Digital humans that look just like us | Doug Roble

翻译人员: shasha zhu 校对人员: Wendy Wu

Hello.

大家好。

I'm not a real person.

我不是一个真人。

I'm actually a copy of a real person.

我实际上是一个真人的复制版本。

Although, I feel like a real person.

但我感觉和真人无异。

It's kind of hard to explain.

这样很难说明白。

Hold on -- I think I saw a real person ... there's one.

稍等， 我看见了一个真人……在这儿。

Let's bring him onstage.

欢迎他来到台前。

Hello.

大家好。

(Applause)

（掌声）

What you see up there is a digital human.

大家刚才看到的是一个数字化的人。

I'm wearing an inertial motion capture suit

我穿着一套能够捕获惯性运动的装备，

that's figuring what my body is doing.

它能够分析出我身体当前的动作。

And I've got a single camera here that's watching my face

这里有一个对准我脸的摄像头，

and feeding some machine-learning software that's taking my expressions,

它不断的把我的面部表情 输入进一个机器学习的软件里，

like, "Hm, hm, hm,"

像这样 ，“嗯，嗯，嗯，”，

and transferring it to that guy.

再把这些表情传输给刚才那个复制品。

We call him "DigiDoug."

我们叫他“数码道格”。

He's actually a 3-D character that I'm controlling live in real time.

他实际上是一个三维人物， 而我可以实时控制他。

So, I work in visual effects.

我的工作方向是视觉效果。

And in visual effects,

在视觉效果这个领域，

one of the hardest things to do is to create believable, digital humans

一件最难的事情就是 制造出一个让人信服的，

that the audience accepts as real.

难辨真假的数字人。

People are just really good at recognizing other people.

人们通常都很擅长辨认人脸。

Go figure!

那就去辨认吧！

So, that's OK, we like a challenge.

我们喜欢面对挑战。

在过去的15年里，

Over the last 15 years,

we've been putting humans and creatures into film

我们一直将人和动物角色 放在电影中，

that you accept as real.

你会认为它们是真实存在的。

If they're happy, you should feel happy.

你会因他们喜而喜，

And if they feel pain, you should empathize with them.

也会因他们的遭遇而心生同情。

We're getting pretty good at it, too.

我们也已经做得很不错了。

But it's really, really difficult.

但这其实很不简单。

Effects like these take thousands of hours

达到这样的效果， 得花费成千上万个小时，

and hundreds of really talented artists.

需要数百个天赋异禀的艺术家。

But things have changed.

但是，时过境迁。

Over the last five years,

在过去的五年，

computers and graphics cards have gotten seriously fast.

电脑和显卡技术发展迅速。

And machine learning, deep learning, has happened.

机器学习和深度学习技术也日趋成熟。

So we asked ourselves:

我们开始自问：

Do you suppose we could create a photo-realistic human,

通过真人面部表情和细节

like we're doing for film,

控制对应的数码人，

but where you're seeing the actual emotions and the details

我们是否能够制作出 像我们一直在电影中做的

of the person who's controlling the digital human

那样逼真的人物，

in real time?

但是是实时产生的？

In fact, that's our goal:

事实上，这就是我们的目标：

If you were having a conversation with DigiDoug

如果让你和“数码道格”

one-on-one,

一对一谈话，

is it real enough so that you could tell whether or not I was lying to you?

他能够真实到 让你能判断我是否在说谎吗？

So that was our goal.

这就是我们的目标。

About a year and a half ago, we set off to achieve this goal.

一年半以前，我们设置了这个目标。

What I'm going to do now is take you basically on a little bit of a journey

现在就给大家简单展示一下 这一路走来，

to see exactly what we had to do to get where we are.

为了达到这个目标， 我们做了怎样的努力。

We had to capture an enormous amount of data.

我们必须获取大量的数据。

In fact, by the end of this thing,

毫不夸张的说，到这个项目结束为止，

we had probably one of the largest facial data sets on the planet.

有关我脸部的数据集可能是

Of my face.

地球上最大的数据集之一了。

(Laughter)

（笑声）

Why me?

为什么选我的脸啊？

Well, I'll do just about anything for science.

因为，我愿意为科学献身啊。

I mean, look at me!

看看我这身！

I mean, come on.

仔细看看。

We had to first figure out what my face actually looked like.

首先得分析出我的脸长什么样子。

Not just a photograph or a 3-D scan,

并不仅仅是一张照片或者是三维扫描，

but what it actually looked like in any photograph,

得分析出它在任何镜头之下呈现的真实面貌，

how light interacts with my skin.

光影如何和我的皮肤互动。

Luckily for us, about three blocks away from our Los Angeles studio

幸运的是，在我们位于洛杉矶的 工作室三个街道之外，

有个叫做ICT的地方。

is this place called ICT.

They're a research lab

ICT是个研究实验室，

它附属于南加州大学。

that's associated with the University of Southern California.

They have a device there, it's called the "light stage."

它里面有个叫做“光平台”的设备。

It has a zillion individually controlled lights

这个设备有超级多个单独控制的灯

and a whole bunch of cameras.

和很多的摄像头。

And with that, we can reconstruct my face under a myriad of lighting conditions.

在这个设备的帮助下，我们可以在 很多种不同光影变化中，重建我的面部。

We even captured the blood flow

我们甚至捕捉到了血液的流通，

and how my face changes when I make expressions.

以及在我做表情时， 我的面部是如何变化的。

This let us build a model of my face that, quite frankly, is just amazing.

说真的，这样制造出的 面部模型真是非常棒。

It's got an unfortunate level of detail, unfortunately.

虽然它呈现了不该呈现的细节， 太让人难堪了。

(Laughter)

（笑）

You can see every pore, every wrinkle.

你能清晰的看见每个毛孔，每条皱纹。

But we had to have that.

但是呢，我们必须有这些东西。

Reality is all about detail.

真实取决于细节。

And without it, you miss it.

没有细节，你的工作就存在缺陷。

We are far from done, though.

但是我们的工作远远不止这些。

This let us build a model of my face that looked like me.

这仅仅是制作了一个 像我面部的模型罢了。

But it didn't really move like me.

它还不能像我一样动。

And that's where machine learning comes in.

这时候，就要用到机器学习了。

And machine learning needs a ton of data.

机器学习需要很多很多的数据。

So I sat down in front of some high-resolution motion-capturing device.

所以呢，我得坐到一个 高分辨率的动态捕捉仪器前。

And also, we did this traditional motion capture with markers.

我们也使用传统的 动作捕捉方法进行标记。

We created a whole bunch of images of my face

我们制作了很多我的面部图片

and moving point clouds that represented that shapes of my face.

以及能呈现我面部轮廓的动态点云。

Man, I made a lot of expressions,

啊，我做了太多表情了，

I said different lines in different emotional states ...

在各种不同的心情下 说了许多不同的台词……

We had to do a lot of capture with this.

我们必须把这些都捕捉下来。

Once we had this enormous amount of data,

一旦我们有了这些巨量的数据，

we built and trained deep neural networks.

就可以开始建造和训练深度神经网络了。

And when we were finished with that,

当我们做完这些，

in 16 milliseconds,

在16微秒内，

the neural network can look at my image

神经网络就可以根据我的图像，

and figure out everything about my face.

分析出我脸上的所有细节。

It can compute my expression, my wrinkles, my blood flow --

它能计算出我的表情， 皱纹，血液流动——

even how my eyelashes move.

甚至于我的睫毛是怎么动的。

This is then rendered and displayed up there

所以之前我们捕捉的细节，

with all the detail that we captured previously.

都会被渲染和呈现出来。

We're far from done.

但这还远远没完成。

This is very much a work in progress.

我们还要再接再厉。

This is actually the first time we've shown it outside of our company.

这是我们第一次在 公司之外展示这一成果。

And, you know, it doesn't look as convincing as we want;

这还没有达到我们想要的效果；

I've got wires coming out of the back of me,

在我背后还可以看见数据线，

and there's a sixth-of-a-second delay

从捕获录影到显示屏展示

between when we capture the video and we display it up there.

这之间还有六分之一秒的延时。

Sixth of a second -- that's crazy good!

六分之一秒啊，这相当厉害了！

But it's still why you're hearing a bit of an echo and stuff.

但是你还是会听到一点回音。

And you know, this machine learning stuff is brand-new to us,

机器学习对我们而言， 是个全新的东西，

有时候很难确保 它做出了正确的判断，

sometimes it's hard to convince to do the right thing, you know?

It goes a little sideways.

它有的时候过于夸张。

(Laughter)

（笑）

But why did we do this?

但是为什么我们要这么做呢？

Well, there's two reasons, really.

实话说，有两个原因。

First of all, it is just crazy cool.

首先，这真是太酷了。

(Laughter)

（笑）

How cool is it?

有多酷呢？

Well, with the push of a button,

按一下按钮，

我就能以完全不同的 人物形象来做这个演说。

I can deliver this talk as a completely different character.

This is Elbor.

这个是阿波尔。

We put him together to test how this would work

我们用他来检测这套系统是如何

with a different appearance.

用另一副形象来工作的。

And the cool thing about this technology is that, while I've changed my character,

这个艺术很炫酷的地方就在于， 虽然呈现的形象变了，

the performance is still all me.

但是演说的人仍然是我。

I tend to talk out of the right side of my mouth;

我习惯用偏右边的嘴巴讲话；

so does Elbor.

阿波尔和我一样。

(Laughter)

（笑）

Now, the second reason we did this, and you can imagine,

第二个原因，你也许能想到，

is this is going to be great for film.

这对电影制作意义非凡。

This is a brand-new, exciting tool

对于艺术家，导演以及说书者，

for artists and directors and storytellers.

它是一个崭新的，令人振奋的工具。

It's pretty obvious, right?

这很明显，对吧？

I mean, this is going to be really neat to have.

一定会让人爱不释手。

But also, now that we've built it,

现在呢，我们已经实现了，

it's clear that this is going to go way beyond film.

很明显，它的应用远远不止在电影界。

But wait.

等一下，

Didn't I just change my identity with the push of a button?

我刚才不是按一下按钮 就改变了我的形象和身份吗？

Isn't this like "deepfake" and face-swapping

这是不是很像我们之前听说过的

that you guys may have heard of?

“deepfake”和换脸呢？

Well, yeah.

对。

In fact, we are using some of the same technology

我们其实应用了一些 和deepfake

that deepfake is using.

同样的技术。

Deepfake is 2-D and image based, while ours is full 3-D

Deepfake是基于二维和平面图像的， 而我们这个是三维的，

and way more powerful.

更强大。

But they're very related.

但二者非常相关。

And now I can hear you thinking,

我猜你一定在想：

"Darn it!

“天啊！

I though I could at least trust and believe in video.

我以为我至少可以信任视频的真实性。

If it was live video, didn't it have to be true?"

实时视频，不应该都是真的吗？”

Well, we know that's not really the case, right?

可是，我们都知道， 事实不是这样的，对吧？

Even without this, there are simple tricks that you can do with video

哪怕没有数码人， 你在视频上运用一下小技巧，

like how you frame a shot

比方说你可以通过布局镜头

that can make it really misrepresent what's actually going on.

去掩盖真正发生的事。

And I've been working in visual effects for a long time,

我在视觉效果研究上花了很长时间，

and I've known for a long time

很久前我们就知道了，

that with enough effort, we can fool anyone about anything.

只要做出足够的努力， 在任何事物上，我们都可以愚弄人。

What this stuff and deepfake is doing

Deepfake和我们研发的工具

is making it easier and more accessible to manipulate video,

可以让视频处理更简单，更容易，

just like Photoshop did for manipulating images, some time ago.

就像不久以前用PS工具处理图像一样。

I prefer to think about

我更倾向于思考

how this technology could bring humanity to other technology

这项技术会怎样让其他技术 变得更加人性化，

and bring us all closer together.

让人与人走得更近。

Now that you've seen this,

现在你们也看到了，

think about the possibilities.

想一想各种可能性吧。

Right off the bat, you're going to see it in live events and concerts, like this.

马上，你就会在演唱会 和直播活动中看到它，就像这样。

Digital celebrities, especially with new projection technology,

特别是随着新的 投影技术的发展，数码化的名人

are going to be just like the movies, but alive and in real time.

将会和电影一样， 但是更鲜活，而且是实时的。

And new forms of communication are coming.

沟通交流的新形式要来临了。

You can already interact with DigiDoug in VR.

你已经能够在虚拟现实中 和 数码道格 交流了。

And it is eye-opening.

这让人大开眼界。

It's just like you and I are in the same room,

就像你和我身处同一个房间一样，

even though we may be miles apart.

即使你我相隔很远。

Heck, the next time you make a video call,

下次你和别人视频通话的时候，

you will be able to choose the version of you

你就可以选择自己的版本，

you want people to see.

选择你想让对方见到的样子。

It's like really, really good makeup.

这就像你化了个很好的妆一样。

I was scanned about a year and a half ago.

我大约一年半以前扫描了我的脸。

I've aged.

我现在比那会儿老了。

DigiDoug hasn't.

但是数码道格却没有。

On video calls, I never have to grow old.

在视频通话中，我一点都不会变老。

And as you can imagine, this is going to be used

你能想象得到，这还可以用来

to give virtual assistants a body and a face.

给虚拟助手提供一个身体和脸。

A humanity.

让它更像一个人。

I already love it that when I talk to virtual assistants,

当我与虚拟助手说话时，我喜欢

they answer back in a soothing, humanlike voice.

它们以令人平静的、 与人类相似的声音来回应。

Now they'll have a face.

现在他们还将会有一张人类的脸。

And you'll get all the nonverbal cues that make communication so much easier.

而且你还能看到让沟通 更方便的非语言的线索。

It's going to be really nice.

效果一定很不错。

You'll be able to tell when a virtual assistant is busy or confused

你可以判断出，虚拟助手是太忙了， 还是迷茫、没听懂指令，

or concerned about something.

亦或是它们在担心什么。

Now, I couldn't leave the stage

不过，不给你们看看我的真容，

without you actually being able to see my real face,

让你们比较一下，

so you can do some comparison.

我是不能下台的。

So let me take off my helmet here.

让我取下头盔。

Yeah, don't worry, it looks way worse than it feels.

别担心，它只是看起来有点糟糕而已。

(Laughter)

（笑）

So this is where we are.

就这样了。

Let me put this back on here.

我再带上头盔。

(Laughter)

（笑）

Doink!

咚！

So this is where we are.

这就是我们目前的进展。

We're on the cusp of being able to interact with digital humans

我们很快就能和数码人互动了，

that are strikingly real,

他们看起来与真人无异，

whether they're being controlled by a person or a machine.

无论是被机器控制或被真人操控。

And like all new technology these days,

和现在所有的新技术一样，

it's going to come with some serious and real concerns

它也会带来一些严峻且真实的顾虑，

that we have to deal with.

这是我们必须去处理的。

But I am just so really excited

但我真的很期待，

about the ability to bring something that I've seen only in science fiction

可以有能力把在我有生之年曾经只在

for my entire life

科幻小说中看到的事

into reality.

变成现实。

Communicating with computers will be like talking to a friend.

和电脑说话，就跟和朋友说话一样。

And talking to faraway friends

和远方的朋友说话

will be like sitting with them together in the same room.

就像坐在同一个房间说话一样。

Thank you very much.

非常感谢。

(Applause)

（掌声）