The Inside Story of the Oculus Rift, VR's Singularity MomentJuly 03, 2014
by Peter Rubin
For the better part of a century, science-fiction writers, technologists and philosophers have contemplated the arrival of a digital realm so immersive that it is neurologically indistinguishable from the outside world. Now it's here. The inside story of the Oculus Rift and how virtual reality became reality.
As he flew from Orange County to Seattle in September 2013, Brendan Iribe, the CEO of Oculus, couldn't envision what the next six months would bring.
The rhapsodic crowds at the Consumer Electronics Show. The around-the-block lines at South by Southwest. Most of all, the $2 billion (£1.2 billion) purchase by Facebook. That autumn, Oculus was still just an ambitious startup chasing virtual reality, a dream that had foiled countless entrepreneurs and technologists for two decades. Oculus's flagship product, the Rift, was widely seen as the most promising virtual-reality device in years, enveloping users in an all-encompassing simulacrum that felt like something out ofSnow Crash orStar Trek. But it faced the same problem that had bedevilled would-be pioneers such as eMagin, Vuzix, even Nintendo: it made people want to throw up.
This was the problem with virtual reality. It couldn't just be really good. It had to be perfect. In a traditional video game, too much latency is annoying -- you push a button and by the time your action registers onscreen you're already dead. But with virtual reality, it's nauseating. If you turn your head and the image that's centimetres from your eyes doesn't adjust instantaneously, your visual system conflicts with your vestibular system and you get sick.
There were a million little problems like that, tiny technical details that would need to be solved if virtual reality were ever to become more than a futurist's fantasy. The Rift had made enough headway to excite long-suffering VR enthusiasts, but it was still a long way from where it needed to be.
But then Iribe got a call from Michael Abrash, an engineer at Valve; the gaming-software company had conducted VR research for a while and had begun collaborating with Oculus. Valve had a new prototype, and it didn't make people sick. In fact, no one who had tried the demonstration had felt any discomfort. Iribe, who was famously sensitive to VR-induced discomfort -- "cold sweat syndrome", he calls it, or sometimes "the uncomfortable valley" -- flew up to Valve's offices outside Seattle to be the ultimate guinea pig. Abrash escorted Iribe into a small room tucked off a hallway. The walls and ceilings were plastered with printouts of QR-code-like symbols called fiducial markers; in the corner, a young engineer named Atman Binstock sat at a computer. Connected to the computer was Valve's prototype headset -- or at least the very beginnings of a headset, all exposed circuit boards and cables. Iribe slipped it over his head and found himself in a room, the air filled with hundreds of small cubes.
He turned his head to look behind him -- more floating cubes. Cubes to the left, cubes to the right, cubes overhead, floating away into infinity. Iribe leaned forward and peered around to see the side of the cube closest to him; he crouched and could see its underside. A small camera on the headset was reading the fiducial markers on the (real) wall and using that spatial information to track his position among the (virtual) cubes. So far, so good; no motion sickness yet.
Binstock tapped some keys and moved the demo to its next stage. Inside the headset, Iribe stood in a giant chamber, a web browser page on each wall. Iribe picked out a word on the wall across from him and started shaking his head back and forth, rotating as fast as he could, waiting for the word to smear across his vision and make him dizzy. Nothing. In any of Oculus's own prototype headsets, Iribe would have become nauseated long ago, but he was still feeling good.
As Binstock continued clicking through the demo, Iribe faded in and out of a series of rooms -- bare-bones virtual worlds filled with cubes and spheres. In all of them he took his time, moving, crouching, panning this way and that, taking in his 360° surroundings. Eventually he came to the grand finale, in which he floated slowly though a vast structure, its interior walls like some glowing mashup of Tron and a Death Star trench. The demo was at an end.
But Iribe couldn't take his headset off. "Again," he said, scarcely able to believe what he was asking for. They ran through the entire series once more. Finally Iribe took off the prototype. His head felt strange -- not dizzy, not displaced, but overwhelmed. "How long was I in there?" he asked Abrash and Binstock.
It had been close to 45 minutes.
"That's it," Iribe thought. "This is going to be bigger than I ever expected."
And that's saying something, because the expectations surrounding the Oculus Rift have always been huge, ever since an 18-year-old named Palmer Luckey hacked together a rough prototype in his parents' garage in Long Beach, California, in 2011. In June 2012, John Carmack, the legendary founder of id Software, the company that created Doom, Quake and the entire concept of 3D gaming -- brought that early prototype to the E3 video game show, reintroducing VR to the popular conversation for the first time since The Lawnmower Man. A year later, Oculus brought an HD prototype to E3 and blew minds all over again. Then it brought another, even more advanced one to CES last January. Then another unit to the Game Developers Conference in March. And finally, the $2 billion purchase by Facebook. All for a company that doesn't even have a commercial product yet and is chasing a dream that most of the tech community had seemingly given up on decades ago.
Oculus has almost single-handedly revived that dream. Luckey's advances have inspired Sony to announce its own forthcoming VR hardware, for now known only as Project Morpheus. Software developers, from Gears of War maker Epic Games to EVE Online studio CCP, have been designing new experiences for the Rift. And it goes beyond gaming: Developers are producing Rift-enabled tools to let users explore everything from molecules to galaxies. Framestore, the London-based visual-effects firm, created a virtual Game of Thrones experience for HBO; Gravity director Alfonso Cuarón has visited Oculus headquarters. Enough Hollywood types have come calling, in fact, that Oculus recently hired a director of film and media.
Beyond that, though, the company and its technology herald nothing less than the dawn of an entirely new era of communication. Mark Zuckerberg gestured at the possibilities himself in a Facebook post in March when he announced the acquisition: "Imagine enjoying a courtside seat at a game, studying in a classroom of students and teachers all over the world, or consulting with a doctor face-to-face -- just by putting on goggles in your home." That's the true promise of VR: going beyond the idea of immersion and achieving true presence -- the feeling of actually existing in a virtual space.
That's because Oculus has found a way to make a headset that does more than just hang a big screen in front of your face. By combining stereoscopic 3D, 360-degree visuals and a wide field of view -- along with a supersize dose of engineering and software magic -- it hacks your visual cortex. As far as your brain is concerned, there's no difference between experiencing something on the Rift and experiencing it in the real world. "This is the first time that we've succeeded in stimulating parts of the human visual system directly," says Abrash, the Valve engineer. "I don't get vertigo when I watch a video of the Grand Canyon on TV, but I do when I stand on a ledge in VR."
Now Oculus is hard at work on its long-awaited headset for consumers to buy, which the company predicts will be released later this year, or more likely early next year, or perhaps even not so early next year. But whenever it comes, we'll finally have something that has eluded us for more than 30 years: the chance to experience immersive, affordable virtual reality. And we'll all know what Brendan Iribe knew standing in that room outside of Seattle.
This is going to be bigger than we ever expected.
If there's a checklist for tech wunderkind, Oculus founder Palmer Luckey leaves no box unticked. There's the shoelessness, for one; he commutes in sandals and regularly pads barefoot around the Oculus offices in Irvine, California. There's the tousled hair, the anachronistic attachment to his 31km/l 2001 Honda Insight, the can of vitamin-enriched sparkling blackberry juice seemingly glued to his hand, and the confidence that comes from knowing a lot of things about a lot of things (or possibly from all that juice).
But most of all, there's the omnivorous curiosity. As a home-schooled teenager in Southern California, Luckey spent much of his free time tinkering with electronics -- modding video game consoles and repairing iPhones for extra cash, then spending the money on high-powered laser systems and upgrades for his gaming PC. The PC, in particular, became an obsession: Luckey found himself pouring tens of thousands of dollars into it. And soon, a hunt for 3D monitors became a search for true immersion. As a kid, he'd been entranced by the idea of getting inside the video games he played on his Game Boy Color. Virtual-world sci-fi such as The Matrix and the anime show Yu-Gi-Oh! intensified the desire. Why, he asked himself, can't we do that yet?
His modding and iPhone repair work had left him with a lot of money, so he bought a Vuzix iWear VR920, then the most cutting-edge consumer VR headset -- enthusiasts call them HMDs, for head-mounted displays -- on the market. Then he moved on to the more expensive eMagin Z800 3DVisor. And he kept looking. Over time, through a combination of government auctions and private resellers, he would spend the money once earmarked for PC upgrades on more than 50 different units, building what he touts as the largest private collection in the world.
But even these couldn't give Luckey the immersion he craved. When he put them on, he felt like he was looking at a play space, not living inside it. "It wasn't garbage," Luckey says, "but it wasn't virtual reality." The image quality was poor because the transmissive LCDs weren't high-contrast. The head-tracking latency was off the charts, causing a nauseating lag every time he turned his head. But most of all, the field of vision was too narrow. He could always see the edge of the screen, which meant his brain could never be truly tricked into thinking it was inside the game.
Luckey figured that he had as good a chance as anyone to solve those problems. So he tinkered, and tinkered some more, and one night in November 2010 he announced to the world -- or at least to the message-board denizens of a 3D-gaming news website called Meant to Be Seen -- the existence of PR1 (for Prototype 1), his first stab at a virtual-reality device. It was a cumbersome beast, built on the shell of a headset from his collection. It displayed only in 2D and was so heavy that it needed a one-kilogram counterweight in the back. But thanks to a massive chassis that could fit a nearly 15-centimetre display, it boasted a 90° field of vision, an angle nearly twice as large as anything else on the market.
Over the next ten months, Luckey kept tinkering, cracking problem after problem. He knew his headset would need a 3D display, but that meant two screens -- projecting slightly different images for each eye -- and even with the explosion of smartphone-ready display panels, there simply wasn't a hi-res panel small enough to fit two side by side in a headset. A few months after announcing the PR1, Luckey was browsing the documentation of a Fujitsu ultramobile PC he owned and noticed that the usable display area was 121 millimetres wide -- just about double the distance between a pair of human eyes. "What if I just used half of it for each image?" he thought. He put a separate lens over each half of the display, and just like that he had a 3D prototype. In September 2011, he announced the wireless PR3. The PR5, which he worked on throughout early 2012, had a gargantuan 270° field of vision (though it was neither wearable nor remotely practical). By that point, Luckey had become something of a celebrity on the Meant to Be Seen forums, whose members eagerly awaited his updates.
One of those members, it turned out, was John Carmack. The Texas engineer is known as the father of the first-person shooter, but games such as Doom and Wolfenstein 3D weren't important just for their violence or perspective. They were technological benchmarks, boasting sophisticated bespoke software engines that could make games faster and more immersive than ever before. Like Luckey, Carmack had always been obsessed with making games as lifelike as possible, an interest that had also led him to virtual reality. And like Luckey, he was routinely disappointed in what he found. "There were two broad camps," he says. "The hardcore academic research people looked down their noses at games. It was all about remote surgery and high-minded things. Then you had the popularisers -- pitching the vision, talking about how wonderful it was going to be, how it was going to change everything, but there wasn't enough technical acumen to get anything accomplished."
Carmack kept tinkering and eventually wound up on the Meant to Be Seen forums. It was there he learned about Palmer Luckey's ongoing project. Carmack was intrigued by the kid, especially when Luckey announced in April 2012 that he was building his sixth-generation unit, which he called the Rift.
"I based it on the idea that the HMD creates a rift between the real world and the virtual world," Luckey wrote on the forums, "though I have to admit that it is pretty silly. " He wrote that he'd be Kickstarting a DIY kit: He'd mail his backers the parts, which they could assemble themselves. After shelling out for the materials, manufacturing, shipping and fees, Luckey wrote, he expected that he'd make a grand total of $10 "for a celebratory pizza and beer". Intrigued, Carmack private-messaged him. Would Palmer consider sending him a loaner unit? Palmer, who idolised Carmack, shipped it off to Texas immediately -- "no NDAs, no signing anything," Carmack says. "It was one of two prototypes that he had."
Carmack got to work on the machine, glueing a motion sensor to it and duct-taping on a ski-goggle strap. But his greatest contribution came in the code he wrote for it. The Rift's biggest selling point was its 90° field of view, which Luckey accomplished by slapping a cheap magnifying lens on the display. The problem was, that lens distorted the image underneath, making it warped and uneven. So Carmack coded a version of Doom 3 that pre-distorted the image, counteracting the effects of the magnifying lens and making the picture appear correct to the viewer. The result was a completely immersive gaming experience, the kind that would otherwise require $10,000 in high-end optics.
Luckey was ecstatic to learn of Carmack's work -- but then Carmack upped the ante. He asked Luckey if he could "show it to some people at E3 in Los Angeles".
"Show it to whoever you want," Luckey told Carmack.
A few weeks later, Luckey was in Boston, attending a trade show about display technology; a friend texted him, asking if he'd seen the article about him. It turned out that what Carmack had meant by "show it to some people" was "take a bunch of meetings with the press to promote virtual reality, the Rift, and Luckey himself."
The reception that the Rift got was rapturous. "The level of immersion was unlike any other gaming experience I've ever had," one site wrote. "It transforms the experience of playing a first-person video-game," another wrote. "When we look at that now," Carmack says, "it was clearly the inflection point." Overnight, the Oculus Rift became the most hotly anticipated gaming device since the Microsoft Kinect.
It was time to get serious. Luckey joined forces with an executive team, Iribe among them, and formally established the company -- he was now the founder of Oculus VR Inc. They also upped the ambition of their Kickstarter campaign: they would still send DIY kits to their early backers, but they couldn't expect developers to start building games for a device they had to construct themselves. So they decided to fund a fully assembled product, promising a complete kit to anyone who pledged $300 or more. The campaign video featured some of the most respected people in the gaming industry, such as Cliff Bleszinski, then design director of Epic Games, and Valve head Gabe Newell, singing Oculus's praises. Hours before the campaign went live, Luckey got nervous and lowered the funding threshold from $500,000 to $250,000. Within hours the company blew past both on its way to more than $2.4 million.
Since then, the team has made even further headway on some of VR's most intractable problems. They hired Nirav Patel, an Apple engineer who had been working on a motion tracker that used a gyroscope, accelerometer and magnetometer to sense players' head motion. At Oculus, Patel helped design the brain of the Rift, a tracker that sampled motion data so quickly that Oculus could use algorithms to predict a player's head movements and pre-render images, shaving latency by precious milliseconds. Oculus also switched from LCDs to AMOLED displays, allowing the Rift to reduce latency and motion blur simultaneously. The team used a small external camera to track the headset itself, doing away with fiducial markers. But perhaps the biggest breakthrough wasn't technical at all. In 2013 Carmack decided to leave id Software, where he had worked since cofounding it in 1991, and join the Oculus team as CTO. It was an eyeball-popping PR coup, but it also meant Carmack could dedicate his engineering skills -- the same ones that madeDoom andQuake such historic landmarks -- to improving the Rift.
By mid-October, the momentum was unstoppable. That month Iribe stood up at a gaming conference and announced that the Oculus Rift would be a "no-motion-sickness experience". It was an audacious promise, and one that caught the attention of Brian Cho, a young partner at Andreessen Horowitz,who was sitting in the audience. The VC firm had turned down an earlier opportunity to invest in Oculus's Series A round. After hearing Iribe's announcement, the firm reached out and asked for another demo. Chris Dixon was among the six Andreessen Horowitz partners who got a look at the new model. "I think I've seen five or six computer demos in my life that made me think the world was about to change," he says. "Apple II, Netscape, Google, iPhone… then Oculus. It was that kind of amazing." By December, Oculus had closed Series B funding -- with Andreessen Horowitz leading -- for $75 million.
It's April 3, nine days after Facebook announced its purchase of Oculus. But not much has changed here at the company's HQ in Irvine. Luckey, now 21, still rolls into the office around 11 (after which he'll work a 12-hour day). The common areas are festooned with all things gaming, from framed posters to signed art to oversized Gears of War figurines. The conference rooms are named after pop culture's greatest virtual-reality dreams -- Star Trek: The Next Generation's holodeck, Snow Crash's Metaverse, Ready Player One's Oasis. The open kitchen, while bountiful, skews engineer: breeze-block-sized containers of liquorice and packets of snacking nuts make it clear there's a Costco nearby. Outside, the April morning is as blue and clear as Orange County usually delivers. On the face of things, last week's acquisition has left the workplace largely untouched.
The Facebook deal moved incredibly fast; Zuckerberg first tried on the latest prototype in February. When Luckey heard about his interest, he was sceptical. "It's not the first thing you think," he says. "'Wow! Facebook! That's exactly who I would have imagined to be a good partner!' So they did run the ring of fire a little bit convincing us."
Over the course of many conversations during the next several weeks, though, Zuckerberg won Oculus over. "I had heard many times that Mark is a laser beam, that Facebook is all he thinks about day in and day out," VP of product Nate Mitchell says. "So when I first met with him, I thought he was going to be like, how do we get News Feed into VR?" Instead, the person who showed up was someone Mitchell calls "Visionary Mark Zuckerberg", who saw virtual reality as not just a gaming tool but as a fully fledged communications platform. The Oculus team agreed; they may have started out trying to build a great gaming device, but they realised now that they were sitting on something much more powerful. Zuckerberg seemed to understand that, and he also seemed to understand that it had potential far beyond being an extension of Facebook's existing social-media service. "This isn't about sharing pictures," Luckey says. "This is about being able to share experiences." The deal was consummated over an eight-day stretch in mid-March. Iribe was so excited about the acquisition that he revested 100 per cent of his own equity for a five-year period, guaranteeing that he'd be with the company for the foreseeable future; Luckey, Carmack and others took similar steps.
But not everyone was so optimistic about the partnership. Within minutes of the announcement, the Oculus website was filled with angry comments. (The top one read simply: "DO NOT WANT".) Backers threatened to cancel their pre-orders, to never buy the Rift, to throw their purchasing power behind Sony's Project Morpheus. Some of this was gamer snobbery, rooted in the assumption that Facebook would dumb down the Oculus experience, loading it with targeted ads and 360° 3D versions of FarmVille. Some of it was fear that their gaming device would wither away in the Facebook catacombs, forgotten by a young billionaire mogul with buyer's remorse. And some of it was the fury of backers spurned, people who had ponied up to support the original Kickstarter campaign, only to see their investments made irrelevant by a deep-pocketed corporation.
But the Oculus team argues that, far from threatening the device's immediate future, Facebook is helping to secure it. "Every VR product has been a failure," Luckey says. "Nobody lending money for manufacturing looks at Oculus and says 'I can loan you $250 million!' Because they know the safe bet is we're going to fail, go bankrupt and take hundreds of millions of dollars with us." Now Oculus doesn't have to worry about getting loans at all. And Facebook's backing has helped the company attract a raft of people from top game studios. Within a week of the acquisition announcement, Michael Abrash, the Valve engineer who spearheaded that company's virtual-reality research, became Oculus's chief scientist -- joining colleague Atman Binstock, who'd gone to Oculus earlier in March. Along with a third former Valve engineer, Aaron Nicholls, they are now working at an Oculus research and development lab in the Seattle area.
Facebook's money also means that Oculus doesn't need to worry about turning an immediate profit -- and that will come in handy as it builds its first consumer product. "Let's say we're trying to pack in everything we can for $300," Mitchell says. If the device needs to be profitable, then the company couldn't spend much more than $100 on the hardware itself. But now that it doesn't need to preserve its profit margin, Mitchell says, "you can take all of that margin money, apply it to components and still keep the price exactly the same." In fact, according to Luckey, the consumer version will be "higher quality in every aspect" than the prototype that Valve showed Iribe last year. While Oculus's internal units have used twin AMOLED 1080p displays from Samsung Galaxy S4s, the company no longer has to depend on the mobile-phone ecosystem; it now has the money and the backing to ask a manufacturer to create custom displays specifically for VR applications.
Oculus is also working on a second, outward-facing camera that will be part of the headset itself. The Valve prototype used such a camera to read fiducial markers on the walls for tracking, but Oculus seems to intend it for some very different applications. For one, Carmack says, it can function as a pass-through camera, allowing Rift-wearing users to see what's happening in the real world -- a kind of external heads-up display that would allow you to pick up a drink, for instance. But it has other, much more interesting potential uses. Right now the Rift allows players to look around a virtual world; to move through it, they use an Xbox controller. But a front-facing camera might allow the Rift some day to track users' gestures instead -- much like a Kinect, but more powerful. "In the early days of virtual reality, it was all goggles and gloves," Carmack says. "Nobody's talking about the gloves now -- it's going to be done with optical tracking. You want it to feel like a virtuoso with an instrument." Add haptic feedback, which the company is also developing, and you've taken a giant step toward achieving true presence. Players will be able to engage with virtual worlds -- and have those worlds engage back -- unencumbered.
But what those worlds end up looking like isn't up to Oculus -- it's up to the partners and developers creating the experiences that we'll have within the Rift. And already they're finding that the future of virtual reality might not look like anything we've been led to expect.
The gamer fantasy of VR tends to involve a full-body first-person shooter. But that's not going to happen for a while: photorealistic games of today simply can't be rendered at the frame rate that current VR technology demands. Instead, Carmack says, much as Angry Birds defined iPhone gaming, Oculus's first breakouts will take advantage of the medium's unique properties. And that's an opportunity for developers. "The magic is not in the 6,000-line GPU shader that's going to make a highlight just right," Carmack says, but in designing games that could have run on a less powerful computer: "It's not like good games are only made when you can throw teraflops of performance at them." For now programmers need to concentrate on the simpler aspects -- how motion works, for example -- rather than the crazy pyrotechnics. Otherwise you're just slapping icing on a cake that no one can eat.
That's just one way in which the logic of mainstream gaming may not pertain to the Oculus. For instance, fast-twitch human locomotion -- the kind that Carmack pioneered with Doom -- becomes overwhelming in VR. (Oculus has hit on 1.4 metres per second as the optimal walking rate.) Similarly, some of the most popular games being shared among developers and early adopters are simulators, in which players drive or parachute or roller-coaster through an otherwise static world but don't move themselves. That's a limited approach, but Binstock says that more profound interactions are harder to design, and they risk breaking the illusion of immersive reality that has been so crucial to Oculus's success so far. "Presence is fragile," Binstock says. "It's easy to do things that break the feeling of being somewhere." That could be a dropped frame that interrupts a game experience or an aesthetic flaw, like an object that looks too flat.
Earlier this year Oculus prepared a 42-page best-practices document, enumerating dozens of design guidelines to help developers avoid such pitfalls. "Consider having your interface elements as intuitive and immersive parts of the 3D world," reads one. "Ammo count might be visible on the user's weapon rather than in a floating HUD." In the past, environmentally integrated game design like this was seen as a perk; on the Rift, it's a must.
But, as Zuckerberg predicted, games are just the beginning. VR could change the way we consume media. Early on, Oculus showcased a VR Cinema application that lets users sit in a virtual cinema and watch Man of Steel on a full-size screen. "Last time I was sick with the flu," Carmack says, "I just lay in bed and watched VR movies on the ceiling."
Teleconferencing is also in the works. It's easy to imagine strapping on a Rift and finding yourself across a table from someone who is thousands of kilometres away. Oculus has VR Chat prototypes in the works, and a demo that Epic Games unveiled in March allows two players wearing Rifts to interact with each other's avatars in the same virtual room. "The key," Abrash says, "is generating the cues that tell us we're in a real place in the presence of another person: eye motion, facial expressions, body language, voice."
The list of potential uses goes on. Bring a classroom full of kids inside any museum in the world. Hell, that goes for holidays too. And let's not even talk about the oft-predicted sex simulators. "Hardware, while essential, is just an enabler," Abrash says. "The future of VR lies in the unique experiences that get created in software, and if I knew what those would be, even in broad outline, I would be very happy."
And that, more than anything, points towards the challenges that lie ahead. New experiences are under development right now -- and each one may well require the same ingenuity, the same willingness to forge an entirely new visual language, that Luckey and his team have called on to get the Rift where it is today.
The hardware problems have been solved, the production lines are almost open and the Rift will be here soon. After that it's anybody's guess. "I've written two million lines of code over the past 20 years, and now I'm starting from a blank page," Carmack says. "But the sense that I'm helping build the future right now is palpable."