Ever wanted to put yourself inside of a video game? It’s now possible to dynamically represent your entire body inside of a virtual world, and this monumental task can be accomplished with off-the-shelf hardware. With three first-gen Kinects and an Oculus Rift dev kit, a developer by the name of Oliver Kreylos has hacked together a true VR prototype that actually makes you feel like you’re in a computer-generated world.
By placing three Kinect sensors in an equilateral triangle around his room, Kreylos is able to generate a live 3D model of his body from the raw video data. By sidestepping the Kinect’s processor-intensive skeletal reconstruction, this visual representation is extremely low-latency. When you put on your Oculus Rift and hold your hand to your face, Kreylos claims that your brain immediately registers the limb as your own.
Keep in mind, all of this is being done with relatively modest gear. In a post on his blog, the developer explains that this entire demo is running on a Linux box with a 3.5GHz Intel Core i7 CPU, 8GB of RAM, and an Nvidia Geforce GTX 770 GPU. He’s also using the low-res Kinect cameras from the last generation, so imagine how much better this could look with the newest model. With three HD cameras and a few months of polish, this could potentially mean big things for affordable VR training, improved teleconferencing, and immersive games.
It’s easy to shrug off this three-camera rig as an impractical solution, but this is a really solid first step towards solving the problem of full-body VR. The Oculus Rift and Project Morpheus are impressive prototypes, but they’re far from complete packages. If we want full-fledged virtual reality in our future, we’re going to need to account for a lot more than just neck movement in our simulations. We need to think bigger than just head-mounted displays.
Comments
Post a Comment