Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:00,000 --> 00:00:02,560
[MUSIC PLAYING]
2
00:00:03,060 --> 00:00:06,390
JAMES CAMERON: When the term is
a computer-generated character,
3
00:00:06,390 --> 00:00:08,430
there's the impression
that the computer's
4
00:00:08,430 --> 00:00:09,750
creating the character.
5
00:00:09,750 --> 00:00:11,430
Computer's not
creating anything.
6
00:00:11,430 --> 00:00:15,210
Computer is doing the number
crunching to generate an image.
7
00:00:15,210 --> 00:00:17,640
But character is being
created by the actor.
8
00:00:17,640 --> 00:00:20,700
It's not animation, it's
performance capture.
9
00:00:20,700 --> 00:00:23,628
[MUSIC PLAYING]
10
00:00:34,370 --> 00:00:38,330
We've talked about
low-budget filmmaking.
11
00:00:38,330 --> 00:00:40,430
Now we're going to talk
about the opposite end
12
00:00:40,430 --> 00:00:41,360
of the spectrum.
13
00:00:41,360 --> 00:00:47,728
In "Avatar," we were-- it was a
highly experimental production.
14
00:00:47,728 --> 00:00:49,520
So one of the things
that's unique about it
15
00:00:49,520 --> 00:00:53,360
is that we went into it starting
years before the release.
16
00:00:53,360 --> 00:00:58,428
We started in 2005, and the
film came out in late 2009,
17
00:00:58,428 --> 00:00:59,970
not knowing exactly
how we were going
18
00:00:59,970 --> 00:01:02,380
to solve most of the problems.
19
00:01:02,380 --> 00:01:05,950
We had a general idea
of how to do things.
20
00:01:05,950 --> 00:01:08,460
We were pretty
thin on specifics.
21
00:01:08,460 --> 00:01:12,270
And I had actually
written "Avatar" in 1995,
22
00:01:12,270 --> 00:01:14,880
as a challenge to my
digital effects company,
23
00:01:14,880 --> 00:01:16,560
Digital Domain.
24
00:01:16,560 --> 00:01:18,060
And they said, we
can't do it yet.
25
00:01:18,060 --> 00:01:20,370
We don't have the
tools to do the muscle
26
00:01:20,370 --> 00:01:22,650
rigs, and the
facial muscle rigs,
27
00:01:22,650 --> 00:01:24,120
and the eyes, and the hands.
28
00:01:24,120 --> 00:01:25,710
And we can't do it.
29
00:01:25,710 --> 00:01:26,610
It's too much.
30
00:01:26,610 --> 00:01:28,470
It's beyond us right now.
31
00:01:28,470 --> 00:01:30,060
I said, all right,
I'll just put it--
32
00:01:30,060 --> 00:01:32,552
I'll just put it in a
shelf for a few years.
33
00:01:32,552 --> 00:01:34,260
And I waited, and I
waited, and I waited.
34
00:01:34,260 --> 00:01:37,500
And then I was watching
"The Two Towers."
35
00:01:37,500 --> 00:01:41,428
And there's the scene where
Gollum talks to himself.
36
00:01:41,428 --> 00:01:42,720
And he's in two places at once.
37
00:01:42,720 --> 00:01:44,970
And he's having this kind of
schizophrenic, out-of-body
38
00:01:44,970 --> 00:01:45,480
experience.
39
00:01:45,480 --> 00:01:47,370
And it's highly expressive,
highly emotional.
40
00:01:47,370 --> 00:01:49,290
It goes on for several minutes.
41
00:01:49,290 --> 00:01:51,480
And it's 100% CG character.
42
00:01:51,480 --> 00:01:53,670
And I said to
myself, if they can
43
00:01:53,670 --> 00:01:55,435
do that, I can make "Avatar."
44
00:01:55,435 --> 00:01:57,810
And that's what set me down
the path of making the movie.
45
00:01:57,810 --> 00:02:00,240
So I was inspired by the
work of another filmmaker.
46
00:02:00,240 --> 00:02:03,690
I was building on what
Peter Jackson had done
47
00:02:03,690 --> 00:02:05,800
and what WETA Digital
had already done.
48
00:02:05,800 --> 00:02:08,160
So I came in with a whole
new set of challenges
49
00:02:08,160 --> 00:02:10,157
and a new high bar they
had to get through.
50
00:02:10,157 --> 00:02:10,990
And they loved that.
51
00:02:10,990 --> 00:02:13,110
They loved that challenge.
52
00:02:13,110 --> 00:02:15,090
We just studied a lot
of close-ups, really
53
00:02:15,090 --> 00:02:17,820
big close-ups of men and
women and looked at them
54
00:02:17,820 --> 00:02:20,473
in different kinds of lighting
and said, well what do you see?
55
00:02:20,473 --> 00:02:22,140
What do you see about
the way they move?
56
00:02:22,140 --> 00:02:24,223
What do you see about the
way they take the light?
57
00:02:24,223 --> 00:02:26,520
What do you see
about the skin pores?
58
00:02:26,520 --> 00:02:28,060
What do you see about the eye?
59
00:02:28,060 --> 00:02:28,560
You know?
60
00:02:28,560 --> 00:02:31,410
And one of the breakthroughs
was the eye is not dry,
61
00:02:31,410 --> 00:02:32,310
the eye is wet.
62
00:02:32,310 --> 00:02:33,570
That's kind of obvious.
63
00:02:33,570 --> 00:02:38,070
But it also means that the
upper part of the lower eyelid
64
00:02:38,070 --> 00:02:39,570
is actually a wet surface.
65
00:02:39,570 --> 00:02:45,510
And there's a little meniscus of
tear fluid between that eyelid
66
00:02:45,510 --> 00:02:47,880
and between the eyeball.
67
00:02:47,880 --> 00:02:49,230
And it catches the light.
68
00:02:49,230 --> 00:02:51,990
And there's a little glint
there, and it's always there.
69
00:02:51,990 --> 00:02:54,480
But if you're not looking
for it, you'll never see it.
70
00:02:54,480 --> 00:02:59,222
The second we put the meniscus
in, the eye came to life.
71
00:02:59,222 --> 00:03:01,430
I know, it sounds like a
little tiny thing, something
72
00:03:01,430 --> 00:03:02,430
you'd never think about.
73
00:03:02,430 --> 00:03:05,720
But when it's missing,
you sense the lack.
74
00:03:05,720 --> 00:03:09,140
And I always felt that
CG characters fell down
75
00:03:09,140 --> 00:03:10,520
around the movement
of the mouth.
76
00:03:10,520 --> 00:03:12,200
It always felt mask-like.
77
00:03:12,200 --> 00:03:14,750
It always felt sort of too
thin at the corner of the mouth
78
00:03:14,750 --> 00:03:17,420
and not interacting
properly with the teeth.
79
00:03:17,420 --> 00:03:21,080
It didn't feel like those
muscles really exist,
80
00:03:21,080 --> 00:03:22,585
like the dimensionality
of the mouth
81
00:03:22,585 --> 00:03:23,752
and the face really existed.
82
00:03:23,752 --> 00:03:27,740
So we spent literally a couple
of years figuring out how
83
00:03:27,740 --> 00:03:29,600
to do this with WETA Digital.
84
00:03:29,600 --> 00:03:32,630
And WETA Digital did the
heavy lifting on this.
85
00:03:32,630 --> 00:03:36,230
And it was iteration after
iteration after iteration
86
00:03:36,230 --> 00:03:38,143
and coming up with
little things.
87
00:03:38,143 --> 00:03:40,310
So the most critical thing
about all visual effects,
88
00:03:40,310 --> 00:03:42,560
and certainly any kind
of CG character design,
89
00:03:42,560 --> 00:03:45,920
and especially if you're trying
to emulate photo reality,
90
00:03:45,920 --> 00:03:47,540
is to observe.
91
00:03:47,540 --> 00:03:48,920
Observe, observe, observe.
92
00:03:48,920 --> 00:03:51,806
[MUSIC PLAYING]
93
00:03:56,620 --> 00:03:58,420
In a nutshell, the
methodology was
94
00:03:58,420 --> 00:04:01,300
that we're going to capture
the performance using
95
00:04:01,300 --> 00:04:03,040
a collection of data.
96
00:04:03,040 --> 00:04:05,320
The idea of motion
capture where we
97
00:04:05,320 --> 00:04:10,013
capture the body movement of
a character had been around--
98
00:04:10,013 --> 00:04:12,180
I used it on "Titanic" for
some of the crowd scenes.
99
00:04:12,180 --> 00:04:13,870
So that had been
around for a while.
100
00:04:13,870 --> 00:04:16,452
It had been around
for at least 10 years.
101
00:04:16,452 --> 00:04:18,160
The idea of how you
were going to capture
102
00:04:18,160 --> 00:04:20,740
the specifics of a
facial performance, which
103
00:04:20,740 --> 00:04:24,440
is where you see the majority
of the emotion in the eyes,
104
00:04:24,440 --> 00:04:28,210
the mouth, all the many,
many muscles in the face.
105
00:04:28,210 --> 00:04:31,510
So it's almost it's
own separate world
106
00:04:31,510 --> 00:04:33,265
from the full body performance.
107
00:04:33,265 --> 00:04:35,140
People that have tried
to do this in the past
108
00:04:35,140 --> 00:04:38,408
tried to do it by using markers
that they put on the face
109
00:04:38,408 --> 00:04:40,450
and capturing the face
the same way they captured
110
00:04:40,450 --> 00:04:44,860
the body, with a bunch of motion
capture cameras and a big array
111
00:04:44,860 --> 00:04:45,830
around it.
112
00:04:45,830 --> 00:04:47,830
But the detail
wasn't fine enough.
113
00:04:47,830 --> 00:04:51,460
They could only work
in very small volumes.
114
00:04:51,460 --> 00:04:52,945
It really limited the action.
115
00:04:52,945 --> 00:04:54,570
It limited how they
were going to-- you
116
00:04:54,570 --> 00:04:58,170
know, the type of action
that they could stage.
117
00:04:58,170 --> 00:04:59,880
And it was very--
118
00:04:59,880 --> 00:05:04,020
I mean you had to literally glue
like 80 little, tiny glass bead
119
00:05:04,020 --> 00:05:05,700
balls onto the actor's face.
120
00:05:05,700 --> 00:05:06,390
It was horrible.
121
00:05:06,390 --> 00:05:07,950
Actors didn't like it.
122
00:05:07,950 --> 00:05:09,870
The end result
wasn't that great.
123
00:05:09,870 --> 00:05:13,270
So we proposed a completely
different solution,
124
00:05:13,270 --> 00:05:16,200
which is to take something that
looked a lot like a boom mic
125
00:05:16,200 --> 00:05:20,010
for a singer in a concert
and put a little tiny video
126
00:05:20,010 --> 00:05:22,200
camera out on the end of
it on a pretty wide lens
127
00:05:22,200 --> 00:05:25,710
and shoot the face, film
the face, or video the face,
128
00:05:25,710 --> 00:05:27,750
while the actor is
doing the scene.
129
00:05:27,750 --> 00:05:29,070
And it's knolled out to them.
130
00:05:29,070 --> 00:05:31,290
It stays with them
wherever they go.
131
00:05:31,290 --> 00:05:34,800
And it's getting a 100%
recording, a close-up
132
00:05:34,800 --> 00:05:39,310
recording, of their performance,
their facial performance.
133
00:05:39,310 --> 00:05:42,150
So now you've got a video data
stream coming out of the face,
134
00:05:42,150 --> 00:05:46,210
and we called that image-based
facial performance capture.
135
00:05:46,210 --> 00:05:46,710
Right?
136
00:05:46,710 --> 00:05:50,810
And that set a trend that's
still in motion to this day.
137
00:05:50,810 --> 00:05:53,010
A lot of other
people have used it.
138
00:05:53,010 --> 00:05:55,320
Different people have
brought different refinements
139
00:05:55,320 --> 00:05:55,990
to the process.
140
00:05:55,990 --> 00:05:58,890
But it hasn't fundamentally
changed since 2005 when
141
00:05:58,890 --> 00:06:01,300
we first launched the concept.
142
00:06:01,300 --> 00:06:04,660
So now you have two data streams
that are recording the actor's
143
00:06:04,660 --> 00:06:05,292
performance.
144
00:06:05,292 --> 00:06:06,750
You're getting the
body performance
145
00:06:06,750 --> 00:06:10,320
off a standard
motion capture system
146
00:06:10,320 --> 00:06:12,600
with a lot of really
tricky bells and whistles
147
00:06:12,600 --> 00:06:13,832
on it to make it better.
148
00:06:13,832 --> 00:06:15,540
And you're getting
the facial performance
149
00:06:15,540 --> 00:06:17,957
from this little camera that's
on this little camera boom.
150
00:06:17,957 --> 00:06:21,420
Here's me with my virtual
camera on "Avatar."
151
00:06:21,420 --> 00:06:24,840
And I'm working with the
actors in their capture suits.
152
00:06:24,840 --> 00:06:26,880
You can see that they
have marker suits.
153
00:06:26,880 --> 00:06:28,500
We're just in a big lit stage.
154
00:06:28,500 --> 00:06:30,930
There's no jungle
there whatsoever.
155
00:06:30,930 --> 00:06:32,620
They're running on a log.
156
00:06:32,620 --> 00:06:34,980
So I think these are
probably more stunt
157
00:06:34,980 --> 00:06:36,382
doubles than anything.
158
00:06:36,382 --> 00:06:37,590
And they're running on a log.
159
00:06:37,590 --> 00:06:39,640
They're just running
basically across the floor.
160
00:06:39,640 --> 00:06:40,710
It's very safe.
161
00:06:40,710 --> 00:06:42,060
And I'm looking--
162
00:06:42,060 --> 00:06:44,400
I'm not looking at
them with the camera,
163
00:06:44,400 --> 00:06:46,380
I'm looking at the
characters that they're
164
00:06:46,380 --> 00:06:50,700
playing in real-time
in the world,
165
00:06:50,700 --> 00:06:54,000
in the CG world all
at low resolution.
166
00:06:54,000 --> 00:06:56,520
It almost looks like an
old, crappy video game
167
00:06:56,520 --> 00:06:57,170
from the '80s.
168
00:06:57,170 --> 00:06:58,962
JAMES CAMERON (ON
VIDEO): (SHOUTING) And 6.
169
00:06:58,962 --> 00:07:00,510
And 7.
170
00:07:00,510 --> 00:07:01,985
And 8.
171
00:07:01,985 --> 00:07:04,110
We're much better at it
these days than we were--
172
00:07:04,110 --> 00:07:08,020
this is all, you know,
obviously 11 years old.
173
00:07:08,020 --> 00:07:08,532
Here's me.
174
00:07:08,532 --> 00:07:10,240
And, once again, I'm
not looking at them.
175
00:07:10,240 --> 00:07:11,250
I'm looking at my image.
176
00:07:11,250 --> 00:07:13,260
But I'm near them, so
I can talk to them.
177
00:07:13,260 --> 00:07:15,720
And I'm seeing what they
look like in the world.
178
00:07:15,720 --> 00:07:18,205
And then here's me playing back.
179
00:07:18,205 --> 00:07:19,580
And what I'm seeing
is they're up
180
00:07:19,580 --> 00:07:22,250
on what we call a vine bridge in
between two floating mountains,
181
00:07:22,250 --> 00:07:24,917
and they're 1,000
feet above the ground.
182
00:07:24,917 --> 00:07:26,500
But, of course,
because the characters
183
00:07:26,500 --> 00:07:29,380
are very mountain goat-like
and can run around
184
00:07:29,380 --> 00:07:31,930
1,000 feet above the
ground, it was okay for them
185
00:07:31,930 --> 00:07:35,530
to not be scared and
balancing and wobbling along.
186
00:07:35,530 --> 00:07:37,300
And it was completely safe.
187
00:07:37,300 --> 00:07:39,640
And so we're creating
a fantasy image
188
00:07:39,640 --> 00:07:45,220
and using relatively
simple setups to do it.
189
00:07:45,220 --> 00:07:47,710
And basically, we found out
that it worked very well.
190
00:07:47,710 --> 00:07:49,180
You could do just
about anything.
191
00:07:49,180 --> 00:07:51,430
You could run, you could
jump, you could ride a horse,
192
00:07:51,430 --> 00:07:53,230
you could have a big argument.
193
00:07:53,230 --> 00:07:55,300
You could even, if
you were careful,
194
00:07:55,300 --> 00:07:57,305
you could drink a beer.
195
00:07:57,305 --> 00:07:58,930
You could do whatever
you needed to do.
196
00:07:58,930 --> 00:08:00,972
The only thing we couldn't
shoot using the system
197
00:08:00,972 --> 00:08:02,710
was a kiss, for obvious reasons.
198
00:08:02,710 --> 00:08:05,360
Because the two booms
would be in the way.
199
00:08:05,360 --> 00:08:08,290
So we had to use a different
kind of harder technique that
200
00:08:08,290 --> 00:08:11,242
was more labor-intensive to
do the kiss scene and a couple
201
00:08:11,242 --> 00:08:11,950
of other moments.
202
00:08:11,950 --> 00:08:13,780
But for the most part
it worked for I'd
203
00:08:13,780 --> 00:08:16,935
say 99% of what we
were trying to shoot.
204
00:08:16,935 --> 00:08:18,310
We had a lot of
proof of concept.
205
00:08:18,310 --> 00:08:20,140
We spent a fair bit of
the studio's money--
206
00:08:20,140 --> 00:08:22,270
the studio being
20th Century Fox--
207
00:08:22,270 --> 00:08:25,430
to prove that we could do it.
208
00:08:25,430 --> 00:08:26,980
And to do that, we did a test.
209
00:08:26,980 --> 00:08:29,110
We did a 1 and 1/2 minute test.
210
00:08:29,110 --> 00:08:31,030
Boy meets girl in the jungle.
211
00:08:31,030 --> 00:08:32,679
And they talk.
212
00:08:32,679 --> 00:08:36,280
He only come
and make problems.
213
00:08:36,280 --> 00:08:37,059
Only.
214
00:08:37,059 --> 00:08:38,142
MAN (AS JAKE): Okay, fine.
215
00:08:38,142 --> 00:08:39,679
Fine.
216
00:08:39,679 --> 00:08:41,460
Fine.
217
00:08:41,460 --> 00:08:43,578
You love your little
forest friends.
218
00:08:43,578 --> 00:08:45,120
So why not just let
them kill my ass?
219
00:08:45,120 --> 00:08:47,940
What's the thinking?
220
00:08:47,940 --> 00:08:49,170
Why save?
221
00:08:49,170 --> 00:08:50,790
MAN (AS JAKE): Yes.
222
00:08:50,790 --> 00:08:51,375
Why save.
223
00:08:55,530 --> 00:08:58,850
You not fear.
224
00:08:58,850 --> 00:09:01,580
You strong heart.
225
00:09:01,580 --> 00:09:05,300
And it was compelling enough
that Fox was willing to pull
226
00:09:05,300 --> 00:09:10,520
the trigger on a $200-plus
million project that then went
227
00:09:10,520 --> 00:09:13,160
somewhat over that budget.
228
00:09:13,160 --> 00:09:14,660
And it was three
years later that we
229
00:09:14,660 --> 00:09:19,900
had the first shot that really
proved that we could do it.
230
00:09:19,900 --> 00:09:21,030
So think about that.
231
00:09:21,030 --> 00:09:26,697
We had our proof of
concept done in 2005.
232
00:09:26,697 --> 00:09:28,530
We got the first shot
that we could actually
233
00:09:28,530 --> 00:09:32,090
use in the movie in 2008.
234
00:09:32,090 --> 00:09:35,370
And everything in between
was a gigantic leap of faith.
235
00:09:35,370 --> 00:09:37,370
And it's easy to look
back on it and say, oh, it
236
00:09:37,370 --> 00:09:38,390
was all a no-brainer.
237
00:09:38,390 --> 00:09:40,250
Come on, that world building,
those great designs,
238
00:09:40,250 --> 00:09:40,875
all that stuff.
239
00:09:40,875 --> 00:09:41,690
Yeah.
240
00:09:41,690 --> 00:09:42,440
That's true.
241
00:09:42,440 --> 00:09:45,150
World building was
a big part of it.
242
00:09:45,150 --> 00:09:46,980
This was the most
important part,
243
00:09:46,980 --> 00:09:49,470
because the audience enters
the film through the eyes
244
00:09:49,470 --> 00:09:51,730
and hearts of the characters.
245
00:09:51,730 --> 00:09:53,743
So we had to have
the characters.
246
00:09:53,743 --> 00:09:56,581
[MUSIC PLAYING]
247
00:09:59,420 --> 00:10:03,350
The key for me is, if the
actor did the performance,
248
00:10:03,350 --> 00:10:07,170
we've captured it, it's going
to be in the character later.
249
00:10:07,170 --> 00:10:10,460
If we want to run a
little splitscreen here
250
00:10:10,460 --> 00:10:13,070
where you can see
Zoe Saldana in one
251
00:10:13,070 --> 00:10:17,020
of her big emotional moments in
the film with Sam Worthington.
252
00:10:17,020 --> 00:10:18,270
Emotional moment for him, too.
253
00:10:18,270 --> 00:10:19,730
I trusted you!
254
00:10:19,730 --> 00:10:20,480
Trust me now.
255
00:10:20,480 --> 00:10:20,980
Please.
256
00:10:20,980 --> 00:10:27,851
[SPEAKING IN NA'VI] That is
why you not Na'vi [INAUDIBLE]..
257
00:10:27,851 --> 00:10:30,790
(ANGRILY) You will never
be one of The People!
258
00:10:30,790 --> 00:10:32,070
He tried to stop them--
259
00:10:32,070 --> 00:10:32,610
ZOE SALDANA (AS NEYTIRI):
[SPEAKING IN NA'VI]
260
00:10:32,610 --> 00:10:33,390
Neytiri, please!
261
00:10:33,390 --> 00:10:35,038
Please!
262
00:10:35,038 --> 00:10:36,580
JAMES CAMERON: So
she's freaking out,
263
00:10:36,580 --> 00:10:38,310
and it's the end of
the relationship.
264
00:10:38,310 --> 00:10:40,740
But I think what's
interesting here
265
00:10:40,740 --> 00:10:45,490
is to look at how
closely we were
266
00:10:45,490 --> 00:10:46,930
able to capture her expression.
267
00:10:46,930 --> 00:10:49,990
It's not perfect, because
her character is not exactly
268
00:10:49,990 --> 00:10:52,660
physiologically
identical to her.
269
00:10:52,660 --> 00:10:53,953
Her nose is very different.
270
00:10:53,953 --> 00:10:55,120
Her eyes are very different.
271
00:10:55,120 --> 00:10:57,010
Her forehead's very different.
272
00:10:57,010 --> 00:10:58,480
But her mouth is very similar.
273
00:10:58,480 --> 00:11:01,720
And you can see that the
performance capture technique
274
00:11:01,720 --> 00:11:05,170
captured what she did.
275
00:11:05,170 --> 00:11:08,650
It's an actor-driven process.
276
00:11:08,650 --> 00:11:11,920
Everything that she does,
everything that Sam does,
277
00:11:11,920 --> 00:11:15,042
translates into the character.
278
00:11:15,042 --> 00:11:16,250
And this is what we hope for.
279
00:11:16,250 --> 00:11:18,200
This is what we worked toward.
280
00:11:18,200 --> 00:11:19,880
And, of course, this
is the head rig.
281
00:11:19,880 --> 00:11:24,620
And this facial performance is
being captured by this camera
282
00:11:24,620 --> 00:11:26,270
and by a video feed,
a live video feed,
283
00:11:26,270 --> 00:11:30,020
we broadcast it off the
body to a record deck.
284
00:11:30,020 --> 00:11:33,170
So you can see that he blinks,
his lips are pulled back
285
00:11:33,170 --> 00:11:35,540
in that exact moment
exactly the same way.
286
00:11:35,540 --> 00:11:39,080
It maps through frame
by frame by frame.
287
00:11:39,080 --> 00:11:43,630
And we don't take
liberties with this.
288
00:11:43,630 --> 00:11:45,790
It took us a long
time to figure out
289
00:11:45,790 --> 00:11:49,780
how to create the
neuromuscular rig, which we now
290
00:11:49,780 --> 00:11:51,970
call the algorithm for Zoe.
291
00:11:51,970 --> 00:11:53,830
Because Zoe's face
did a lot of things
292
00:11:53,830 --> 00:11:56,680
that a lot of people's
faces don't do.
293
00:11:56,680 --> 00:11:58,600
The way in which her
lips would curl out,
294
00:11:58,600 --> 00:12:00,460
and her-- the tension,
and the way her lips
295
00:12:00,460 --> 00:12:03,130
would thin out over her
teeth in a very feral way.
296
00:12:03,130 --> 00:12:04,750
It's a very expressive face.
297
00:12:04,750 --> 00:12:08,680
And we needed to figure out how
to get the pseudo-muscles that
298
00:12:08,680 --> 00:12:12,100
exist entirely
mathematically to do that.
299
00:12:12,100 --> 00:12:13,540
And that was our goal.
300
00:12:13,540 --> 00:12:17,210
And that's what the technical
teams at WETA were able to do.
301
00:12:17,210 --> 00:12:20,230
But the hand motion,
everything, you
302
00:12:20,230 --> 00:12:22,750
can see how it maps
frame by frame.
303
00:12:22,750 --> 00:12:26,473
And there's absolutely zero
compromise in that system.
304
00:12:26,473 --> 00:12:28,390
You have to get the
performance in the moment.
305
00:12:28,390 --> 00:12:30,100
You can't make it better later.
306
00:12:30,100 --> 00:12:31,475
There are a couple
of times where
307
00:12:31,475 --> 00:12:32,980
we tried to alter
a moment slightly,
308
00:12:32,980 --> 00:12:35,230
maybe to have a tiny bit more
of a smile or a little bit
309
00:12:35,230 --> 00:12:35,800
more fear.
310
00:12:35,800 --> 00:12:38,800
[SCOFFS] It was a
horrible failure.
311
00:12:38,800 --> 00:12:41,880
It has to be the
actor has to do it.
312
00:12:41,880 --> 00:12:43,800
And then it's just
up to us to honor it
313
00:12:43,800 --> 00:12:47,010
in the downstream pipeline,
which takes about a year.
314
00:12:47,010 --> 00:12:49,170
From the moment you
capture it, you're at least
315
00:12:49,170 --> 00:12:52,470
a year away from
seeing a shot, which
316
00:12:52,470 --> 00:12:54,270
is a thing in and of itself.
317
00:12:54,270 --> 00:12:56,255
I don't recommend making
a movie like "Avatar,"
318
00:12:56,255 --> 00:12:57,630
but it's good for
people to know,
319
00:12:57,630 --> 00:13:01,160
even if they're just starting
out, what can be done.
320
00:13:01,160 --> 00:13:03,980
[MUSIC PLAYING]
321
00:13:07,270 --> 00:13:12,038
If you're shooting a scene with
camera, a photographic scene,
322
00:13:12,038 --> 00:13:14,330
you're going to shoot a
master, some kind of wide shot.
323
00:13:14,330 --> 00:13:16,950
And you're going to probably
shoot some kind of a two-shot,
324
00:13:16,950 --> 00:13:19,703
or a closer shot that
ties both actors in.
325
00:13:19,703 --> 00:13:21,870
And then you're going to
shoot an over-the-shoulder.
326
00:13:21,870 --> 00:13:24,412
And then you're probably going
to shoot either a tighter over
327
00:13:24,412 --> 00:13:26,280
or a tight single
on this side of it.
328
00:13:26,280 --> 00:13:27,810
Then you're going to turn
the lighting all around,
329
00:13:27,810 --> 00:13:29,290
you're going to come
around over here,
330
00:13:29,290 --> 00:13:31,530
then you're going to shoot
the corresponding tight over,
331
00:13:31,530 --> 00:13:32,905
and then probably
a tight single.
332
00:13:32,905 --> 00:13:34,050
That's classic coverage.
333
00:13:34,050 --> 00:13:35,800
A lot of filmmakers
don't like to do that,
334
00:13:35,800 --> 00:13:36,930
because it's too classic.
335
00:13:36,930 --> 00:13:37,782
But it works.
336
00:13:37,782 --> 00:13:39,240
And there are
reasons why it works,
337
00:13:39,240 --> 00:13:41,305
and people still
do it all the time.
338
00:13:41,305 --> 00:13:42,930
But let's just take
that as an example,
339
00:13:42,930 --> 00:13:44,680
because it is classic coverage.
340
00:13:44,680 --> 00:13:47,550
So that's six setups.
341
00:13:47,550 --> 00:13:53,820
Across those six setups, maybe
you do five takes of each shot,
342
00:13:53,820 --> 00:13:57,473
just for the actors, just to
figure out in the master what
343
00:13:57,473 --> 00:13:59,640
you're doing, figure out
in the two-shot the timing,
344
00:13:59,640 --> 00:14:01,655
figure out in the
close-up the emotion,
345
00:14:01,655 --> 00:14:03,030
the moment that
somebody is going
346
00:14:03,030 --> 00:14:06,900
to cry or have some big
reaction or whatever it is.
347
00:14:06,900 --> 00:14:07,890
The actors need time.
348
00:14:07,890 --> 00:14:09,307
They need to work
their way to it.
349
00:14:09,307 --> 00:14:11,610
And so let's say five--
let's say five takes.
350
00:14:11,610 --> 00:14:13,710
So that's six setups
times five takes.
351
00:14:13,710 --> 00:14:19,090
That's 30 times that the actors
have to do it the same way,
352
00:14:19,090 --> 00:14:23,000
so that you can cut it together.
353
00:14:23,000 --> 00:14:25,650
On the other hand, I could do--
354
00:14:25,650 --> 00:14:29,580
I think I did seven
takes of this scene.
355
00:14:29,580 --> 00:14:31,460
And none of them were the same.
356
00:14:31,460 --> 00:14:34,085
But it didn't matter,
because for this scene--
357
00:14:34,085 --> 00:14:36,350
I'm gonna show you something
really interesting here.
358
00:14:36,350 --> 00:14:37,933
And this is the
fundamental difference
359
00:14:37,933 --> 00:14:40,180
I think with the
performance capture.
360
00:14:40,180 --> 00:14:40,680
All right.
361
00:14:40,680 --> 00:14:42,420
So we start in a
fairly tight two-shot.
362
00:14:42,420 --> 00:14:44,295
They're both barely in
frame with each other,
363
00:14:44,295 --> 00:14:48,840
but it emphasizes the way
he's trying to hold on to her,
364
00:14:48,840 --> 00:14:50,380
and she's pushing him away.
365
00:14:50,380 --> 00:14:51,870
All right?
366
00:14:51,870 --> 00:14:55,370
Now we're going to
cut to her close-up.
367
00:14:55,370 --> 00:14:56,280
All right?
368
00:14:56,280 --> 00:14:57,942
That's a separate set up, right?
369
00:14:57,942 --> 00:14:59,900
Then we're going to
briefly be on that two-shot
370
00:14:59,900 --> 00:15:01,298
again while she slams him away.
371
00:15:01,298 --> 00:15:02,840
Then we're going to
continue with her
372
00:15:02,840 --> 00:15:04,528
in a slightly looser close-up.
373
00:15:04,528 --> 00:15:06,320
And then we're going
to cut to his reaction
374
00:15:06,320 --> 00:15:08,210
as he tries to explain it.
375
00:15:08,210 --> 00:15:11,690
And then we're going to
cut to a medium shot,
376
00:15:11,690 --> 00:15:13,700
so that we could see
her hand and her gesture
377
00:15:13,700 --> 00:15:15,020
as she pulls away.
378
00:15:15,020 --> 00:15:17,820
Again, I went wider here
to suggest distance.
379
00:15:17,820 --> 00:15:19,483
She's already pulled
away from him.
380
00:15:19,483 --> 00:15:20,900
We see a little
bit of her mother.
381
00:15:20,900 --> 00:15:22,108
Her mother tries to stop her.
382
00:15:22,108 --> 00:15:24,070
She pushes the mother away.
383
00:15:24,070 --> 00:15:25,000
Guess what?
384
00:15:25,000 --> 00:15:27,290
It's all from the same take.
385
00:15:27,290 --> 00:15:30,350
All those shots are from
the same take-- take seven.
386
00:15:30,350 --> 00:15:32,900
I happen to remember that,
because I was very happy when
387
00:15:32,900 --> 00:15:35,400
I realized that there was one
take that had everything in it
388
00:15:35,400 --> 00:15:37,705
that I wanted, where Sam
was great, Zoe was great,
389
00:15:37,705 --> 00:15:38,580
everything was great.
390
00:15:38,580 --> 00:15:40,850
We had good reference coverage.
391
00:15:40,850 --> 00:15:42,380
So all of those
shots were extracted
392
00:15:42,380 --> 00:15:43,940
from the same captured take.
393
00:15:43,940 --> 00:15:45,870
Now we had done other takes.
394
00:15:45,870 --> 00:15:47,100
I didn't like them as well.
395
00:15:47,100 --> 00:15:50,147
And so I didn't want to
shoot my virtual camera
396
00:15:50,147 --> 00:15:51,230
coverage from those takes.
397
00:15:51,230 --> 00:15:53,260
I did it all from that one take.
398
00:15:53,260 --> 00:15:54,260
We don't always do that.
399
00:15:54,260 --> 00:15:57,040
But it shows what's possible.
29098
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.