training, things where it's too expensive, too dangerous to train in the real world. So you get that initial training virtually before moving out into the real environments.
[00:01:37.018] Kent Bye: And so yeah, maybe you could talk a bit about the balls and the tracking that you need to put on a HMD and how that then is being multiplexed and sensor fused into the inputs to be able to actually track into something like a Unity 3D program. Okay, sure.
[00:01:52.154] Matt Oughton: So the way the system works, we have a variety, or a number of cameras, in this case we have six cameras, and they emit a near-infrared wavelength. That wavelength of light hits the retroreflective markers, which is then returned back to the camera. Each camera has been calibrated so it knows where it is in relationship to the others, and therefore we can triangulate each marker in 3D. In the Bicron software we have a number of markers which are modelled together in an object, so in this case the Oculus Rift for example, and we model those orientation of markers continuously. So as the user walks around with the Oculus on their head, we are continuously modelling where they are, where they're looking at. And that data is streamed out either as individual markers or in this case just the central point of the route, if you like, of the model. That data point gets streamed through for a variety of methods into third-party software, in this case Eon Reality, and then it's up to that third-party software to create the rendering and use that data essentially. So we are, in very simple terms and doing ourselves a huge disservice, a tool to the means to an end. What it enables you to do is have very high quality tracking. We are precise to within 0.5 millimeters and that enables you to have a really immersive environment. Low latency means that you have a more immersive environment but you feel less nauseous which you do with other tracking systems. The Oculus system for example on its own does have limited tracking in it but it's quite imprecise and the latency can make the user feel nauseous because you move your head and then a perceptible moment of time later the display moves with you and that cannot have other knock-on effects. It makes you realize that you're in an immersive world rather than just getting on and learning the program that you're supposed to be doing.
[00:03:38.152] Kent Bye: And what's the frequency that, you know, how fast of an update rate are these infrared sensors and how many per second, how fast is it going?
[00:03:45.915] Matt Oughton: So the Vicon system, these cameras here will run up to 250 Hertz. We have other cameras that will run up to 2,000 Hertz. So it really depends on what you want. For most VR world, we run, tend to run about 240, 250. It doesn't need to be that high, but one of the benefits of running that fast is that the movement of the marker between each frame is a lot lower. And that means that we can track more accurately, more precisely, with less dropouts. It just makes for a better user experience. We have plenty of customers around the world who limit their tracking capabilities to 60Hz or 100Hz. There's no right or wrong answer. It's whatever you feel best as a user. And we leave that entirely up to you. So the slowest you can run the system up is 30Hz. and these cameras will go up to 250, but like I said, other cameras will go higher.
[00:04:29.507] Kent Bye: What would be the use case for having an update rate of up to 2 kHz?
[00:04:33.875] Matt Oughton: For a VR world, probably not very many, but if you are in life science world and you're looking at how the foot impacts in sprinting, for example, that's a very high speed movement or a golf swing, how the golf shaft distorts when it hits the golf ball. That's the type of thing where you want to have a really, really high frame rate. But for your standard VR world, you're not going to want to go that high. So you have the cameras like this, which are limited to 250.
[00:04:59.657] Kent Bye: And what is the upper limit in terms of how many individual markers you could track on a body or how many objects you could have? Is it per ball or what's the upper threshold of how many things you can track?
[00:05:10.608] Matt Oughton: So that's dependent on the camera and the software that you're using. The system we have here, we could probably track quite happily 50 objects and we define an object as four or five markers together. So the four or five markers on the Oculus Rift we count as one object. So, 50 objects is absolutely fine. However, with our other cameras, the larger T-Series cameras, which we do not have here, but they're available for you to view on the website, they have onboard processing and we can track, I believe it's something of the order of 130,000 individual markers per frame. So, if you're tracking at 100 hertz, that's a hell of a lot of markers. And if you're using that many markers, I want to come see what you're doing.
[00:05:47.916] Kent Bye: Yeah, I'm curious what kind of use cases you've seen in terms of either the military or industry, and especially in the context of virtual reality, how you see people using these trackers.
[00:05:58.918] Matt Oughton: So the things that we see more often than not are design reviews. You're an automotive maker and you want to understand the latest design from the design department. Does that meet your engineering spec? Does that meet your marketing spec? Is it usable? So as you're sat in the driver's seat, for example, can you comfortably reach the radio? Can you comfortably reach across and let your passenger in and out of the door? By doing it virtually in the virtual world, it's a lot quicker. There is still obviously time invested in making that digital model, but it's a lot quicker and certainly a lot cheaper to do it virtually than to hack about pieces of plastic or wood, which is how they used to do it. And by doing it virtually, you can learn and collaborate with design offices across the world or across the country and just have a much faster turnaround between initial design to finished piece. In terms of real-world uses for large numbers of markers, again, probably not in the VR world, but we are also involved quite heavily in other engineering markets. So there's an institute in Japan which creates life-size or half-life-size buildings on earthquake shaker tables, puts markers on every brick or every number of bricks, sets off a simulation of an earthquake and watches how the building copes with that, how it falls, how it crumbles. Obviously that's a large number of markers. There's also plenty of, in the clinical world, plenty of experiments where they look at skin deformation. So they'll literally pepper somebody in markers and then get them to walk around, move, and understand how the skin stretches over the bones and over the muscles. The VR world is relatively straightforward from a tracking point of view because it tends to be at least just the head, or you may go full body, but that's still only a limited number of markers.
[00:07:35.915] Kent Bye: And what is the cost in terms of the minimum amount of cameras that you would need to do in order to do a VR tracking scenario?
[00:07:44.118] Matt Oughton: Well, that really depends upon the environment that you want to track within and the number of objects or people that you want to track in. So I guess the simplest thing would be stood in front of a power screen or a 3D projection TV, and that minimum would be two cameras. And you're probably looking at, certainly in the UK, a cost of around the £5,000 mark for that. As you go up to more complex scenarios, you're going to want to have many more cameras. So again, not necessarily in the VR world, but in the entertainment world, for example, where they're capturing entire football teams playing soccer against each other for computer games, they may have 250 cameras. And you'll be looking at an investment of well into over a million pounds for that. And there's everything in between. If you can think of it, we've probably captured it, or our customers have probably captured it. There's some pretty amazing things if you check out YouTube and search for Vicon 100. There's plenty of real-world examples out there with a whole number of cameras from, like I say, from 2 to 250.
[00:08:41.897] Kent Bye: Yeah, and it seems like that, you know, with the virtual reality world, that there's a lot of consumer market level of stuff. I know the valves lighthouse system of being able to kind of at 100 hertz, use lasers with X and Y to be able to track a 15 by 15 space. And I'm curious if Vicon is sort of looking to perhaps in the future, produce a more consumer range in terms of perhaps it's not, you know, at the spec of industry, but, you know, kind of a lower range that might be available for the consumer market.
[00:09:12.663] Matt Oughton: It's certainly an interesting idea. We're aiming for the very high precision marketplace. Costs are falling all the time. When I first joined Vicon eight years ago, a system like we have here would be well in the order of £100,000. Obviously, we're now down to the £10,000 to £20,000 mark. That's a huge decrease in a relatively short period of time. Who's to say that in another eight years' time this system wouldn't be available for £1,000 or £2,000? The process of reducing the cost of technology and whilst equally increasing the ability of technology is advancing at such a rapid rate. I really wouldn't want to be the person to predict it. If you think about the mobile phone you have in your pocket right now, if someone said to you 10 years ago, you'd be able to do all these things, have the internet on there, you'd laugh at them. But here it is now, and it's commonplace. So who's to say? Hopefully.
[00:10:01.239] Kent Bye: Great. And finally, what do you see that Vicon's able to really contribute to creating a sense of presence and immersion within virtual reality experiences?
[00:10:09.822] Matt Oughton: I think that comes down to the high fidelity and precision of the markers and also some of the unique things that we're able to do. So we're able to track through occlusion, basically. So if, say, you've got five markers on your Oculus and you turn and position your head in such a way where, say, two of those markers are occluded, we can still give you full, reliable tracking from those remaining three markers because we're projecting that 3D model onto what we're seeing. It also enables us to do single camera tracking. So once you've booted from multiple cameras, you can go to the very corner of your cave, and as long as one camera can see, you can still maintain that full 3D tracking. So essentially, with the Vicon system, you're able to do more with less. With our competitors out there in the motion capture, motion tracking area, you'd still need to have at least two cameras on there. So the number of cameras that you'd need to do the same thing would be different.
[00:10:56.212] Kent Bye: Okay, great. Well, thank you so much.
[00:10:57.653] Matt Oughton: That's all. Thanks very much.
