Three years ago, when Facebook bought Oculus for $2 billion, many scratched their heads in befuddlement. Social networks and virtual reality seem like such strange bedfellows; one is about connecting you to the world, while the other appears to do the opposite. But CEO Mark Zuckerberg envisioned a world where VR would be a place for communication, not isolation. And, many years later, that vision is much closer to reality. Facebook Spaces is the company's answer to social VR and it is, as I was amazed to find, surprisingly compelling.
I had the chance to try out Spaces for myself a few hours after it was announced at F8. As soon as I placed the Oculus Rift headset on my head and the Touch controllers were in my hands, I was transported to what appeared like a beautiful park with cherry blossom trees. On my right was my virtual helper, Justin, who appeared in the form of an animated cartoon avatar. In front of me was a tableau of sorts, with a little dashboard in front of me.
Justin told me to select Appearance, and voila, I could customize the appearance of an animated cartoon avatar of myself. You could design one from scratch by customizing individual features like your nose or your hair, but I decided to just have one automatically generated for me. I grabbed one of my profile photos, which were already on display, and Spaces was smart enough to translate it into a cartoon version of me. From there, I also changed the color of my glasses and my shirt, which you can do in the Appearance tab too.
What I noticed almost immediately is how real it seemed, which is really weird considering I was speaking to an animated avatar. Mike Booth, the product manager leading the Spaces development team, says that's because the Rift and the Touch creates a little motion capture studio. "You get your actual body language," he said. "It captures head movements, even hand gesticulation."
What's more, Spaces also infers what your eyes are looking at, creating what appears to be eye contact, which is integral to face-to-face communications. One of the reasons Spaces can do this so well is because these avatars are stylized and cartoon-like. "They're not hyper-realistic, where you can find every little flaw," says Booth.
The same goes with mouth movement. It actually listens to the voice coming through the Rift microphone and then it tries to guess what mouth shapes you're making. It's not always accurate -- it sort of snaps the mouth around like a Wallace & Gromit cartoon -- but that's entirely on purpose. "We're not trying to be super photorealistic," says Booth. "We just want to show that you're talking."