HTC Vive Pro Eye delivers eye-tracking, while Vive Cosmos is untethered

HTC is already the lead dog in the VR world, and it continued its effort to separate from the pack by announcing two new headsets at CES 2019. One delivers eye tracking, and the other is an untethered device with inside-out tracking.   

HTC Vive Pro Eye

This headset is essentially the same as the HTC Vive Pro that's already on the market, but with the secret sauce of Tobii eye tracking. Eye tracking has long been one of the holy grails for VR headsets, and Tobii is an ideal partner for HTC: Last year at CES, Tobii's VR HMD eye-tracking demo was possibly the best of its kind we've ever seen. 

Eye tracking has a couple benefits that can dramatically improve VR experiences. First is that this allows for foveated rendering, which is a technique that delivers the highest graphical details only where your eye is focused. This follows how your eyes actually work: You can see fine details only front and center of what you're looking at; the further out you get in your peripheral vision, the less detail you perceive. 

Applied to a VR context, then, if an image is fully and completely rendered, precious GPU resources are wasted because you literally can't see most of those details at any given time. But if your eyes are being tracked, the system knows where the detail needs to be highest at all time, and it intelligently and dynamically adapts. It's a simple but effective way to reduce graphical demands in VR, made possible by eye tracking.

Eye tracking can also be applied to menu navigation and more natural gaming. HTC suggests, for example, that eye tracking could be used to automate shooting, so that you merely look at your enemies without having to target them with a controller. Whether that's any fun or not will be seen, but simply looking at things to win doesn't sound like much of a "game." Menu selection, done by looking at the item you want to choose, is a perfectly plausible application for the technology. 

The "Pro" in the name really does mean "professional," because this new headset isn't designed for gaming. This piece of kit is for the workers of the world, and it's used for actions like selling cars through virtual experiences, training employees, and more. At this point, HTC says that the enterprise and professional market is a "healthy place for us to go," so more of this kind of partnership is likely. 

We don't yet know how much the Vive Pro Eye will cost, but given that the earlier Vive Pro currently retails for $799, we expect it to be more than that.

The Vive Cosmos

The other new device in the Vive world is the Cosmos, an untethered headset that also boasts inside-out tracking. That means it's a complete system unto itself and doesn't need to be, well, tethered to a PC to operate, and it doesn't need external trackers to know its location and angle in an environment. It can, though, track the two Vive controllers you hold in your hands. 

The Vive Cosmos is also apparently designed with comfort first and foremost, with breathable fabric around the crown of the headset and an effort to achieve optimal weight distribution. You can flip up the front of the Vive Cosmos if you need to quickly look down at something in the real world.

Otherwise, HTC is keeping the details of this device under wraps. Nobody at CES could try it, no specs were released, and there's no price. 




Comments closed
    • Chrispy_
    • 9 months ago

    IMO inside-out tracking is the future but they still have a slight tendency to drift.

    They shouldn’t, since mine could easily be using the LED lights on the keyboard and monitor as static reference points, but it doesn’t. In saying that, it could just be awful software from Samsung/Microsoft to blame, so I’ll be keeping an eye on the performance of the Cosmos, purely from a drift perspective.

    It’s funny how despite having such massive headstarts, both HTC and Oculus have failed to really provide any meaningful 2nd-gen products. That mantle has fallen to relative newcomers to the scene like Samsung and Pimax.

      • psuedonymous
      • 9 months ago

      Markerless inside-out tracking [i<]to a sufficient performance level acceptable for VR[/i<] is an extremely hard problem. It's tempting to see markerless SLAM demos from a decade or two ago running on a toaster and thinking "gee, it's so easy! VR inside-out markerless tracking must be just around the corner!", but like with eye-tracking (easy to do with a webcam, well beyond even high-end CotS systems to do acceptably for foveated rendering in a HMD) actually getting the performance up to adequate levels is much deceptively difficult. [quote<]It's funny how despite having such massive headstarts, both HTC and Oculus have failed to really provide any meaningful 2nd-gen products. That mantle has fallen to relative newcomers to the scene like Samsung and Pimax.[/quote<] Neither of which have produced 2nd gen products either. Gen '1.1' at most, just doing the exact same things as current HMDs but cramming newer panels in (and in the case of PiMax, larger singlet lenses they are unable to correct for the distortion of). No advances in haptics, optics, variable focus (or lightfield display), eye tracking (for lens calibration or gaze target selection, let alone foveated rendering), full-body markerless skeletal tracking*, etc. HTC are reliant on Valve for R&D, and Valve are content to noodle about at their own pace. *Partial credit for Oculus Quest, with more than one Quest in the environment they can perform full-body tracking of other people in the environment, and limited body tracking of the wearer in a solo environment, but it's a dev feature rather than one aimed for general usage.

        • DavidC1
        • 9 months ago

        The problem is people are expecting everything to advance as fast as devices that depend on Moore’s Law, and that only applies to processors(of any form).

        Most of the things that are asked of in a future VR headset like improved lenses, better comfort, lower weight, and haptics are things where the pace of improvement is much slower, and sometimes don’t even get better over time. If you are lucky, and the developers do everything right, then we might see a headset that is as a whole, very good.

        Another thing Moore’s Law did well was exponential reduction of cost per compute power. You will not see such things happen here. Maybe in science fiction movies.

        Don’t expect to see something that’s a vast improvement in all areas over what we have today. You will just set yourself up for disappointment otherwise.

          • psuedonymous
          • 9 months ago

          On top of that, the ‘early’ VR development cycle (that is, from the pre-DK1 duct tape prototype series in ~2011/2012 to today) benefited from the fruits of massive Smartphone R&D funds in development of displays, MEMS IMUs, and ICs. Today, it looks like HMDs will be the driver for further increases in panel density (the “4K smartphone” never materialised beyond one flop, which had an unsuitable LCD panel), for increases in SoC GPU performance, increases in IMU quality, advances in compact high-power optics, etc.
          That makes a significant step change from a few years of technology-reuse driven by tens of billions in R&D powered by mass-market sales, to technology development driven by a few hundred million (maybe one or possibly even two single-digit billions) in R&D with the (reasonable) assumption of future mass-market sales.

Pin It on Pinterest

Share This