Valve Explains How It Built the Steam Controller

Discussion in 'Geek Tech Forum' started by News Bot, Nov 4, 2013.

  1. News Bot

    News Bot Your News Bitch

    3,316
    0
    36
    Aug 5, 2013
    Admin Post
    Valve isn’t getting into the hardware business lightly. While some consider the Steam giant’s sudden interest in physical boxes to be out of left field, the company has actually been researching the hardware space for nearly two years, making strategic hires and preparing itself for longer than most consumers realize. With its own OS in the works and the recent news that Steam has 65 million accounts, the house that Gabe Newell built is set to make a big splash when it enters the hardware race sometime next year.
    Following recent announcements that Valve will*work with partners to create Steam Machines and will build its own Steam controller, IGN visited Valve’s headquarters to speak with Greg Coomer, Eric Hope, and Anna Sweet to learn about the process behind how Valve built its debut hardware, plus the problems it still needs to solve moving forward.
    “We began thinking about input in an open-ended, not terribly directed way," Coomer told IGN. "We were thinking a lot about VR and how input needed to evolve in order to accommodate wearable computing and virtual reality. Not very many people were working on it, but it was an interesting area of exploration. We were coming at it just kind of throwing darts at the wall, trying stuff out to see what seemed like it made sense.”
    According to Coomer, the process didn’t begin by looking at something like the Xbox 360 controller and thinking about how to improve it, but instead by thinking about the nature of input and what made the most sense on a fundamental level.

    We like to go out and tell people, ‘PC’s awesome. It’s the home of innovation.’ But in the area of input that just wasn’t really true.

    “We were thinking about the fact that innovation just hadn’t happened in the PC space around input for a really long time,” he explained. “It was happening on other platforms. To us that was weird, because the PC seems like a place where innovation typically happens and is ahead of a lot of other platforms, in hardware and in software. We like to go out and tell people, ‘PC’s awesome. It’s the home of innovation.’ But in the area of input that just wasn’t really true. Another reason that we just started these open-ended experiments was from that high-level notion. It just seemed like it should be happening for PC games. Since Steam was becoming the de facto place where PC gaming happened, it felt like it was sort of incumbent upon us to lead that and to try to do some of that innovation ourselves. None of those are very directed reasons and goals around input. It was more just, ‘hey, let’s try stuff.’”
    And “try stuff” is exactly what Valve did. In the nearly two years Valve has been working with hardware and thinking about the living room experience, dozens of prototype controllers were considered, and to this day the team is still experimenting with wearable computing and all kinds of various other input devices. To start, though, they knew they had to make a gamepad if they wanted to offer Steam users a more focused, pick-up-and-play experience.
    [​IMG]A small sample of Valve’s controller prototypes.

    “We wanted to accommodate the entire Steam catalog and give game players a way to control that catalog by holding something in their hands,” Coomer explained. “And we were focusing that particularly on helping people play games on their TV, in the living room. It was that clarity that really made us focus our efforts, not on open-ended experiments and ‘what could we do if the sky’s the limit?’ but instead, ‘let’s do an inventory of what is required for playing all Steam games, even if you’re doing it from the sofa.’ In the den, you have about 104 keys on a standard keyboard, and you have a massively high-resolution 2D input surface. That’s pretty awesome for PC gaming, but it’s a lot to try to replicate or bring over to a space where you don’t have those devices and you’re holding everything in your hand. But it did help us focus our design efforts.”

    We wanted to accommodate the entire Steam catalog and give game players a way to control that catalog by holding something in their hands.

    With that focus in mind, Valve began the long process of figuring out exactly what a controller needed to offer. At first, that meant adding a trackball to an otherwise traditional controller alongside a standard analog stick, which initially seemed promising.
    “It did a lot of what we were looking for,” Coomer said of the physical trackball. “It let you push a mouse cursor around a screen. It let you aim more precisely than a thumb stick let you aim in first-person shooter games. It was rough, but it was a step in a pretty interesting direction for us. It felt like it was a successful first prototype.”
    From there, Valve experimented with trackballs of different sizes, and even tried embedding a long, thin touch screen at the bottom of one prototype that let players swipe through pages of virtual keys to perform various commands. “Paging through things wouldn’t allow as quick of access as you would have on your desk with a physical keyboard, but it would go a long way toward providing at least accessibility for playing games that wouldn’t otherwise be playable with just a handheld device,” Coomer explained.
    [​IMG]An early design that included a physical trackball and traditional analog stick.

    Valve looked closely at traditional gamepads and decided that the small overall percentage of Steam games that support that type of input are already served well by existing devices like the Xbox 360 controller. Instead, the team wanted to focus on having all Steam games comfortably playable with the new controller.

    We’ve never shipped hardware or written firmware ourselves, but we had a lot of confidence that we could attack that problem.

    “It wasn’t so much about, ‘let’s build something better than the competition.’ It was a completely different design proposition, where our goal was around, ‘let’s serve this whole catalog and make something that enough Steam gamers can make use of, and let’s improve upon the capability of a traditional game pad in many genres of games, like first-person shooters,’" Coomer said. “’Let’s not just meet that bar. Let’s build something that can exceed it by quite a ways, especially in the fidelity of pointing and aiming.’ That was always a second part of our goal. The whole catalog, and then that improvement.”
    Valve soon found that by simulating a trackball rather than using a physical one, it had more control over the experience. “This kind of device is one that is more controlled by software than by physics,” Coomer said, holding up a controller more reminiscent of the final product. “We could modify the kind of experience that’s possible when you’re touching this sort of surface, much more than we felt like we could modify the experience that’s possible when you’re touching [a trackball]. By making this more of a software problem, we felt like it was far closer to home for us, and far more likely that we were going to build the right kind of solution for customers. We know how to build software. We’ve never shipped hardware or written firmware ourselves, but we had a lot of confidence that we could attack that problem and start making it better.”
    [​IMG]A mock-up of Valve’s touch-only prototype.

    Once the team realized that a virtual trackball run by firmware worked better than a physical one, it decided to replace the traditional analog stick with a second trackball instead. Valve eventually brought back a new version of the narrow touch-strip it experimented with earlier, and at one point considered making the controller a touch-only device altogether.
    “We were sort of giddy at this point. We said, ‘touch is fantastic. Let’s throw out all buttons and make a touch-only controller,’” Coomer remembers. “’Maybe even touch surfaces on the back. Every button is virtualized. Let’s have all 104 keys.’ We were kind of entranced by it at first. But that was short-lived. There was even the notion of a phone. Not really a phone, but a Steam communication core that we were going to put in different input devices. We built some functional prototypes of that, not just mockettes like this, and tested it out. It was pretty cool, what we thought we could accomplish, but it felt pretty quickly like we’d gone off the reservation a little bit. We’re focusing now on accomplishing different goals. This felt like a future controller that might serve a new generation of games that was meant to target this kind of device while you’re looking at a TV. It’s not really what we were after. So we reined ourselves in and we came back to something that was much more like [the final controller].”

    We said, ‘touch is fantastic. Let’s throw out all buttons and make a touch-only controller.

    With Valve’s final design, ergonomics were a huge priority. The controller is angled to provide a natural hand position in order to prevent your hands from getting tired, and the controller is able to provide haptic feedback to make sure you always feel like you’re in control.
    “Early, when we were doing the trackball experiments, we quickly realized how valuable physical feedback was,” Hope explained. “There’s this data channel through the tips of your fingers that your brain immediately taps into. It’s really valuable and interesting. When we went to the track pads, which were more scalable and mutable – they could become anything – we lost that physicality. And so then we started looking at ways to bring that physicality back. We ultimately ended up with linear resonant actuators. And so they’re like small solenoid pistons that can deploy at various frequencies and intensities. That let us simulate lots of different textures or zones. It lets you know if you’re touching a virtual button or crossing a virtual zone. We got a lot of that physical haptic feedback back once we employed those in the track pads.”
    [​IMG]A mock-up of Valve’s final controller design.

    Hope elaborated on Valve’s previous claims that the trackpads can also operate as speakers, explaining that while a trackpad “doesn’t necessarily make for the most ideal speaker surface,” it is able to emit a small amount of sound. “It’s better than a piezo kind of effect,” he said. “If you were to play a human voice it would be somewhat distorted. It’d sound more like GLaDOS than a normal person. But we think it’ll be interesting when partners start to play with the direct API and they experiment with it.”

    If we weren’t [open to feedback], it would kind of be a facade to open up the beta.

    Valve is excited to get that API to partners, which will allow developers to provide feedback and make changes, but it’s also excited to get its prototype into the hands of Steam users. Roughly 300 units will be sent out as part of Valve’s hardware beta, and, just like Valve’s Linux-based OS will be open source, the idea of the beta is to allow users to contribute feedback and ideas that will essentially make the development of future hardware “just as much a collaboration as an open source effort really is,” according to Coomer.
    Coomer and Hope explained that Valve is “absolutely” open to feedback from the beta, with Hope explaining that “if we weren’t, it would kind of be a facade to open up the beta.”
    “If we find out that there’s a necessary change that means a radical redesign of the controller, we’re going to make that decision just the same way that we have with all of our other products,” Coomer said. “We may find that out. We feel confident that we’ve done enough internal testing to know that we’ve got a good platform and that we’re probably going to be enabling enough good experiences. People aren’t going to want to just throw it out the window entirely. But we think we’re going to learn a lot. We’re super open to that.”
    [​IMG]The pre-touch screen controller that will be included in Valve’s hardware beta.

    As for Valve’s ambitions for the future of hardware, Coomer explained that the team isn’t necessarily looking to win over casual gamers who don’t own a PC, but instead is catering to Steam users first.
    “The focus of most of this work has really been on bringing value to Steam customers,” he said. “Even if we’re only serving a fraction of them, we feel like we’ll be very successful if that fraction is having a great experience in the living room. That number will probably grow over time. We’re not even trying to push our existing users toward the living room or the TV if they don’t want to have that experience. All those customers are currently pretty happy doing what they’re doing in the den with their PCs or laptops. This is just an extra avenue for them, if they want to sit on the sofa like a lot of them do, I think. Then it’s great to have this other option through which you can access Steam. It’s not an attempt to go very far, or really at all at first, beyond our initiated customer base. We’re going to learn a bunch from the people who already value Steam really highly. We don’t feel like we have to jump exponentially outside that group just to be successful in that realm.”

    The focus of most of this work has really been on bringing value to Steam customers.

    “Usually, when a platform like this gets brought out, it’s a very different working method and proposition to customers,” he continued. “It looks more like a team that’s much larger than us has worked at perfecting something and finishing it, and then reducing risk as much possible and locking down that design, making it ready for a massive initial manufacturing push, spending billions of dollars on marketing. Very different from what we’re trying to do. We don’t have to be so risk-averse. We intentionally are operating this way because we think it will result in a much better product, in the short term and the long term, to be public about this, and to have it iterated with us and with partners and with users. But it lets us start small and grow over time.”
    Hope explained that while Valve’s controller is obviously designed to work natively with SteamOS, it can also be detected as a traditional controller by a PC running Steam but would have extremely limited functionality. “Steam constantly updates the firmware that is running in this device,” he said. “Even game by game, and eventually moment by moment, it’s going to be updating the code that’s running in the firmware. For that reason and a few other reasons, like updating configs and so forth, Steam is a required component wherever it runs, PC or Mac. Without Steam, if you plug this into one of those devices and run some game that’s not part of Steam, you would only get the most basic functionality. It wouldn’t actually be very valuable as a controlling device.”
    “It would basically be a mouse with 16 buttons,” Coomer added.
    [​IMG]A render of Valve's final designs for its controller and a Steam Machine.

    In Legacy Mode, Valve will allow users to assign specific bindings to the Steam Controller to allow it to work with games that weren’t originally design with Valve’s device in mind. “There’s lots of ways that the pads can be configured,” Coomer said. “One of the ways that we most recently started playing with, and are actually enjoying is where both pads are mouse input, because you don’t really need [the trackpad] to be an analog stick or WASD or whatever. We do some interesting things where we take the values of both pads and we average and multiply and divide based on what we think you would want, because if we just took the velocity of both at the same time, it would probably be too much.”

    You can play Dota with the controller; you just can’t win.

    Any key on a keyboard or any action available via a mouse can be assigned to any button on the controller. More importantly, friends will be able to share bindings on Steam, or pick from bindings that have already been voted up by the community. Bindings can also be created by developers, and Valve will eventually allow users to sort by the newest or most popular.
    Valve will also allow users to adjust things like sensitivity, which can be adjusted on a per-game basis in SteamOS. “On SteamOS, sensitivity is really easy to solve, because we can just change what the operating system’s sensitivity is on the fly, based on what you’re doing in a game,” Hope said. However, in Legacy mode or in Windows, sensitivity won’t be able to be adjusted. “Windows has its own sensitivity settings for input. The game has its own sensitivity settings, and our controller has its own sensitivity settings. We have to find this sweet spot where we’re averaging all three inputs into this spectrum,” he said. “It’s mostly solvable. There hasn’t been any game on Windows that we haven’t been able to play. But it takes a little bit more tweaking.”
    [​IMG]The internals of Valve's Steam Machine prototype.

    So far, Valve has been able to successfully play (and win) games in every genre -- except for one.
    “Somewhat ironically, our flagship at the moment that we spend a lot of time thinking about is Dota 2,” Coomer said. “You can definitely play Dota 2 with this controller, but we wouldn’t say to anybody who’s a serious DOTA player that this is an acceptable replacement for a mouse and keyboard for that game.”
    “We have an internal joke. You can play Dota with the controller; you just can’t win,” Hope laughed. “Unless you’re playing against everybody else who has a controller, and then it’s a tractable problem.”

    The entire catalog, we still think we can call it playable.

    “There are definitely games that, although playable, aren’t good enough for the existing players to spend much time playing them with this controller,” Coomer explained. “There aren’t very many games like that on Steam, though. So when we say we’ve managed to accommodate the whole catalog, there are some places where, at the highest level, that’s not complete, because of things like Dota. But the entire catalog, we still think we can call it playable. There just aren’t that many in that category.” Hope said Valve has also “been doing a lot of experiments” regarding fighting games and will have more to reveal soon.
    Valve is also working with developers to make sure that they can successfully port their games to Linux in order to run in SteamOS.
    “That’s something we’ve been talking to them about for a couple of years and getting them ready for what we’re thinking and where we’ve been thinking about heading,” Sweet said. “In general, they’re all pretty excited. They like Steam for most of the same reasons we like Steam. It gets them really close to their customers. They’re able to update their game as much as they want. All of those pieces that they love about the platform. Any time there’s a new platform to develop for, people think through how much work it’s going to be, how much time it will take. They’re certainly going through that analysis.”
    “It really comes down to the type of game and the tech they’re using,” she continued. “If they’re using Unity, it’s totally done. I think we’re benefiting from the fact that there are a lot of platforms right now, because a lot of people have already built for Mac or built for Android, and that is actually a lot of the work of switching from DirectX to OpenGL. We’ve also done a lot of work as we’ve moved our own games over to Linux. We’ve learned a lot about building for it. We’re working on tools to help make partners’ lives easier as they’re doing that. We see our partners as pretty excited, working toward building out their games.”
    [​IMG]A powered-on (and unpainted) version of Valve's Steam Machine prototype.

    Looking ahead, Valve is confident that players will like the controller, but the team is willing to admit that players will have to learn to adjust. “It does take some getting used to. We’ve noticed that it takes about 15 to 30 minutes depending on whether you’ve spent most of your time on a keyboard and mouse or whether you’ve spend most of your time on a game pad with sticks,” Hope said. “The sticks are a tougher transition, because you’re used to gripping the controller really hard and pushing really hard with your thumbs. You have to lighten up your grip and get up on your thumbs.”

    It does take some getting used to. We’ve noticed that it takes about 15 to 30 minutes.

    Valve is also working on the wireless protocols the controller will use, with Coomer telling us, “we’re pretty settled on the category of protocol. There’s a Nordic wireless chipset that we’ve been starting to use. It’s like Bluetooth, but it’s a custom stack that has lower latency than Bluetooth. The boxes do need to be able to speak to our controller through that wireless channel, so we’re working with all our partners to be able to make sure that they do.”
    “One of the interesting things about going that route with Bluetooth is that we can use the low-latency Bluetooth channel when it’s available, and we can also flip over to the generic Bluetooth, vanilla Bluetooth,” Hope added. “Any device that has that kind of Bluetooth can still talk to the controller.”
    “Just with a latency penalty, a small one,” Coomer clarified. “Part of why that’s important is because you can use this controller with a PC. It doesn’t have to be a Steam machine box. So with other PCs, you could just communicate via regular Bluetooth and have it be a tiny bit slower.” Every controller shipped separately will also include a dongle “so that you can get the high speed Bluetooth protocol,” Hope said.
    Valve’s controller will ship to users selected for the hardware beta in 2014, and specific hardware partners and Steam designs will be announced at CES in January. For much more on Valve’s plans, be sure to read why it won’t create any SteamOS-exclusive games, plus hands-on impressions from developers.
    Andrew Goldfarb is IGN’s news editor. Keep up with pictures of the latest food he’s been eating by following @garfep on Twitter or garfep on IGN.


    More...
     
  2. DM OUTL4W

    DM OUTL4W Administrator

    6,495
    26
    48
    Feb 17, 2007
    OneDrunkOUTL4W
    DM_OUTL4W
    Admin Post
    looks like a re-skinned xbox one....