拍一张照片就能变身“阿凡达”

你的自拍很快就能从2D变成3D了。

Loom.ai是一家旧金山的科技创业公司，该公司致力于使虚拟通讯更加个性化。该平台凭借130万美元的种子资金于本周二正式上线了。只需要一张简单的数码相片，loom.ai就能自动生成一个栩栩如生的3D头像，它可以用于短信、社交媒体、VR游戏甚至是电子商务中（类似于前一阵很是火了一阵的Bitmoji，只不过它是3D版的，而且更加传神）。

这家人工智能创业公司计划将该技术授权给第三方平台，以将其整合到各种新的或现有的应用中。

Loom.ai是由拥有多年从业经验的视觉效果专家马赫什·拉玛萨伯拉玛尼安和基兰·巴特两人创办的。拉玛萨伯拉玛尼安曾在梦工厂担任电影视效主管，《怪兽史莱克》和《马达加斯加3》等动画大片都是出自他的手笔。巴特曾为卢卡斯影业做过《复仇者联盟》和《加勒比海盗》的面部动画设计工作。

直到现在，制作逼真的3D头像都是非常昂贵的，而且基本上只能在好莱坞的工作室里完成。拉玛萨伯拉玛尼安和巴特想改变这种情况。

巴特表示：“我们发现，观众不满足于仅仅坐在那里看电影——他们想要互动。我们认为，你应该把用户带入到那种体验当中。也就是说，我们是怎样把电影明星数字化的，就应该以同样的方法把普通用户也数字化。”

Loom.ai公司今年三月成立后，立即进入了美国知名创业孵化器Y Combinator的法眼——Y Combinator一般会向处于原型产品阶段的创业公司提供创业指导和2万美元的资金，以换取对方1.5%的股权。

Loom.ai能够通过机器学习技术捕捉面部细节。它的软件收集的关于面部运动的数据越多，就越擅长推测人的其它面部表情。

拉玛萨伯拉玛尼安表示：“就算我们不知道所有细节，我们也能推测出关于面部运动规律的很多其他信息。”

这家创业公司背后的投资人主要有丹华资本、Y Combinator、Presence Capital、Anorak Ventures以及GV合伙人乔·克劳斯等。（Anorak Ventures的任事股东格雷格·卡斯尔也是VR公司Oculus的一名种子投资人——目前Oculus已被Facebook收购）。

Loom.ai不愿透露目前有哪些第三方平台正在测试它的API，不过该公司表示，它的目标是像Facebook这样拥有大量用户群的平台。

拉玛萨伯拉玛尼安表示：“Facebook有大量用户群，将来如果他们愿意用我们的技术将2D照片变成3D头像，我们还是很乐意的。”

今年10月，Facebook的CEO马克·扎克伯格贴出了一段视频。从视频中可以看出，Facebook正在研究未来通过Oculus Rift设备让人们通过3D化身在VR世界中实互相聊天。 （财富中文网）

作者：Polina Marinova

译者：朴成奎

Your selfies are about to get a whole lot more realistic.

Loom.ai, a San Francisco-based tech startup that wants to make virtual communication more personal, emerged from stealth mode Tuesday with $1.3 million in seed funding. Using basic digital photos, Loom.ai automatically generates detailed 3D avatars that can be used in text messages, social media, virtual reality games, and e-commerce (think Bitmoji, but 3D and more expressive).

The artificial intelligence startup plans to license the technology to third-party platforms that will incorporate it into new or existing applications.

Loom.ai was founded by seasoned visual effects artists Mahesh Ramasubramanian and Kiran Bhat. Ramasubramanian previously worked at DreamWorks Animation as a visual effects supervisor on movies such as Shrek and Madagascar 3. Bhat worked on facial animation at LucasFilm for movies such as The Avengers and Pirates of the Caribbean.

Until now, creating realistic 3D avatars was an expensive process done exclusively in a Hollywood studio. Ramasubramanian and Bhat want to change that.

“What we found is that the audience no longer wants to sit back and watch a movie — they want to interact,” Bhat says. “We believe that you need to bring the user into the experience, which means figuring out a way to digitize the everyday user just like we used to digitize the movie stars.”

After launching in March, Loom.ai was selected for the Y Combinator fellowship, a remote program that gives prototype-stage startups guidance and $20,000 in funding for 1.5% equity.

Loom.ai relies on machine learning to capture the subtleties of the human face. As the company’s software collects more data about facial movements, it becomes better at guessing and inferring other facial expressions as well.

“Even if we can’t see everything, we can predict a lot of other things about how the face should move,” Ramasubramanian says.

The startup is backed by investors such as Danhua Capital, Y Combinator, Presence Capital, Anorak Ventures, and GV partner Joe Kraus. (Greg Castle, managing partner at Anorak Ventures, was also a seed investor in Oculus, the virtual-reality company now owned by Facebook.)

Loom.ai would not disclose the third-party platforms currently testing its API, but said it’s aiming for platforms like Facebook with a massive user base.

“Facebook has a ton of users, and we would love it if they would use our technology to turn a 2D photograph into a 3D avatar in the future,” Ramasubramanian says.

In October, Facebook CEO Mark Zuckerberg released a video showing how the future of work would involve 3D avatars talking to each other through the Oculus Rift device.