VR is the Next Big Thing in computing right now, but I don’t own my own VR hardware yet. I consider myself an early—but cautious—adopter. I didn’t back the Rift’s Kickstarter or preorder HTC’s Vive, but not for a lack of interest. I’m actually fascinated by this new wave of VR and augmented-reality, or AR, technology. One reason I find them so intriguing is the vast untapped potential of these fields. My personal interest was piqued when I read The Atopia Chronicles in early 2013. That book’s modern take on AR got my head spinning. I think it should be required reading for VR and AR developers, but I digress.
Another large part of my fascination with VR stems from watching PC hardware and software developers grapple with the problems that VR introduces. I know I’m not alone in feeling thankful for something that’s pushing the envelope on optimization of software and drivers. Many of the usual suspects are already doing their part. Nvidia and AMD are working closely with Oculus, as well as HTC and Valve, to release drivers that improve the VR experience. Heavy hitting game developers like Epic, Unity Technologies, and Crytek have added VR support to their engines to varying degrees, as well.
Gato 2.0 approves – or does he? Also, he heard something about gerbils?
Just because the biggest players are already firmly seated doesn’t mean there isn’t room for newcomers, though. Whether you think VR is a new industry, a bunch of companies turning an existing one on its ear, or just a fad, there’s no denying it attracts a lot of attention. Tons of investors, inventors, and entrepreneurs are trying to make VR the next big thing.
One of those inventors, Aaron Schradin, happens to be a good friend of mine. His device, the Turris, is meant to change the way we traverse VR environments. I’ve had the opportunity to watch the Turris develop as it’s made the rounds to SXSW and other conventions for over a year now. I thought it was about time I took advantage of my insider connection to share a TR-exclusive in-depth preview.
If I had one of these, I’d display it a little more prominently.
A little background information
Thanks in part to TR, Aaron got connected with AMD during the launch-hype surrounding that company’s Dragon Platform back in 2008. I still remember him telling some of AMD’s best extreme overclockers how they were doing it wrong, and it still makes me chuckle to this day. From there, Aaron consulted with AMD and eventually developed a number of custom cooling pots for extreme overclocking, including a see-through lexan pot for LN2 and a pot specifically designed for liquid helium that was eventually used to set a world record for CPU clock speed when Bulldozer launched. The industry connections he made around this time eventually pulled him into the world of VR, where another “you’re doing it wrong” moment lead to the founding of his company, Praevidi, and the creation of the Turris.
The concept behind the Turris is pretty straightforward: you sit on it, and your movement in VR space mimics your real-world movement. Tilting forward, backward, left and right moves your character in those directions, and rotating the chair rotates your character 1:1 in VR. The buzzword that makes these abilities important is de-coupling. In engines that support it, de-coupling means that your torso’s position and direction is tracked separately from where your head is looking. It’s a function that sounds simple on paper, but it’s kind of a big deal for the industry, especially for seated applications like the Rift. Take a look at the impromptu video below for a demonstration (my apologies for the small on-screen character size—it’s probably best to watch in full-screen so you can actually see that detail).
TL;DR version right here…
How does it work?
The Turris appears as an Xinput-compatible device when it’s hooked up to the PC (in short, like an Xbox 360 controller). The directional inputs from the seat are interpreted as digital, WASD-style movement of the left thumbstick. The rotational movement is similarly interpreted as left or right movement of the right thumbstick, depending on the direction of rotation. According to Aaron, this means that any developer in possession of an Xbox controller already owns a Turris dev kit.
In its current form, the switches on the Turris are mounted into 3D-printed springboards designed to prevent them from being overly stressed when they are actuated. There’s a few other details hidden under the seat, as well. Bolts with rubber caps act as the seat’s fulcrums. These bolts can be adjusted to fine-tune the amount of tilt required to initiate movement, kind of like adjustable weights in a mouse. The same adjustments can also be used to compensate for differences in user height, weight, and seating posture, since those variables all affect the location of the user’s center of gravity.
A glimpse under the cushion.
There are also hard stops around the outside edge of the base that line up with adjustable steps under the top of the seat. These stops work in conjunction with the fulcrums to limit the maximum amount of tilt. That adjustment range is important when you’re using your body weight to control fine movement in VR, because you want to make the subtlest adjustments possible. If your center of gravity is allowed to shift too far in any direction, it requires at least that much more movement to change directions back again.
Aaron tells me that he’s also tested switches that have two different actuation points. Those switches could mean that shifting further in a direction could change movement from a walk to a run. That makes nailing the per-user adjustability of the seat all the more critical. He’s also gotten the Turris working with gyroscopic-motion-controls (an MPU-6050, for the curious) to allow for more analog-style control of speed and direction, but users have so far preferred the simpler and more predictable movement that comes from the on/off toggle of digital switches.
Aaron explains the Turris to TR gerbil Sheldomaus.
The Turris tracks rotational movement with a black-and-white ring in its base. An optical encoder in the free-spinning section tracks this ring as the chair moves. The encoder tracks the absolute angle and position of the chair in 2.8-degree increments. If you do the math, that means the chair can occupy 128 unique positions throughout its 360º rotation. Aaron says that the 2.8-degree value is the sweet spot for balancing granularity of movement and avoiding unintentional judder that could lead to motion sickness. Since the Turris is still in development, though, these parameters are still being tweaked.
Connecting it to a PC
There are two ways to hook up a PC to the Turris. The most common method will probably be to use an existing PC setup and treat the Turris as a peripheral. That setup is made possible by the rotary slip connector in the base of the chair. The connector carries power, USB, and HDMI signals to the chair, and it’s what makes the Turris’s goal of tangle-free, 360-degree rotation possible. This connector doesn’t pass Ethernet signals, though. A wired network connection over USB should work fine, but that method hasn’t been tested. The prototype Turris also has three spare “developer” lines open for experimentation. Aaron took delivery of an HTC Vive last week, and he’s already testing that HMD with the Turris. Handling the 90-Hz, 2160×1080 signalling demands of retail VR is more difficult than the development VR hardware that’s been used until now, though. The Turris grew up alongside the 75-Hz, 1920×1080 screen of the Oculus DK2, and that headseat remains the Turris’ primary demo platform for the moment.
The Turris connected to an external PC.
The second option for using a PC with the Turris allows it to work without passing video from an external source to the VR headset. As it happens, the Turris itself doubles as a PC enclosure. With a PC built right into the chair, all the computing power goes along for the ride when the chair is spun around. A PC in the base of the unit lets the whole setup work without the need to transmit video signals through the slip connector. The connector still has to carry power for the PC, as well as any USB connections for things like tracking cameras, a keyboard, or a mouse. With the PC on the inside of the Turris, the HDMI cable in the slip connector can still be used to send an image out to an external display. Any PC built inside the Turris should probably skip mechanical hard drives and go 100% solid-state, for obvious reasons. Aaron didn’t make that suggestion directly, probably because it goes without saying for the typical PC enthusiast.
PC basics installed.
As a PC case, the Turris is extremely bare-bones. In fact, it’s closer to an open desktop test bench than a case in most respects. A Turris with a PC assembled inside wasn’t available when I took my other photographs, but I was able to bring a Turris base home and toss a spare motherboard and graphics card in it to give you an idea of how it comes together. The Turris is only compatible with Mini-ITX motherboards right now, but certain microATX boards fit, as well. An eventual design goal for the Turris is to accomodate every microATX board, but space remains tight.
The Turris has room for double-wide graphics cards—a requirement for anything powerful enough to run a VR headset right now. As for board lengths, a Radeon R9 Nano fits easily, as would most sawed-off GeForce GTX 970s. The 10.5″ sweet spot for many reference graphics cards would leave the card hanging 0.25″ past the outside of the t-rail structure, though. The shroud around the chair has standoffs that leave a gap, though, so there should be room for those cards so long as the PCIe power plugs into the side of the card instead of the front edge. Cards longer than 10.5″ are probably chancing clearance issues. The flexibility inherent to t-rail construction means that there are options to include any number of mounting brackets that could work with a PCIe riser card or extension cable, but none of those add-ons are finalized yet. While it’s not pictured, the Turris offers a bracket for supporting the graphics card with screws, as it would be in a traditional case.
Finally, a legit reason to own an R9 Nano!
Other core case features that are MIA are fan mounts, a finalized PSU bracket, and any form of drive cage. I wouldn’t expect 5.25″ or 3.5″ bays to fit in the final Turris, but that’s probably not an issue for most people. Some trickier issues for the Turris are dust filtration and cable management. I’ve seen various incarnations of mesh grills covering the Turris’ internals, but nothing that could be called a filter yet. The design of the shroud does offer room for large intake and exhaust vents, though. Cable management can be tough for any small-form-factor system, and the extra cables that come with VR compound that problem. There is space to cram cables into various hidey-holes in the Turris, but the chair doesn’t have any explicit cable routing solution for now.
Based on its current state, I would much rather use the Turris with an external PC. If I were to build a dedicated Turris VR PC, I would want to use a 65W Skylake Core i5 CPU paired with a short GTX 970. A motherboard that afforded any version of M.2 storage and a modular PSU would reduce cabling concerns, too. A system like that, combined with the open air nature of the “case,” wouldn’t make me worry too much about heat (especially with a tower cooler on the CPU). It wouldn’t be a beast, but it would get the job done without breaking the bank.
The Turris and Gear VR
The Turris is designed and optionally configured to work with mobile VR devices, most notably Samsung’s Gear VR. The mobile Turris connects to the phone using Bluetooth LE. All the hardware inside the mobile version, which includes the identical switches and encoder of the PC version, is powered by an off-the-shelf USB battery pack. Aaron says the 10,000-mAh pack he uses for testing can power the chair for “days” as long as “you don’t plug the LED accent lights into it.” The mobile version works more or less the same as the PC version, including the critical de-coupled functionality in engines that support it. I’ve run the demo shown in the video at the beginning of the story on both the PC and Gear VR. The mobile version looks pretty good, despite the inherent hardware limitations of a smartphone. The functionality of the Turris is the same across both platforms.
Using the Turris
To make a long story short, I’ve taken the Turris for a spin a number of times, and it works exactly as described. The only fiddly part of using the chair is the initial “butt calibration” required to find a natural resting center that doesn’t trigger the switches. Since I’m evaluating a chair used for public demos, it’s configured with a one-size-fits-all approach. That calibration worked OK, but not great.
Fine-tuning the chair for my natural seating position took a few minutes of trial and error. I alternated between sitting on the unit and kneeling next to the chair to adjust the stops and pivot points. I’d estimate that those few minutes got me about 90% of the responsiveness I wanted. If I had unlimited access to the Turris, I would want to go another round or two with it to really get it dialed in.
I’ve mostly used the Turris with the Oculus DK2 in two different demos that showcased its compatibility with two different engines: Unity and Unreal. Turris has a CryEngine demo in the works, but it’s not quite ready yet.
The first demo I tried, PolyWorld, is a Unity-powered project that supports the de-coupled tracking of the headset and the Turris. I found it simple and intuitive to traverse the demo’s small town and its surrounding caves and bridges, all without using my hands. Since PolyWorld is a demo for new Turris users—and newbies to VR in general— the PolyWorld experience is slow and exploratory in nature.
The Gear VR version of PolyWorld.
You’ll recall that the forward, backward, left, and right controls on the Turris are binary in nature. Because it’s seen as an Xinput device, though, the travel speed when each switch is triggered can be set to different levels in software. During my demo, Aaron pranked me by cranking up the speed behind the scenes. That change surprised and disoriented me a bit, but it didn’t make me nauseous. I’ve played seated VR demos with the DK2 before, and that experience has taught me that I have motion sickness issues. I suppose that the de-coupled awareness of the Turris made the difference in this case, though. With the Turris keeping tabs on my torso, my virtual face-camera wasn’t able to make any promises to my brain that my body couldn’t keep.
The second demo I tried was built in-house at Turris using the Unreal Engine. It’s tailor-made to make the user immediately aware of their presence in the game and highlight the difference that de-coupled movement makes. It’s ingeniously simple— just a mirror in VR that you can use to see your virtual self—but it’s strangely compelling. Even without the mirror, you can see your body in this demo. If you turn your head, you can look at your shoulders and see your head move in the mirror. If you keep your head still and rotate the Turris, you can see that your feet are pivoting on the ground without your head moving. Running through the maze after you’ve checked yourself out in the mirror is also a smart way to show off how effectively the Turris operates.
The PC Unreal engine mirror and maze demo.
Even though the body-weight-shifting or torso-tilting motions one makes to activate the Turris work naturally for the most part, there are situations where something you want to do in VR might be at odds with the inputs the Turris offers. A good example of this is leaning to peak around a virtual corner. With keyboard FPS controls, you would typically use the Q or E key for this motion. In VR, well, you just actually peek around a corner. If you’re seated on a Turris, that lean could send your character strafing out from behind cover and into danger. Developers might have to build in a feature that temporarily disables body control and only registers head tracking to get around this issue.
The Unreal engine supports the Leap Motion controller, as well, which theoretically allows for fully-articulated hand tracking in VR without a controller. Unfortunately, I wasn’t able to try out the Turris and the Leap working together. I have tried the Leap with its most recent software update separately, though, and it’s fairly impressive. I’ve also tried the hand controllers included with HTC’s Vive since running the Turris demos, and the Vive controllers are quite effective as virtual graspers. I can see how either of these two technologies could go hand in hand with the Turris to create a much more dynamic seated VR experience than an Xbox One controller in a normal desk chair offers.
This picture included to embarrass Sheldomaus.
Before wrapping up, I want to make three other small notes about the Turris. First, the appearance of the chair seems to be trying to straddle professional and “gamer” aesthetics. I know many TR readers aren’t impressed by the trappings of traditional gaming peripherals ,and I generally feel the same way. The design of the chair isn’t final, but in its present state I think it could blend in with many interior environments without issue. Second, I should mention the noise the Turris makes while it’s rotating. It’s quiet, but not silent—somewhere between noiselessly swiveling on an office chair and the sound the same office chair would make rolling over carpet. Lastly, I think it’s important to mention that even though the Turris is a specialty VR peripheral, it’s also just a chair, so it doesn’t require any dedicated storage space or elaborate setup process to use.
Praevidi, Aaron’s company, spends a lot of time explaining why de-coupled movement is so important on its site, but after my experience with the device I think I can describe its value more succinctly. The short version is that when simulating something as all-encompassing as reality, tracking head and torso position independently could be just as important as tracking your hands for creating a sense of immersion and presence (and seriously, devs, put some mirrors in your games—that helps, too). I now know how true this is for seated VR experiences, but my sense is that torso-tracking will need to be addressed in room-scale VR, as well.
The pieces of the VR puzzle are still coming together. I haven’t quite seen them all put together yet, even though I’ve come closer than most. I believe that bringing your hands into VR with hardware like the HTC Vive controllers or hand-tracking sensors like the Leap is a critical piece of the puzzle. I’m not sure I would go so far as to call the Turris “seated room-scale VR,” but the way that the Turris subtly involves one’s feet and torso in the VR experience makes a strong impression that it can be an edge or maybe even a corner piece of the VR puzzle.
The Turris doesn’t have a release date or price yet, but Praevidi is targeting the fourth quarter of this year for the release of the Gear VR-compatible version. The company is also making units available to developers that want to incorporate the Turris into their VR experiences.