TED2009
Ed Ulbrich: How Benjamin Button got his face
Ed Ulbrich揭露班傑明巴頓做臉過程
Filmed:
Readability: 4.4
1,080,448 views
Digital Domain團隊的數位特效大師Ed Ulbrich，介紹他的團隊是如何創造出奧斯卡得獎片"班傑明的奇幻旅程"裡，年輕和衰老的布萊德彼特。
Ed Ulbrich - Visual storyteller
Ed Ulbrich works at the leading edge of computer-generated visuals. On a recent project, filmmakers, artists, and technologists have been working at a breakthrough point where reality and digitally created worlds collide. Full bio
Ed Ulbrich works at the leading edge of computer-generated visuals. On a recent project, filmmakers, artists, and technologists have been working at a breakthrough point where reality and digitally created worlds collide. Full bio
Double-click the English transcript below to play the video.
00:18
I'm here today representing a team of artists and technologists and filmmakers
0
0
5000
今天我代表一群藝術家、技術人員、電影製作團隊
00:23
that worked together on a remarkable film project for the last four years.
1
5000
3000
過去四年間，我們共同完成了一項非凡的電影工程
00:26
And along the way they created a breakthrough in computer visualization.
2
8000
4000
過程中也突破了電影繪圖技術
00:30
So I want to show you a clip of the film now.
3
12000
3000
我先讓大家看個電影片段
00:33
Hopefully it won't stutter.
4
15000
3000
希望播放時不會卡住
00:36
And if we did our jobs well, you won't know that we were even involved.
5
18000
3000
如果我們做好工作的話，你們不會看出我們動了什麼手腳
00:39
Voice (Video): I don't know how it's possible ...
6
21000
3000
影片: 我不知道這是怎麼回事
00:42
but you seem to have more hair.
7
24000
3000
但你頭髮好像變多了
00:45
Brad Pitt: What if I told you that I wasn't getting older ...
8
27000
3000
班傑明巴頓: 如果我告訴你，我沒有越來越老
00:48
but I was getting younger than everybody else?
9
30000
2000
反而是越來越年輕了呢?
00:53
I was born with some form of disease.
10
35000
3000
我出生就有一種病
00:56
Voice: What kind of disease?
11
38000
2000
什麼樣的病?
00:58
BP: I was born old.
12
40000
2000
我出生就是老人的樣子
01:01
Man: I'm sorry.
13
43000
2000
我很遺憾
01:03
BP: No need to be. There's nothing wrong with old age.
14
45000
3000
不用遺憾，老又沒有錯
01:08
Girl: Are you sick?
15
50000
2000
你生病了嗎?
01:10
BP: I heard momma and Tizzy whisper,
16
52000
3000
我聽到媽和Tizzy講悄悄話
01:13
and they said I was gonna die soon.
17
55000
2000
說我活不了多久
01:15
But ... maybe not.
18
57000
3000
但... 也許他們說錯了
01:18
Girl: You're different than anybody I've ever met.
19
60000
3000
我沒見過像你這樣的人
01:22
BB: There were many changes ...
20
64000
3000
我身體開始變化
01:25
some you could see, some you couldn't.
21
67000
3000
有些看的見，有些則不行
01:28
Hair started growing in all sorts of places,
22
70000
3000
體毛開始到處長
01:31
along with other things.
23
73000
3000
還有一些其他東西
01:34
I felt pretty good, considering.
24
76000
3000
其實這感覺還不賴
01:38
Ed Ulbrich: That was a clip from "The Curious Case of Benjamin Button."
25
80000
4000
Ed Ulbrich: 這是電影"班傑明的奇幻旅程"片段
01:42
Many of you, maybe you've seen it or you've heard of the story,
26
84000
4000
相信很多人已經看過、或聽過這故事
01:46
but what you might not know
27
88000
2000
但你也許不知道
01:48
is that for nearly the first hour of the film,
28
90000
2000
電影前將近一小時
01:50
the main character, Benjamin Button, who's played by Brad Pitt,
29
92000
3000
由布萊德彼特飾演的主角 班傑明巴頓
01:53
is completely computer-generated from the neck up.
30
95000
3000
頸部以上完全是電腦合成的
01:56
Now, there's no use of prosthetic makeup
31
98000
3000
沒有靠任何化妝技術
01:59
or photography of Brad superimposed over another actor's body.
32
101000
3000
也不是布萊德的臉，蓋過另一個演員的身體
02:02
We've created a completely digital human head.
33
104000
3000
我們全以數位方式製造出人頭
02:05
So I'd like to start with a little bit of history on the project.
34
107000
3000
我先稍微說一下製作起源
02:08
This is based on an F. Scott Fitzgerald short story.
35
110000
2000
電影改編自史考特．費滋傑羅的短篇故事
02:10
It's about a man who's born old and lives his life in reverse.
36
112000
3000
關於一個出生時80歲的男人，越活越年輕的故事
02:13
Now, this movie has floated around Hollywood
37
115000
2000
這部電影在好萊塢就謠傳了很久
02:15
for well over half a century,
38
117000
2000
超過50年有吧
02:17
and we first got involved with the project in the early '90s,
39
119000
3000
我們第一次接觸這計劃是在1990年初
02:20
with Ron Howard as the director.
40
122000
2000
導演是朗霍華
02:22
We took a lot of meetings and we seriously considered it.
41
124000
3000
我們討論了很久，認真考慮要接
02:25
But at the time we had to throw in the towel.
42
127000
2000
到後來真的必須放棄的時候
02:27
It was deemed impossible.
43
129000
2000
這已經被認為是不可能的任務
02:29
It was beyond the technology of the day to depict a man aging backwards.
44
131000
4000
當時的科技無法刻畫出返老還童的樣子
02:33
The human form, in particular the human head,
45
135000
3000
整個人，特別是頭的部份
02:36
has been considered the Holy Grail of our industry.
46
138000
3000
在我們這一行，稱作是聖杯傳奇
02:39
The project came back to us about a decade later,
47
141000
3000
約十年後又有人提起這計畫
02:42
and this time with a director named David Fincher.
48
144000
3000
這次的導演是大衛芬奇
02:45
Now, Fincher is an interesting guy.
49
147000
3000
芬奇真的是怪咖一個
02:48
David is fearless of technology,
50
150000
2000
一點都不怕技術上困難
02:50
and he is absolutely tenacious.
51
152000
2000
就是堅持一定要做
02:52
And David won't take "no."
52
154000
2000
沒有人能拒絕他
02:54
And David believed, like we do in the visual effects industry,
53
156000
3000
大衛、和我們做特效的持同樣看法
02:57
that anything is possible
54
159000
3000
沒有什麼是不可能的
03:00
as long as you have enough time, resources and, of course, money.
55
162000
3000
只要有足夠的時間、資源，當然還有資金
03:03
And so David had an interesting take on the film,
56
165000
4000
大衛對這部電影有獨特的想法
03:07
and he threw a challenge at us.
57
169000
3000
丟給我們這項考驗
03:10
He wanted the main character of the film to be played
58
172000
3000
他要這部電影的主角，從出生到死亡
03:13
from the cradle to the grave by one actor.
59
175000
2000
都要是同一個人來演
03:15
It happened to be this guy.
60
177000
2000
也就是這個人
03:17
We went through a process of elimination and a process of discovery
61
179000
3000
我們經過了一連串淘汰、一連串新發現
03:20
with David, and we ruled out, of course, swapping actors.
62
182000
3000
由於大衛堅持，我們排除使用不同演員的方法
03:23
That was one idea: that we would have different actors,
63
185000
3000
本來是想利用不同演員
03:26
and we would hand off from actor to actor.
64
188000
2000
分別演出不同階段的人生
03:28
We even ruled out the idea of using makeup.
65
190000
2000
我們也排除利用化妝
03:30
We realized that prosthetic makeup just wouldn't hold up,
66
192000
3000
因為用石膏來弄根本做不出來
03:33
particularly in close-up.
67
195000
2000
尤其是近拍的時候
03:35
And makeup is an additive process. You have to build the face up.
68
197000
3000
況且也要先做出一張臉才能化
03:38
And David wanted to carve deeply into Brad's face
69
200000
3000
大衛想要在布萊德彼特的臉上刻畫出皺紋
03:41
to bring the aging to this character.
70
203000
2000
來表現出這角色的年老樣貌
03:43
He needed to be a very sympathetic character.
71
205000
2000
他必須是個能讓人感同身受的角色
03:45
So we decided to cast a series of little people
72
207000
3000
所以我們決定選出一批小人物
03:48
that would play the different bodies of Benjamin
73
210000
3000
來演出班傑明在不同人生階段
03:51
at the different increments of his life
74
213000
2000
的各種身體樣貌
03:53
and that we would in fact create a computer-generated version of Brad's head,
75
215000
3000
然後我們利用電腦合成出布萊德的頭
03:56
aged to appear as Benjamin,
76
218000
2000
弄出不同年紀的班傑明
03:58
and attach that to the body of the real actor.
77
220000
3000
在接上其他演員的身體
04:01
Sounded great.
78
223000
2000
聽起來很可行
04:03
Of course, this was the Holy Grail of our industry,
79
225000
3000
當然，這可是我們這行的聖杯傳奇
04:06
and the fact that this guy is a global icon didn't help either,
80
228000
3000
加上這傢伙沒有人不認識
04:09
because I'm sure if any of you ever stand in line at the grocery store,
81
231000
3000
只要去過雜貨店一定看過他的臉
04:12
you know -- we see his face constantly.
82
234000
3000
這可不是什麼好處
04:15
So there really was no tolerable margin of error.
83
237000
2000
正因如此我們不能出錯
04:17
There were two studios involved: Warner Brothers and Paramount.
84
239000
3000
過程中由兩個製片公司一起合作: 華納兄弟和派拉蒙
04:20
And they both believed this would make an amazing film, of course,
85
242000
3000
當然，他們都深信這會是一部很讚的電影
04:23
but it was a very high-risk proposition.
86
245000
3000
無庸置疑也是很冒險的計劃
04:26
There was lots of money and reputations at stake.
87
248000
3000
攸關大把金錢與名聲
04:29
But we believed that we had a very solid methodology
88
251000
3000
我們認為我們有很可靠的方法
04:32
that might work ...
89
254000
3000
有可能成功的
04:35
But despite our verbal assurances,
90
257000
3000
但我們的口頭保證不夠
04:38
they wanted some proof.
91
260000
2000
他們要看到證據
04:40
And so, in 2004, they commissioned us to do a screen test of Benjamin.
92
262000
3000
所以在2004年，他們要求看到班傑明的試鏡畫面
04:43
And we did it in about five weeks.
93
265000
3000
我們做了約五週
04:46
But we used lots of cheats and shortcuts.
94
268000
3000
用了些旁門左道
04:49
We basically put something together to get through the meeting.
95
271000
3000
基本上我們東加西減地混過會談
04:52
I'll roll that for you now. This was the first test for Benjamin Button.
96
274000
3000
現在放給各位看。這是第一個班傑明巴頓
04:55
And in here, you can see, that's a computer-generated head --
97
277000
3000
在這裡可以看出是電腦合成出來的頭
04:58
pretty good -- attached to the body of an actor.
98
280000
3000
其實不錯，可以接到演員的身體
05:01
And it worked. And it gave the studio great relief.
99
283000
3000
也算成功，製片那邊終於放下大石
05:04
After many years of starts and stops on this project,
100
286000
3000
經過多年走走停停
05:07
and making that tough decision,
101
289000
3000
作出一些困難的決定
05:10
they finally decided to greenlight the movie.
102
292000
3000
他們終於同意了
05:13
And I can remember, actually, when I got the phone call to congratulate us,
103
295000
3000
我還記得，當我接到電話
05:16
to say the movie was a go,
104
298000
2000
恭喜我們電影終於可以開拍
05:18
I actually threw up.
105
300000
2000
我還吐了
05:20
(Laughter)
106
302000
2000
(笑聲)
05:22
You know, this is some tough stuff.
107
304000
2000
這是艱難的工作
05:24
So we started to have early team meetings,
108
306000
3000
我們開始初期的團隊會議
05:27
and we got everybody together,
109
309000
2000
集合大家
05:29
and it was really more like therapy in the beginning,
110
311000
3000
剛開始有點像在諮商治療
05:32
convincing each other and reassuring each other that we could actually undertake this.
111
314000
3000
告訴彼此，我們一定做得到的
05:35
We had to hold up an hour of a movie with a character.
112
317000
3000
要能撐過電影一個小時
05:38
And it's not a special effects film; it has to be a man.
113
320000
3000
而這不是特效片，必須用真人
05:41
We really felt like we were in a -- kind of a 12-step program.
114
323000
3000
這真的感覺像是戒酒那種12步計畫
05:44
And of course, the first step is: admit you've got a problem. (Laughter)
115
326000
3000
當然第一步就是: 承認問題的存在
05:48
So we had a big problem:
116
330000
2000
我們問題可大了
05:50
we didn't know how we were going to do this.
117
332000
3000
不曉得到底怎麼開始
05:53
But we did know one thing.
118
335000
2000
但我們知道
05:55
Being from the visual effects industry,
119
337000
3000
作視覺特效這行
05:58
we, with David, believed that we now had enough time,
120
340000
3000
我們和大衛相信的，只有足夠時間
06:01
enough resources, and, God, we hoped we had enough money.
121
343000
3000
足夠資源，天啊，希望我們有足夠資金
06:04
And we had enough passion to will the processes and technology into existence.
122
346000
5000
我們也有足夠的熱誠，成功的決心
06:09
So, when you're faced with something like that,
123
351000
2000
當你面對這樣的情況
06:11
of course you've got to break it down.
124
353000
2000
當然得先開始分析問題
06:13
You take the big problem and you break it down into smaller pieces
125
355000
2000
你面前有個大問題，把它拆成小問題
06:15
and you start to attack that.
126
357000
1000
個個攻破
06:16
So we had three main areas that we had to focus on.
127
358000
2000
主要集中在三個點
06:18
We needed to make Brad look a lot older --
128
360000
2000
我們要布萊德看起來很老
06:20
needed to age him 45 years or so.
129
362000
2000
我們要幫他加個45歲左右
06:22
And we also needed to make sure that we could take Brad's idiosyncrasies,
130
364000
6000
我們得確定它有布萊德的臉部特徵
06:28
his little tics, the little subtleties that make him who he is
131
370000
2000
臉部肌肉的抽動，各種他特有的細微變化
06:30
and have that translate through our process
132
372000
2000
經過我們的製作過程
06:32
so that it appears in Benjamin on the screen.
133
374000
3000
讓班傑明能夠呈現於大螢幕
06:35
And we also needed to create a character
134
377000
2000
我們也需要創造一個
06:37
that could hold up under, really, all conditions.
135
379000
3000
能夠做出每件事的角色
06:40
He needed to be able to walk in broad daylight,
136
382000
2000
他要能走在陽光下、
06:42
at nighttime, under candlelight,
137
384000
3000
夜晚、燭光下
06:45
he had to hold an extreme close-up,
138
387000
2000
近拍不能露餡
06:47
he had to deliver dialogue,
139
389000
1000
要能唸台詞
06:48
he had to be able to run, he had to be able to sweat,
140
390000
2000
要能跑、能流汗
06:50
he had to be able to take a bath, to cry,
141
392000
2000
能洗澡、能哭
06:52
he even had to throw up.
142
394000
1000
甚至要能嘔吐
06:53
Not all at the same time --
143
395000
1000
當然不是同時
06:54
but he had to, you know, do all of those things.
144
396000
2000
但他要能做到每件事
06:56
And the work had to hold up for almost the first hour of the movie.
145
398000
3000
而且要能撐住電影前一小時
06:59
We did about 325 shots.
146
401000
2000
我們拍了325個景
07:01
So we needed a system that would allow Benjamin
147
403000
3000
我們需要一個系統
07:04
to do everything a human being can do.
148
406000
3000
讓班傑明做出人類能做的所有事
07:07
And we realized that there was a giant chasm
149
409000
3000
我們知道2004年當時最先進的技術
07:10
between the state of the art of technology in 2004
150
412000
3000
和我們所想要的技術
07:13
and where we needed it to be.
151
415000
2000
仍有很大一段距離
07:15
So we focused on motion capture.
152
417000
3000
所以我們專注於動態捕捉技術
07:18
I'm sure many of you have seen motion capture.
153
420000
2000
我相信大部分的人都看過動態捕捉
07:20
The state of the art at the time
154
422000
2000
當時最先進的技術是
07:22
was something called marker-based motion capture.
155
424000
2000
利用標記點的動態捕捉
07:24
I'll give you an example here.
156
426000
1000
給大家看個例子
07:25
It's basically the idea of, you wear a leotard,
157
427000
2000
主要就是，穿著緊身衣
07:27
and they put some reflective markers on your body,
158
429000
2000
在身上標出一些反光標記點
07:29
and instead of using cameras,
159
431000
2000
物體四周放置紅外線感應
07:31
there're infrared sensors around a volume,
160
433000
2000
而不是攝影機
07:33
and those infrared sensors track the three-dimensional position
161
435000
2000
這些紅外線感應，會追蹤標記點
07:35
of those markers in real time.
162
437000
2000
當時的3D位置
07:37
And then animators can take the data of the motion of those markers
163
439000
3000
接著動畫師把標記點的數據
07:40
and apply them to a computer-generated character.
164
442000
2000
輸入到電腦做出來的角色上
07:42
You can see the computer characters on the right
165
444000
3000
可以看到右邊的動畫角色
07:45
are having the same complex motion as the dancers.
166
447000
3000
也能做出和舞者一樣的複雜動作
07:48
But we also looked at numbers of other films at the time
167
450000
2000
我們也看了一些
07:50
that were using facial marker tracking,
168
452000
2000
使用臉部標記點的電影
07:52
and that's the idea of putting markers on the human face
169
454000
2000
就是把點標記在臉上
07:54
and doing the same process.
170
456000
1000
過程相同
07:55
And as you can see, it gives you a pretty crappy performance.
171
457000
4000
可以看到，感覺蠻假的
07:59
That's not terribly compelling.
172
461000
3000
這樣可沒法說服別人
08:02
And what we realized
173
464000
2000
我們了解到
08:04
was that what we needed
174
466000
1000
我們所需要的
08:05
was the information that was going on between the markers.
175
467000
2000
是點與點之間的數據
08:07
We needed the subtleties of the skin.
176
469000
3000
我們要皮膚的微妙變化
08:10
We needed to see skin moving over muscle moving over bone.
177
472000
3000
要能看到骨頭到肌肉到皮膚的變化
08:13
We needed creases and dimples and wrinkles and all of those things.
178
475000
2000
需要皺紋、酒窩、細紋等等
08:15
Our first revelation was to completely abort and walk away from
179
477000
3000
我們了解現在需要完全放棄
08:18
the technology of the day, the status quo, the state of the art.
180
480000
3000
當時最先進的技術
08:21
So we aborted using motion capture.
181
483000
3000
我們放棄使用動態捕捉
08:24
And we were now well out of our comfort zone,
182
486000
3000
丟下了自己所熟悉的技術
08:27
and in uncharted territory.
183
489000
2000
到了陌生的領域
08:29
So we were left with this idea
184
491000
3000
我們有了一種想法
08:32
that we ended up calling "technology stew."
185
494000
3000
後來我們稱為科技大雜燴
08:35
We started to look out in other fields.
186
497000
2000
我們開始往別的領域找
08:37
The idea was that we were going to find
187
499000
3000
目的是要尋找出
08:40
nuggets or gems of technology
188
502000
2000
其他科技領域像是醫學影像
08:42
that come from other industries like medical imaging,
189
504000
2000
或電玩空間技術
08:44
the video game space,
190
506000
1000
之中的稀世珍品
08:45
and re-appropriate them.
191
507000
2000
拿來重新研究
08:47
And we had to create kind of a sauce.
192
509000
3000
雜燴也需要點醬料
08:50
And the sauce was code in software
193
512000
3000
就是軟體程式碼
08:53
that we'd written to allow these disparate pieces of technology
194
515000
3000
讓這些不同的領域的技術
08:56
to come together and work as one.
195
518000
2000
能整合在一起
08:58
Initially, we came across some remarkable research
196
520000
2000
起先，我們找到一項傑出的研究
09:00
done by a gentleman named Dr. Paul Ekman in the early '70s.
197
522000
3000
是70年代初期，Paul Ekman博士提出的
09:03
He believed that he could, in fact,
198
525000
3000
他相信他有辦法
09:06
catalog the human face.
199
528000
2000
將人臉目錄化
09:08
And he came up with this idea of Facial Action Coding System, or FACS.
200
530000
3000
他想到使用臉部表情辨識系統(FACS)
09:11
He believed that there were 70 basic poses
201
533000
3000
他相信人的臉型
09:14
or shapes of the human face,
202
536000
3000
有70種基本表情動作
09:17
and that those basic poses or shapes of the face
203
539000
3000
只要把基本動作或臉型
09:20
can be combined to create infinite possibilities
204
542000
3000
結合起來就能創造出
09:23
of everything the human face is capable of doing.
205
545000
2000
人臉做的出的無限表情
09:25
And of course, these transcend age, race, culture, gender.
206
547000
3000
不論年齡、種族、文化、性別，這70種動作都是一樣的
09:28
So this became the foundation of our research as we went forward.
207
550000
4000
這變成我們往後的基礎點
09:32
And then we came across some remarkable technology
208
554000
3000
之後我們發現了項非凡的技術
09:35
called Contour.
209
557000
1000
叫做輪廓提取(Contour)
09:36
And here you can see a subject having phosphorus makeup
210
558000
3000
現在看到的，是在人臉上
09:39
stippled on her face.
211
561000
2000
輕輕塗上磷
09:41
And now what we're looking at is really creating a surface capture
212
563000
3000
接下來是表面捕捉
09:44
as opposed to a marker capture.
213
566000
2000
正好與點的捕捉相反
09:46
The subject stands in front of a computer array of cameras,
214
568000
2000
接著站在一排電腦攝影機前面
09:48
and those cameras can, frame-by-frame,
215
570000
2000
攝影機就會一景一景的
09:50
reconstruct the geometry of exactly what the subject's doing at the moment.
216
572000
3000
重製出人當時的動作
09:53
So, effectively, you get 3D data in real time of the subject.
217
575000
5000
立刻就能得到3D數據
09:58
And if you look in a comparison,
218
580000
3000
各位可以看看比較
10:01
on the left, we see what volumetric data gives us
219
583000
3000
左邊是體積數據的結果
10:04
and on the right you see what markers give us.
220
586000
2000
右邊是標記點的
10:07
So, clearly, we were in a substantially better place for this.
221
589000
2000
可以清楚看出，我們有很大的進步
10:09
But these were the early days of this technology,
222
591000
2000
但這還是初期的結果
10:11
and it wasn't really proven yet.
223
593000
2000
還沒完全成功
10:13
We measure complexity and fidelity of data
224
595000
2000
但我們依據多邊形數量來計算
10:15
in terms of polygonal count.
225
597000
2000
數據的複雜性和準確性
10:17
And so, on the left, we were seeing 100,000 polygons.
226
599000
3000
左邊可以看到有十萬的多邊形
10:20
We could go up into the millions of polygons.
227
602000
2000
還可以做到幾十萬個
10:22
It seemed to be infinite.
228
604000
2000
甚至無限多
10:24
This was when we had our "Aha!"
229
606000
2000
這就是我們"啊哈"的時候
10:26
This was the breakthrough.
230
608000
1000
這是個大突破
10:27
This is when we're like, "OK, we're going to be OK,
231
609000
2000
這時我們才告訴自己
10:29
This is actually going to work."
232
611000
1000
一切都會沒事
10:30
And the "Aha!" was, what if we could take Brad Pitt,
233
612000
4000
"啊哈"就是，我們或許可以請布萊德彼特
10:34
and we could put Brad in this device,
234
616000
3000
使用這個裝置
10:37
and use this Contour process,
235
619000
2000
再搭配輪廓提取
10:39
and we could stipple on this phosphorescent makeup
236
621000
2000
塗上發磷光的顏料
10:41
and put him under the black lights,
237
623000
1000
用螢光燈照射
10:42
and we could, in fact, scan him in real time
238
624000
3000
拍攝Ekman的FACS臉部動作時
10:45
performing Ekman's FACS poses.
239
627000
2000
就能即時掃描出來
10:47
Right? So, effectively,
240
629000
2000
對吧，所以最後
10:49
we ended up with a 3D database
241
631000
2000
我們有了所有布萊德彼特的臉
10:51
of everything Brad Pitt's face is capable of doing.
242
633000
3000
能做的所有表情
10:54
(Laughter)
243
636000
2000
(笑聲)
10:56
From there, we actually carved up those faces
244
638000
3000
之後，將他的這些臉
10:59
into smaller pieces and components of his face.
245
641000
3000
劃分成小塊小塊的
11:02
So we ended up with literally thousands and thousands and thousands of shapes,
246
644000
3000
最後有了成千上萬的形狀
11:05
a complete database of all possibilities
247
647000
3000
就是這張臉可以做的
11:08
that his face is capable of doing.
248
650000
3000
所有表情
11:11
Now, that's great, except we had him at age 44.
249
653000
3000
看來很不錯，但我們只有44歲的他
11:14
We need to put another 40 years on him at this point.
250
656000
3000
現在我們需要的是40年後的他
11:17
We brought in Rick Baker,
251
659000
2000
我們找來了Rick Baker
11:19
and Rick is one of the great makeup and special effects gurus
252
661000
2000
Rick是我們這行的化妝、特效
11:21
of our industry.
253
663000
1000
大師之一
11:22
And we also brought in a gentleman named Kazu Tsuji,
254
664000
3000
我們也請來一位Kazu Tsuji先生
11:25
and Kazu Tsuji is one of the great photorealist sculptors of our time.
255
667000
3000
Kazu Tsuji是位很厲害的超現實雕刻家
11:28
And we commissioned them to make a maquette,
256
670000
3000
我們請他做班傑明的雕塑模型
11:31
or a bust, of Benjamin.
257
673000
2000
也可以說半身像
11:33
So, in the spirit of "The Great Unveiling" -- I had to do this --
258
675000
3000
本著盛大揭幕的精神
11:36
I had to unveil something.
259
678000
2000
我一定要揭露一下
11:38
So this is Ben 80.
260
680000
2000
這是80歲的班傑明
11:40
We created three of these:
261
682000
2000
我們製作了三個
11:42
there's Ben 80, there's Ben 70, there's Ben 60.
262
684000
2000
80歲、70歲、60歲
11:44
And this really became the template for moving forward.
263
686000
3000
這就是之後我們依照的模板
11:47
Now, this was made from a life cast of Brad.
264
689000
2000
這是以布萊德的模樣刻的
11:49
So, in fact, anatomically, it is correct.
265
691000
3000
所以當然是正確的樣貌
11:52
The eyes, the jaw, the teeth:
266
694000
3000
眼睛、下巴、牙齒...
11:55
everything is in perfect alignment with what the real guy has.
267
697000
3000
每一樣都是真人比例
11:58
We have these maquettes scanned into the computer
268
700000
2000
我們把雕塑圖掃瞄到電腦
12:00
at very high resolution --
269
702000
2000
使用超高解析
12:02
enormous polygonal count.
270
704000
2000
很多的多邊形
12:04
And so now we had three age increments of Benjamin
271
706000
4000
現在電腦裡有了三種班傑明
12:08
in the computer.
272
710000
2000
年老的樣子
12:10
But we needed to get a database of him doing more than that.
273
712000
3000
但我們需要更多的數據
12:13
We went through this process, then, called retargeting.
274
715000
3000
接下來的過程，叫做重定向
12:16
This is Brad doing one of the Ekman FACS poses.
275
718000
2000
這是布萊德做的Ekman FACS
12:18
And here's the resulting data that comes from that,
276
720000
3000
這是獲得的數據
12:21
the model that comes from that.
277
723000
2000
弄出來的樣子
12:23
Retargeting is the process of transposing that data
278
725000
3000
重定向就是將這些數據轉移到
12:26
onto another model.
279
728000
2000
另一個模型上
12:28
And because the life cast, or the bust -- the maquette -- of Benjamin
280
730000
3000
利用依照布萊德做出的
12:31
was made from Brad,
281
733000
2000
班傑明的半身像
12:33
we could transpose the data of Brad at 44
282
735000
3000
我們就能把44歲布萊德的數據轉移
12:36
onto Brad at 87.
283
738000
2000
到87歲的布萊德
12:38
So now, we had a 3D database of everything Brad Pitt's face can do
284
740000
3000
所以我們有了布萊德彼特
12:41
at age 87, in his 70s and in his 60s.
285
743000
4000
87、70、60歲的所有臉部3D資料
12:45
Next we had to go into the shooting process.
286
747000
3000
接下來就是進行拍攝了
12:48
So while all that's going on,
287
750000
1000
同時間
12:49
we're down in New Orleans and locations around the world.
288
751000
2000
我們去了紐奧良和其他各地
12:51
And we shot our body actors,
289
753000
2000
我們先拍了身體演員
12:53
and we shot them wearing blue hoods.
290
755000
2000
讓他們穿了藍色頭套
12:55
So these are the gentleman who played Benjamin.
291
757000
2000
這位先生飾演班傑明
12:57
And the blue hoods helped us with two things:
292
759000
2000
藍色頭套的功用是
12:59
one, we could easily erase their heads;
293
761000
2000
第一，可以輕易消除他們的頭
13:01
and we also put tracking markers on their heads
294
763000
2000
還有，可以直接標上追蹤點
13:03
so we could recreate the camera motion
295
765000
2000
以便讓我們能在片廠重製
13:05
and the lens optics from the set.
296
767000
2000
攝影動作和鏡片光學
13:07
But now we needed to get Brad's performance to drive our virtual Benjamin.
297
769000
3000
現在需要拍布萊德的表演，來結合虛擬的班傑明
13:10
And so we edited the footage that was shot on location
298
772000
2000
我們重新編輯了實體拍攝的
13:12
with the rest of the cast and the body actors
299
774000
3000
其他演員片段
13:15
and about six months later
300
777000
2000
約半年後
13:17
we brought Brad onto a sound stage in Los Angeles
301
779000
3000
布萊德來到了洛杉磯片廠
13:20
and he watched on the screen.
302
782000
3000
他看了已經拍好的畫面
13:23
His job, then, was to become Benjamin.
303
785000
2000
他的工作就是變成班傑明
13:25
And so we looped the scenes.
304
787000
1000
我們重複放了片段
13:26
He watched again and again.
305
788000
1000
他看了又看
13:27
We encouraged him to improvise.
306
789000
2000
我們鼓勵他即興唸台詞
13:29
And he took Benjamin into interesting and unusual places
307
791000
3000
他也把班傑明演的有趣獨特
13:32
that we didn't think he was going to go.
308
794000
2000
出乎我們料想
13:34
We shot him with four HD cameras
309
796000
2000
我們用四台高清攝影機拍攝
13:36
so we'd get multiple views of him
310
798000
1000
為了取得各種角度的畫面
13:37
and then David would choose the take of Brad being Benjamin
311
799000
3000
然後大衛會選最符合該片段
13:40
that he thought best matched the footage
312
802000
3000
的臉部角度
13:43
with the rest of the cast.
313
805000
1000
以搭配身體部分
13:44
From there we went into a process called image analysis.
314
806000
3000
再來的過程是影像分析
13:47
And so here, you can see again, the chosen take.
315
809000
3000
可以看到，選好的片段
13:50
And you are seeing, now, that data being transposed on to Ben 87.
316
812000
3000
資料轉換成班傑明87歲
13:53
And so, what's interesting about this is
317
815000
3000
最有趣的就是
13:56
we used something called image analysis,
318
818000
2000
我們使用畫面分析
13:58
which is taking timings from different components of Benjamin's face.
319
820000
3000
擷取各時間點班傑明臉部的每個部分
14:01
And so we could choose, say, his left eyebrow.
320
823000
3000
所以我們可以隨便選，例如他的左眉
14:04
And the software would tell us that, well,
321
826000
2000
軟體就會告訴我們
14:06
in frame 14 the left eyebrow begins to move from here to here,
322
828000
2000
在第14景，左眉從這裡移到這裡
14:08
and it concludes moving in frame 32.
323
830000
2000
然後在第32景停止
14:10
And so we could choose numbers of positions on the face
324
832000
2000
可以隨便取得任何一個表情的
14:12
to pull that data from.
325
834000
2000
臉部數據
14:14
And then, the sauce I talked about with our technology stew --
326
836000
2000
再來就是剛剛提過，科技大雜燴的醬料
14:16
that secret sauce was, effectively, software that allowed us to
327
838000
3000
獨門配方就是，一個軟體能讓我們
14:19
match the performance footage of Brad
328
841000
3000
用布萊德的表演
14:22
in live action with our database of aged Benjamin,
329
844000
4000
配上電腦裡班傑明年老的
14:26
the FACS shapes that we had.
330
848000
2000
FACS數據
14:28
On a frame-by-frame basis,
331
850000
3000
有了各景的基礎
14:31
we could actually reconstruct a 3D head
332
853000
3000
我們可以重建出完全符合
14:34
that exactly matched the performance of Brad.
333
856000
3000
布萊德演出的3D頭
14:37
So this was how the finished shot appeared in the film.
334
859000
3000
這就是電影最終畫面
14:40
And here you can see the body actor.
335
862000
2000
可以看到身體演員
14:42
And then this is what we called the "dead head," no reference to Jerry Garcia.
336
864000
3000
這個我們稱呼為死人頭，跟Jerry Garcia無關
14:45
And then here's the reconstructed performance
337
867000
3000
這個是重建的演出
14:48
now with the timings of the performance.
338
870000
3000
配上正確的時間點
14:51
And then, again, the final shot.
339
873000
2000
最後，完成的畫面
14:54
It was a long process.
340
876000
2000
是很冗長的過程
14:56
(Applause)
341
878000
3000
(掌聲)
15:07
The next section here, I'm going to just blast through this,
342
889000
2000
下一段我只能稍微帶過
15:09
because we could do a whole TEDTalk on the next several slides.
343
891000
4000
要細說可能需要再做一次TEDTalk
15:13
We had to create a lighting system.
344
895000
3000
我們必須創造出燈光系統
15:16
So really, a big part of our processes was creating a lighting environment
345
898000
3000
這部分最難的是，創造班傑明每一場景
15:19
for every single location that Benjamin had to appear
346
901000
2000
的各種燈光環境
15:21
so that we could put Ben's head into any scene
347
903000
3000
到時候把頭接上才不會怪怪的
15:24
and it would exactly match the lighting that's on the other actors
348
906000
3000
而且才能完美搭配真實演員的
15:27
in the real world.
349
909000
1000
身體演出部份
15:28
We also had to create an eye system.
350
910000
3000
我們也需要創造眼睛系統
15:31
We found the old adage, you know,
351
913000
2000
因為俗話說的對
15:33
"The eyes are the window to the soul,"
352
915000
2000
"眼睛是靈魂之窗"
15:35
absolutely true.
353
917000
1000
真是一點也沒錯
15:36
So the key here was to keep everybody looking in Ben's eyes.
354
918000
2000
關鍵就是，要讓大家能在班傑明眼中
15:38
And if you could feel the warmth, and feel the humanity,
355
920000
2000
感覺的到溫暖，感覺的到人性
15:40
and feel his intent coming through the eyes,
356
922000
3000
從眼中感受的到熱誠
15:43
then we would succeed.
357
925000
1000
這樣就是成功了
15:44
So we had one person focused on the eye system
358
926000
3000
我們有個同仁專門弄眼睛系統
15:47
for almost two full years.
359
929000
2000
快兩年的時間
15:49
We also had to create a mouth system.
360
931000
2000
再來是嘴巴系統
15:51
We worked from dental molds of Brad.
361
933000
2000
我們從布萊德的齒模下手
15:53
We had to age the teeth over time.
362
935000
2000
再來弄不同年紀的牙齒
15:55
We also had to create an articulating tongue that allowed him to enunciate his words.
363
937000
3000
還有要個能夠清晰發音的舌頭
15:58
There was a whole system written in software to articulate the tongue.
364
940000
2000
有個完整程式負責清晰地發音
16:00
We had one person devoted to the tongue for about nine months.
365
942000
2000
有位同仁花九個月的時間，致力於舌頭
16:02
He was very popular.
366
944000
2000
他很受歡迎
16:04
Skin displacement: another big deal.
367
946000
3000
皮膚替換: 另一個挑戰
16:07
The skin had to be absolutely accurate.
368
949000
2000
皮膚必須完全精確
16:09
He's also in an old age home, he's in a nursing home
369
951000
3000
他待過安養院、療養院
16:12
around other old people,
370
954000
2000
身邊有其他老人
16:14
so he had to look exactly the same as the others.
371
956000
2000
所以他必須看起來跟其他人一樣
16:16
So, lots of work on skin deformation,
372
958000
1000
老化皮膚是項大工程
16:17
you can see in some of these cases it works,
373
959000
1000
可以看到有時候成功
16:18
in some cases it looks bad.
374
960000
1000
有時候失敗
16:19
This is a very, very, very early test in our process.
375
961000
2000
這是我們非常早期的成果
16:21
So, effectively we created a digital puppet
376
963000
3000
蠻成功的做出數位人偶
16:24
that Brad Pitt could operate with his own face.
377
966000
3000
讓布萊德彼特可以自己控制
16:27
There were no animators necessary to come in and interpret behavior
378
969000
4000
不需要動畫師來製作
16:31
or enhance his performance.
379
973000
2000
或加強表演
16:33
There was something that we encountered, though,
380
975000
3000
我們還經歷了一個過程
16:36
that we ended up calling "the digital Botox effect."
381
978000
3000
我們稱作是數位肉毒桿菌
16:39
So, as things went through this process,
382
981000
3000
在過程中
16:42
Fincher would always say, "It sandblasts the edges off of the performance."
383
984000
3000
芬奇說，這修飾了最後演出
16:45
And thing our process and the technology couldn't do,
384
987000
3000
製作過程與技術唯一做不到的
16:48
is they couldn't understand intent,
385
990000
3000
是無法理解意圖
16:51
the intent of the actor.
386
993000
2000
演員的意圖
16:53
So it sees a smile as a smile.
387
995000
2000
微笑就是微笑
16:55
It doesn't recognize an ironic smile, or a happy smile,
388
997000
3000
它不會分辨是諷刺的笑、或開心的笑
16:58
or a frustrated smile.
389
1000000
1000
或挫折的笑
16:59
So it did take humans to kind of push it one way or another.
390
1001000
3000
沒辦法理解其中的情緒意思
17:02
But we ended up calling the entire process
391
1004000
3000
我們最後稱呼這個過程
17:05
and all the technology "emotion capture,"
392
1007000
2000
情緒捕捉技術
17:07
as opposed to just motion capture.
393
1009000
1000
與動態捕捉相反
17:08
Take another look.
394
1010000
2000
再看一次
17:11
Brad Pitt: Well, I heard momma and Tizzy whisper,
395
1013000
2000
布萊德彼特: 我聽到媽和Tizzy講悄悄話
17:13
and they said I was gonna die soon,
396
1015000
2000
說我活不了多久
17:15
but ... maybe not.
397
1017000
2000
但是... 也許他們說錯了
17:37
EU: That's how to create a digital human in 18 minutes.
398
1039000
3000
這就是18分鐘內解釋完
17:40
(Applause)
399
1042000
3000
(掌聲)
17:48
A couple of quick factoids;
400
1050000
2000
再補件有趣的事
17:50
it really took 155 people over two years,
401
1052000
4000
整個過程大概花了155人兩年的時間
17:54
and we didn't even talk about 60 hairstyles and an all-digital haircut.
402
1056000
4000
我們甚至沒聊到60個髮型和數位剪髮
17:58
But, that is Benjamin. Thank you.
403
1060000
3000
但，這就是大家的班傑明，謝謝
ABOUT THE SPEAKER
Ed Ulbrich - Visual storytellerEd Ulbrich works at the leading edge of computer-generated visuals. On a recent project, filmmakers, artists, and technologists have been working at a breakthrough point where reality and digitally created worlds collide.
Why you should listen
Ed Ulbrich spoke at TED2009 representing a team of filmmakers, artists and technologists who've been working on a significant breakthrough in visual storytelling -- a startling blurring of the line between digital creation and actor.
Ulbrich was the long time executive VP of production at Digital Domain, for whom he executive-produced Academy Award-winning visual effects for Titanic, What Dreams May Come, Fight Club, Zodiac, Adaptation and other features, as well as music videos and more than 500 commercials. He has recently exited this position but has entered into a creative consultant arrangement with the company. In 2007, he was named to the Creativity 50 -- top innovators in advertising and design.
Ed Ulbrich | Speaker | TED.com