Torsten Reil: Using biology to make better animation

33,203 views ใƒป 2008-07-08

TED


์•„๋ž˜ ์˜๋ฌธ์ž๋ง‰์„ ๋”๋ธ”ํด๋ฆญํ•˜์‹œ๋ฉด ์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค.

๋ฒˆ์—ญ: Jong-In Won ๊ฒ€ํ† : InHyuk Song
00:15
I'm going to talk about a technology that we're developing at Oxford now,
0
15160
4000
์ €๋Š” ์ง€๊ธˆ ์˜ฅ์Šคํฌ๋“œ์—์„œ ๊ฐœ๋ฐœ์ค‘์ธ ๊ธฐ์ˆ ์— ๊ด€ํ•ด ์–˜๊ธฐํ• ๊นŒ ํ•ฉ๋‹ˆ๋‹ค.
00:19
that we think is going to change the way that
1
19160
3000
์ด ๊ธฐ์ˆ ์€ ์•ž์œผ๋กœ ์ปดํ“จํ„ฐ ๊ฒŒ์ž„๊ณผ ํ• ๋ฆฌ์šฐ๋“œ ์˜ํ™”๊ฐ€
00:22
computer games and Hollywood movies are being made.
2
22160
3000
๋งŒ๋“ค์–ด ์ง€๋Š” ๋ฐฉ๋ฒ•์„ ๋ฐ”๊ฟ€๊บผ๋ผ๊ณ  ์ƒ๊ฐํ•ฉ๋‹ˆ๋‹ค.
00:26
That technology is simulating humans.
3
26160
3000
์ด ๊ธฐ์ˆ ์€ ๋ฐ”๋กœ ์ธ๊ฐ„์˜ ํ–‰๋™์„ ํ‰๋‚ด๋‚ด๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
00:29
It's simulated humans with a simulated body
4
29160
3000
์ธ๊ฐ„์˜ ๋ชธ๊ณผ ๊ทธ ๋ชธ์„ ์ œ์–ดํ•˜๋Š” ์‹ ๊ฒฝ์„ ๊ฐ€์ง„
00:32
and a simulated nervous system to control that body.
5
32160
3000
์• ๋‹ˆ๋ฉ”์ด์…˜์† ์ธ๊ฐ„์— ๊ด€ํ•œ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
00:36
Now, before I talk more about that technology,
6
36160
3000
์ด ๊ธฐ์ˆ ์— ๋Œ€ํ•ด ์–˜๊ธฐํ•˜๊ธฐ ์ „์—,
00:39
let's have a quick look at what human characters look like
7
39160
3000
์ปดํ“จํ„ฐ ๊ฒŒ์ž„์— ๋‚˜์˜ค๋Š” ์‚ฌ๋žŒ์€ ์–ด๋–ป๊ฒŒ ๋ณด์ด๋Š”์ง€
00:42
at the moment in computer games.
8
42160
3000
์ž ์‹œ ๋ณด์‹œ์ฃ .
00:45
This is a clip from a game called "Grand Theft Auto 3."
9
45160
3000
์ด๊ฒƒ์€ Grand Theft Auto3๋ผ๋Š” ๊ฒŒ์ž„์— ๋‚˜์˜ค๋Š” ํ•œ ์žฅ๋ฉด์ž…๋‹ˆ๋‹ค.
00:48
We already saw that briefly yesterday.
10
48160
2000
์ €๋Š” ์–ด์ œ ์ž ๊น ๋ดค๊ตฌ์š”.
00:50
And what you can see is -- it is actually a very good game.
11
50160
3000
์ง€๊ธˆ ๋ณด์‹œ๋Š” ๊ฒƒ์€ ์‚ฌ์‹ค ์ƒ๋‹นํžˆ ์ข‹์€ ๊ฒŒ์ž„์ž…๋‹ˆ๋‹ค.
00:53
It's one of the most successful games of all time.
12
53160
3000
์ตœ๊ณ ์˜ ๊ฒŒ์ž„์ค‘ ํ•˜๋‚˜์ฃ .
00:56
But what you'll see is that all the animations in this game are very repetitive.
13
56160
4000
๊ทธ๋Ÿฐ๋ฐ ๋ณด์‹œ๋ฉด ์•„์‹œ๊ฒ ์ง€๋งŒ ๊ฒŒ์ž„์†์˜ ๋ชจ๋“  ์• ๋‹ˆ๋ฉ”์ด์…˜์ด ๋งค์šฐ ๋ฐ˜๋ณต์ ์ž…๋‹ˆ๋‹ค.
01:00
They pretty much look the same.
14
60160
2000
๋ชจ๋‘๋“ค ๋น„์Šทํ•˜๊ฒŒ ๋ณด์ด์ฃ .
01:02
I've made him run into a wall here, over and over again.
15
62160
3000
์ด์‚ฌ๋žŒ์„ ๋ฒฝ์— ๋ถ€๋”ช์น˜๊ฒŒ ํ•ด ๋ณด์•˜์Šต๋‹ˆ๋‹ค. ๋ฐ˜๋ณตํ•ด์„œ.
01:05
And you can see he looks always the same.
16
65160
2000
์–ธ์ œ๋‚˜ ๊ฐ™์€ ์›€์ง์ž„์„ ๋ณด์‹ค ์ˆ˜ ์žˆ์ฃ .
01:07
The reason for that is that these characters
17
67160
3000
๊ทธ ์ด์œ ๋Š” ๋ฐ”๋กœ ์—ฌ๊ธฐ ๊ฒŒ์ž„์† ์บ๋ฆญํ„ฐ๋“ค์ด
01:10
are actually not real characters.
18
70160
2000
์‹ค์ œ ์‚ฌ๋žŒ์ด ์•„๋‹ˆ๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค.
01:12
They are a graphical visualization of a character.
19
72160
4000
์‚ฌ๋žŒ์„ ์ปดํ“จํ„ฐ ๊ทธ๋ž˜ํ”ฝ์œผ๋กœ ์‹œ๊ฐํ™”ํ•œ ๊ฒƒ ๋ฟ์ž…๋‹ˆ๋‹ค.
01:16
To produce these animations, an animator at a studio has to anticipate
20
76160
5000
์ด๋Ÿฐ ์˜์ƒ์„ ์ŠคํŠœ๋””์˜ค์—์„œ ๋งŒ๋“ค๊ธฐ ์œ„ํ•ด์„œ๋Š”
01:21
what's going to happen in the actual game,
21
81160
3000
์ œ์ž‘์ž๋Š” ์‹ค์ œ ๊ฒŒ์ž„์—์„œ ์–ด๋–ค ์ผ๋“ค์ด ๋ฒŒ์–ด์งˆ์ง€ ์˜ˆ์ธกํ•˜๊ณ ,
01:24
and then has to animate that particular sequence.
22
84160
3000
์—ฐ์†๋™์ž‘๋“ค์„ ๊ทธ๋ ค๋‚ด์•ผ ํ•˜์ฃ .
01:27
So, he or she sits down, animates it, and tries to anticipate what's going to happen,
23
87160
4000
๊ทธ๋ž˜์„œ ์ œ์ž‘์ž๋Š” ์ž๋ฆฌ์— ์•‰์•„ ์–ด๋–ค์ผ์ด ์ƒ๊ธธ์ง€ ๊ณ ๋ฏผํ•˜์—ฌ ๊ทธ๋ ค๋ณด๊ณ 
01:31
and then these particular animations are just played back
24
91160
3000
๊ทธ๋ ‡๊ฒŒ ๋งŒ๋“ค์–ด์ง„ ์˜์ƒ์ด ๊ฒŒ์ž„์†์—์„œ ํŠน์ •์‹œ์ ์—
01:34
at appropriate times in the computer game.
25
94160
3000
๋ฐ˜๋ณต์ ์œผ๋กœ ์žฌํ˜„๋˜๋Š” ๊ฒƒ ๋ฟ์ด์ฃ .
01:37
Now, the result of that is that you can't have real interactivity.
26
97160
5000
ํ˜„์‹ค๊ฐ ์žˆ๋Š” ์ƒํ˜ธ์ž‘์šฉ์ด ์—†๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
01:42
All you have is animations that are played back
27
102160
3000
๋˜‘๊ฐ™์€ ๋ชจ์Šต๋“ค์ด ๊ณ„์†์ ์œผ๋กœ ๋ฐ˜๋ณต๋˜๋Š”
01:45
at more or less the appropriate times.
28
105160
2000
๋ชจ์Šต๋“ค๋งŒ ๋ณด๊ฒŒ๋˜๋Š” ๊ฑฐ์ฃ 
01:47
It also means that games aren't really going to be as surprising as they could be,
29
107160
5000
์ด๋Ÿฐ์ ๋“ค๋•Œ๋ฌธ์— ๊ฒŒ์ž„๋“ค์ด ๋ณ„๋กœ ๋†€๋ผ์šธ๊ฒŒ ์—†๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
01:52
because you only get out of it, at least in terms of the character,
30
112160
3000
์™œ๋ƒํ•˜๋ฉด ์• ๋‹ˆ๋ฉ”์ด์…˜์† ์‚ฌ๋žŒ์€
01:55
what you actually put into it.
31
115160
2000
์ž…๋ ฅํ•œ ๋Œ€๋กœ๋งŒ ์›€์ง์ด๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค.
01:57
There's no real emergence there.
32
117160
2000
๊ธ‰๋ฐ•ํ•œ๊ฒŒ ์—†๋‹ค๊ณ  ํ• ๊นŒ์š”.
01:59
And thirdly, as I said, most of the animations are very repetitive because of that.
33
119160
4000
๊ทธ๋ž˜์„œ ๊ฒฐ๊ตญ ๋ฐ˜๋ณต์ ์ธ ํ–‰๋™๋งŒ ํ•˜๊ฒŒ ๋˜์ฃ .
02:03
Now, the only way to get around that
34
123160
2000
์ด๊ฑธ ํ•ด๊ฒฐํ•˜๋Š” ์œ ์ผํ•œ ๋ฐฉ๋ฒ•์€
02:05
is to actually simulate the human body
35
125160
2000
๋ฐ”๋กœ ์‚ฌ๋žŒ์˜ ๋ชธ๊ณผ
02:07
and to simulate that bit of the nervous system of the brain that controls that body.
36
127160
5000
๊ทธ ๋ชธ์„ ์ œ์–ดํ•˜๋Š” ๋‡Œ์‹ ๊ฒฝ์„ ํ‰๋‚ด๋‚ด๋Š” ์ˆ˜๋ฐ–์— ์—†์Šต๋‹ˆ๋‹ค.
02:12
And maybe, if I could have you for a quick demonstration
37
132160
3000
์ฐจ์ด๊ฐ€ ๋ญ”์ง€ ์—ฌ๋Ÿฌ๋ถ„๊ป˜ ์ž ๊น ๋ณด์—ฌ๋“œ๋ฆฌ์ž๋ฉด,
02:15
to show what the difference is --
38
135160
2000
๊ทธ ์ฐจ์ด๋Š”
02:17
because, I mean, it's very, very trivial.
39
137160
4000
์•„์ฃผ ์‚ฌ์†Œํ•œ ๊ฒ๋‹ˆ๋‹ค.
02:21
If I push Chris a bit, like this, for example, he'll react to it.
40
141160
3000
์ œ๊ฐ€ Chirs๋ฅผ ์กฐ๊ธˆ ๋ฐ€๋ฉด ๋ฐ˜์‘ํ•˜์ฃ .
02:24
If I push him from a different angle, he'll react to it differently,
41
144160
3000
์ œ๊ฐ€ ๋‹ค๋ฅธ ๋ฐฉํ–ฅ์—์„œ ๋ฐ€๋ฉด ๋‹ค๋ฅด๊ฒŒ ๋ฐ˜์‘ํ•˜์ฃ 
02:27
and that's because he has a physical body,
42
147160
2000
๊ทธ๊ฑด Chris๊ฐ€ ์‹ค์ œ ๋ชธ์„ ๊ฐ€์ง€๊ณ 
02:29
and because he has the motor skills to control that body.
43
149160
3000
๊ทธ ๋ชธ์„ ํ†ต์ œํ•˜๋Š” ๊ทผ์œก๋„ ๊ฐ€์ง€๊ณ  ์žˆ๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค.
02:32
It's a very trivial thing.
44
152160
2000
์•„์ฃผ ์‚ฌ์†Œํ•˜์ง€๋งŒ
02:34
It's not something you get in computer games at the moment, at all.
45
154160
2000
๊ฒŒ์ž„์†์—์„œ๋Š” ์ „ํ˜€ ์ฐพ์•„๋ณผ ์ˆ˜๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค.
02:36
Thank you very much. Chris Anderson: That's it?
46
156160
2000
๊ณ ๋งˆ์™€์š”. ํฌ๋ฆฌ์Šค์•ค๋”์Šจ: ์ด๊ฒŒ ๋‹ค์ธ๊ฐ€์š”?
02:38
Torsten Reil: That's it, yes.
47
158160
2000
Torsten: ๋„ค ๊ทธ๊ฒŒ ๋‹ค์ž…๋‹ˆ๋‹ค.
02:40
So, that's what we're trying to simulate --
48
160160
1000
์šฐ๋ฆฌ๊ฐ€ ํ‰๋‚ด๋‚ด๊ณ ์ž ํ•˜๋Š”๊ฒƒ์€
02:41
not Chris specifically, I should say, but humans in general.
49
161160
4000
Chris๊ฐ€ ์•„๋‹ˆ๋ผ ์ผ๋ฐ˜์ ์ธ ์‚ฌ๋žŒ์ด๋ผ๊ณ  ํ•  ์ˆ˜ ์žˆ์ฃ .
02:46
Now, we started working on this a while ago at Oxford University,
50
166160
5000
์˜ฅ์Šคํฌ๋“œ ๋Œ€ํ•™์—์„œ ์–ผ๋งˆ์ „๋ถ€ํ„ฐ ์ด ์—ฐ๊ตฌ๋ฅผ
02:51
and we tried to start very simply.
51
171160
2000
์•„์ฃผ ๊ฐ„๋‹จํ•œ๊ฒƒ ๋ถ€ํ„ฐ ์‹œ์ž‘ํ–ˆ์Šต๋‹ˆ๋‹ค.
02:53
What we tried to do was teach a stick figure how to walk.
52
173160
3000
๋ง‰๋Œ€๊ธฐ ์บ๋ฆญํ„ฐ์—๊ฒŒ ๊ฑท๋Š” ๋ฐฉ๋ฒ•์„ ๊ฐ€๋ฅด์ณค๋ณด์•˜์Šต๋‹ˆ๋‹ค.
02:56
That stick figure is physically stimulated. You can see it here on the screen.
53
176160
3000
์—ฌ๊ธฐ ์Šคํฌ๋ฆฐ์† ๋ง‰๋Œ€๊ธฐ๋Š” ์‹ค์ œ ์‚ฌ๋žŒ์„ ํ‰๋‚ด๋‚ด๊ณ  ์žˆ๋Š”๋ฐ ๋ณด์ด์‹œ์ฃ .
02:59
So, it's subject to gravity, has joints, etc.
54
179160
3000
์ค‘๋ ฅ์„ ๋ฐ›์œผ๋ฉฐ, ๊ด€์ ˆ๋“ค๋„ ์žˆ์ฃ .
03:02
If you just run the simulation, it will just collapse, like this.
55
182160
3000
์‹œ๋ฎฌ๋ ˆ์ด์…˜์„ ๋Œ๋ฆฌ๋ฉด ๋ง‰๋Œ€๊ธฐ๋Š” ์ด๋ ‡๊ฒŒ ์ฃผ์ €์•‰์Šต๋‹ˆ๋‹ค.
03:05
The tricky bit is now to put an AI controller in it
56
185160
4000
์ด์ œ ์ข€ ๋ณต์žกํ•œ๋ฐ ์ธ๊ณต์ง€๋Šฅ ์ œ์–ด๊ธฐ๋Šฅ์„ ์ถ”๊ฐ€ํ•˜๋ฉด
03:09
that actually makes it work.
57
189160
2000
์‹ค์ œ ๊ฑธ์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
03:11
And for that, we use the neural network, which we based on
58
191160
3000
์ด๋ฅผ ์œ„ํ•ด์„œ ์‹ ๊ฒฝํšŒ๋กœ๋ฅผ ํ™œ์šฉํ–ˆ๋Š”๋ฐ,
03:14
that part of the nervous system that we have in our spine
59
194160
2000
์šฐ๋ฆฌ ์ฒ™์ถ”์— ์žˆ๋Š” ์‹ ๊ฒฝ์˜ ์ผ๋ถ€๋กœ
03:16
that controls walking in humans.
60
196160
2000
๊ฑท๋Š” ๋™์ž‘์„ ํ†ต์ œํ•˜์ฃ .
03:18
It's called the central pattern generator.
61
198160
2000
์†Œ์œ„ ์ค‘์ถ”์œ ํ˜•๋ฐœ์ƒ๊ธฐ(CPG)๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
03:20
So, we simulated that as well, and then the really tricky bit
62
200160
3000
๊ทธ๊ฒƒ๋„ ํ‰๋‚ด๋‚ด๋„๋ก ํ–ˆ๋Š”๋ฐ, ์–ด๋ ค์šด ์ ์€
03:23
is to teach that network how to walk.
63
203160
2000
์‹ ๊ฒฝํšŒ๋กœ๊ฐ€ ๊ฑท๋„๋ก ๊ฐ€๋ฅด์น˜๋Š” ๊ฒ๋‹ˆ๋‹ค.
03:25
For that we used artificial evolution -- genetic algorithms.
64
205160
4000
์ด๋ฅผ ์œ„ํ•ด์„œ ์œ ์ „์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ์ด์šฉํ•˜์—ฌ ์ธ๊ณต์ ์œผ๋กœ ์ง„ํ™”ํ•˜๋„๋ก ํ–ˆ์ฃ 
03:29
We heard about those already yesterday,
65
209160
2000
์–ด์ œ ์ด ๋‚ด์šฉ์€ ์ด๋ฏธ ๋“ค์—ˆ๊ณ ,
03:31
and I suppose that most of you are familiar with that already.
66
211160
3000
์•„๋งˆ ๋Œ€๋ถ€๋ถ„ ์ต์ˆ™ํ•˜์‹ค ๋“ฏ ํ•ฉ๋‹ˆ๋‹ค.
03:34
But, just briefly, the concept is that
67
214160
2000
๊ฐ„๋žตํžˆ ๊ฐœ๋…์„ ์„ค๋ช…ํ•˜์ž๋งŒ
03:36
you create a large number of different individuals --
68
216160
3000
๋งŽ์€ ์ธ๊ณต์˜ ์‚ฌ๋žŒ์„ ๋งŒ๋“ค๊ณ 
03:39
neural networks, in this case --
69
219160
2000
์—ฌ๊ธฐ์„œ ์‹ ๊ฒฝํšŒ๋กœ๋Š”
03:41
all of which are random at the beginning.
70
221160
2000
์ž„์˜๋กœ ์‹œ์ž‘ํ•ฉ๋‹ˆ๋‹ค.
03:43
You hook these up -- in this case, to the virtual muscles
71
223160
2000
์ด์ œ ๊ฐ€์ƒ์˜ ๋‘๋ฐœ๋‹ฌ๋ฆฐ ๋ชจ์–‘์˜
03:45
of that two-legged creature here --
72
225160
3000
๊ทผ์œก์—๋‹ค๊ฐ€ ๊ฑธ์–ด์ฃผ๊ณ 
03:48
and hope that it does something interesting.
73
228160
3000
๋ญ”๊ฐ€ ์žฌ๋ฏธ๋‚œ ์ผ์ด ๋ฒŒ์–ด์ง€๊ธฐ๋ฅผ ๊ธฐ๋Œ€ํ•ฉ๋‹ˆ๋‹ค.
03:51
At the beginning, they're all going to be very boring.
74
231160
2000
์ฒ˜์Œ์—๋Š” ์ •๋ง ์ง€๋ฃจํ•˜์ฃ 
03:53
Most of them won't move at all,
75
233160
2000
๋Œ€๋ถ€๋ถ„์€ ์›€์ง์ด์ง€๋„ ์•Š๊ณ 
03:55
but some of them might make a tiny step.
76
235160
2000
๊ฐ„ํ˜น ์ผ๋ถ€๋Š” ํ•œ๋ฐœ์ง์„ ๋–ผ๊ธฐ๋„ ํ•ฉ๋‹ˆ๋‹ค.
03:57
Those are then selected by the algorithm,
77
237160
2000
์•Œ๊ณ ๋ฆฌ์ฆ˜์— ๋”ฐ๋ผ ์ผ๋ถ€๋ฅผ ์„ ํƒํ•ด์„œ
03:59
reproduced with mutation and recombinations to introduce sex as well.
78
239160
4000
๋ณต์ œ์™€ ์žฌ์กฐํ•ฉ์„ ํ†ตํ•ด ์žฌ์ƒ์‚ฐํ•˜๊ณ  ์„ฑ๋ณ„๊นŒ์ง€ ๋ถ€์—ฌํ•ฉ๋‹ˆ๋‹ค.
04:03
And you repeat that process over and over again,
79
243160
2000
๊ทธ๋ฆฌ๊ณ  ๋‹ค์‹œ ๊ทธ ๊ณผ์ •์„ ๊ณ„์†ํ•ด์„œ ๋ฐ˜๋ณตํ•˜๋Š”๋ฐ
04:05
until you have something that walks --
80
245160
2000
์–ธ์ œ๊นŒ์ง€ ํ•˜๋ƒ๋ฉด
04:07
in this case, in a straight line, like this.
81
247160
2000
์ด๋ ‡๊ฒŒ ๋˜‘๋ฐ”๋กœ ๊ฑท๋Š” ๋…€์„์ด ๋‚˜์˜ฌ๋•Œ ๊นŒ์ง€ ํ•ฉ๋‹ˆ๋‹ค.
04:09
So that was the idea behind this.
82
249160
2000
์ด๊ฒŒ ๋ฐ”๋กœ ์˜ค๋Š˜ ์–˜๊ธฐํ•  ๋‚ด์šฉ์ž…๋‹ˆ๋‹ค.
04:11
When we started this, I set up the simulation one evening.
83
251160
3000
์ฒ˜์Œ ํ• ๋•Œ๋Š” ์ €๋…์— ์‹œ๋ฎฌ๋ ˆ์ด์…˜์„ ์ž‘๋™ํ•˜๊ณ ,
04:14
It took about three to four hours to run the simulation.
84
254160
3000
์„œ๋„ˆ์‹œ๊ฐ„ ๊ฑธ๋ฆฌ๋‹ˆ๊นŒ
04:17
I got up the next morning, went to the computer and looked at the results,
85
257160
4000
๋‹ค์Œ๋‚  ์•„์นจ ์ผ์–ด๋‚˜์„œ ์ปดํ“จํ„ฐ๋กœ ๊ฐ€์„œ ๊ฒฐ๊ณผ๋ฅผ ๋ด…๋‹ˆ๋‹ค.
04:21
and was hoping for something that walked in a straight line,
86
261160
3000
์ง€๊ธˆ ๋ฐฉ๊ธˆ ์ œ๊ฐ€ ๊ฑธ์€ ๊ฒƒ์ฒ˜๋Ÿผ ๋ญ”๊ฐ€๊ฐ€
04:24
like I've just demonstrated,
87
264160
2000
๋˜‘๋ฐ”๋กœ ๊ฑธ์–ด๊ฐ€๊ธฐ๋ฅผ ๊ธฐ๋Œ€ํ•˜๋ฉด์„œ ๋ง์ด์ฃ 
04:26
and this is what I got instead.
88
266160
2000
๊ทธ๋Ÿฐ๋ฐ ๋‚˜์˜จ ๊ฒƒ์€ ์ด๋ ‡์Šต๋‹ˆ๋‹ค.
04:28
(Laughter)
89
268160
10000
์›ƒ์Œ
04:38
So, it was back to the drawing board for us.
90
278160
3000
์ €ํฌ๋Š” ์ฒ˜์Œ๋ถ€ํ„ฐ ๋‹ค์‹œ ์‹œ์ž‘ํ•ด์„œ
04:42
We did get it to work eventually,
91
282160
3000
์—ฌ๊ธฐ์ €๊ธฐ ์ข€ ์กฐ์ •ํ•œ ํ›„์— ๊ฒฐ๊ตญ
04:45
after tweaking a bit here and there.
92
285160
2000
๊ฑท๋„๋ก ๋งŒ๋“ค์—ˆ์Šต๋‹ˆ๋‹ค.
04:47
And this is an example of a successful evolutionary run.
93
287160
3000
๊ทธ๋ฆฌ๊ณ  ์ด๊ฑด ์„ฑ๊ณต์ ์œผ๋กœ ์ง„ํ™”ํ•œ ์ƒ˜ํ”Œ์ž…๋‹ˆ๋‹ค.
04:50
So, what you'll see in a moment is a very simple biped
94
290160
3000
์ž ์‹œํ›„ ๋ณด์‹ค๊ฒƒ์€ ์ธ๊ณต์ง„ํ™”๊ณผ์ •์„ ํ†ตํ•ด ๊ฑท๋Š” ๊ฒƒ์„ ๋ฐฐ์šด
04:53
that's learning how to walk using artificial evolution.
95
293160
3000
๋‘๋ฐœ๋‹ฌ๋ฆฐ ์บ๋ฆญํ„ฐ ์ž…๋‹ˆ๋‹ค.
04:56
At the beginning, it can't walk at all,
96
296160
2000
์ฒ˜์Œ์—๋Š” ์ „ํ˜€ ๊ฑธ์„ ์ˆ˜ ์—†๋‹ค๊ฐ€
04:58
but it will get better and better over time.
97
298160
2000
์‹œ๊ฐ„์ด ๊ฐˆ์ˆ˜๋ก ์ ์  ์ข‹์•„์ง€๊ฒŒ ๋  ๊ฒ๋‹ˆ๋‹ค.
05:02
So, this is the one that can't walk at all.
98
302160
3000
์ด๊ฑด ์ „ํ˜€ ๊ฑธ์„ ์ˆ˜ ์—†๋Š” ๊ฑฐ๊ตฌ์š”
05:05
(Laughter)
99
305160
6000
(์›ƒ์Œ)
05:11
Now, after five generations of applying evolutionary process,
100
311160
3000
5์„ธ๋Œ€๊ฐ€ ์ง„ํ™”ํ•˜๋ฉด
05:14
the genetic algorithm is getting a tiny bit better.
101
314160
3000
์œ ์ „์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์•„์ฃผ ์กฐ๊ธˆ ํ–ฅ์ƒ๋ฉ๋‹ˆ๋‹ค.
05:17
(Laughter)
102
317160
8000
(์›ƒ์Œ)
05:25
Generation 10 and it'll take a few steps more --
103
325160
2000
10์„ธ๋Œ€๊ฐ€ ๋˜๋ฉด ๋ช‡ ๋ฐœ์ง ๊ฑท์ง€๋งŒ
05:31
still not quite there.
104
331160
2000
์•„์ง ๋ฉ€์—ˆ์ฃ 
05:34
But now, after generation 20, it actually walks in a straight line without falling over.
105
334160
5000
ํ•˜์ง€๋งŒ 20์„ธ๋Œ€๊ฐ€ ์ง€๋‚˜์„œ๋Š” ๋„˜์–ด์ง€์ง€ ์•Š๊ณ  ๋˜‘๋ฐ”๋กœ ๊ฑท๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
05:40
That was the real breakthrough for us.
106
340160
3000
๋†€๋ผ์šด ๋ฐœ์ „์ด์ฃ 
05:43
It was, academically, quite a challenging project,
107
343160
3000
ํ•™๋ฌธ์ ์œผ๋กœ๋„ ๋งค์šฐ ์–ด๋ ค์šด ์ผ์ด๊ธฐ๋„ ํ–ˆ์Šต๋‹ˆ๋‹ค.
05:46
and once we had reached that stage, we were quite confident
108
346160
3000
์—ฌ๊ธฐ๊นŒ์ง€ ๋ฐœ์ „ํ•˜์ž ์ž์‹ ๊ฐ์ด ๋“ค๋”๊ตฐ์š”
05:49
that we could try and do other things as well with this approach --
109
349160
3000
์‚ฌ๋žŒ์˜ ๋ชธ๊ณผ, ์‹ ๊ฒฝํšŒ๋กœ๋ฅผ ํ‰๋‚ด๋‚ด๋Š” ์ด ๋ฐฉ๋ฒ•์„
05:52
actually simulating the body
110
352160
2000
ํ™œ์šฉํ•˜์„œ
05:54
and simulating that part of the nervous system that controls it.
111
354160
3000
๋‹ค๋ฅธ ์ผ๋“ค๋„ ํ•  ์ˆ˜ ์žˆ๊ฒ ๋‹ค๋Š” ์ƒ๊ฐ์ด ๋“ค์—ˆ์Šต๋‹ˆ๋‹ค.
05:57
Now, at this stage, it also became clear that this could be very exciting
112
357160
3000
์ง€๊ธˆ ์ด๋‹จ๊ณ„ ๋งŒ์œผ๋กœ๋„ ์ปดํ“จํ„ฐ ๊ฒŒ์ž„์ด๋‚˜ ์˜จ๋ผ์ธ ์ƒ์—์„œ์˜
06:00
for things like computer games or online worlds.
113
360160
3000
์ผ๋“ค์ด ๋งค์šฐ ์žฌ๋ฏธ์žˆ์–ด ์งˆ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
06:03
What you see here is the character standing there,
114
363160
2000
์ง€๊ธˆ ๋ณด์‹œ๋Š” ๊ฒƒ์€ ํ•œ ์บ๋ฆญํ„ฐ๊ฐ€ ์„œ ์žˆ๋Š”๋ฐ,
06:05
and there's an obstacle that we put in its way.
115
365160
2000
์•ž์—๋‹ค๊ฐ€ ์žฅ์• ๋ฌผ์„ ๋†“์œผ๋ฉด
06:07
And what you see is, it's going to fall over the obstacle.
116
367160
5000
์žฅ์• ๋ฌผ์— ๊ฑธ๋ ค ๋„˜์–ด์ง€๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
06:12
Now, the interesting bit is, if I move the obstacle a tiny bit to the right,
117
372160
3000
์žฌ๋ฏธ์žˆ๋Š” ๊ฒƒ์€ ์žฅ์• ๋ฌผ์„ ์กฐ๊ธˆ ์šฐ์ธจ์œผ๋กœ ์˜ฎ๊ธฐ๋ฉด
06:15
which is what I'm doing now, here,
118
375160
2000
์ด๋ ‡๊ฒŒ ๋ง์ด์ฃ 
06:17
it will fall over it in a completely different way.
119
377160
4000
์ „ํ˜€ ๋‹ค๋ฅด๊ฒŒ ๋„˜์–ด์ง€๋Š” ๊ฒƒ์„ ๋ณผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
06:24
And again, if you move the obstacle a tiny bit, it'll again fall differently.
120
384160
5000
๊ทธ๋ฆฌ๊ณ  ๋‹ค์‹œ ์กฐ๊ธˆ ์›€์ง์ด๋ฉด, ์—ญ์‹œ ๋‹ค๋ฅด๊ฒŒ ๋„˜์–ด์ง‘๋‹ˆ๋‹ค.
06:29
(Laughter)
121
389160
2000
(์›ƒ์Œ)
06:31
Now, what you see, by the way, at the top there,
122
391160
2000
์ด๋ฒˆ์— ๋ณด์‹ค ์—ฌ๊ธฐ ์œ„์— ์žˆ๋Š” ๊ฒƒ์€
06:33
are some of the neural activations being fed into the virtual muscles.
123
393160
3000
์‹ ๊ฒฝํ™œ์„ฑํ™”๊ธฐ๋Šฅ์ด ์žˆ๋Š” ๊ฐ€์ƒ์˜ ๊ทผ์œก์ž…๋‹ˆ๋‹ค.
06:36
Okay. That's the video. Thanks.
124
396160
2000
๋„ค ๋ฐ”๋กœ ์ด๊ฒ๋‹ˆ๋‹ค. ๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
06:38
Now, this might look kind of trivial, but it's actually very important
125
398160
3000
๋ณ„๊ฑฐ ์—†์–ด๋ณด์ด์ง€๋งŒ ์‚ฌ์‹ค ๋งค์šฐ ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค.
06:41
because this is not something you get at the moment
126
401160
2000
์™œ๋ƒํ•˜๋ฉด ์ด์ œ๊นŒ์ง€ ์–ด๋– ํ•œ ๊ฐ€์ƒ์„ธ๊ณ„์—์„œ๋„
06:43
in any interactive or any virtual worlds.
127
403160
2000
์กด์žฌํ•˜์ง€ ์•Š์•˜์œผ๋‹ˆ๊น์š”
06:48
Now, at this stage, we decided to start a company and move this further,
128
408160
3000
์ด๋‹จ๊ณ„์—์„œ ์šฐ๋ฆฌ๋Š” ํšŒ์‚ฌ๋ฅผ ์„ธ์šฐ๊ณ  ์ข€๋” ์—ฐ๊ตฌํ•˜๊ธฐ๋กœ ํ–ˆ์Šต๋‹ˆ๋‹ค.
06:51
because obviously this was just a very simple, blocky biped.
129
411160
3000
์ด๊ฑด ์•„์ฃผ ๊ฐ„๋‹จํ•œ ์งค๋”ธ๋ง‰ํ•œ ๋‘๋ฐœ๋‹ฌ๋ฆฐ ์บ๋ฆญํ„ฐ์— ๋ถˆ๊ณผํ•˜์ง€๋งŒ์š”
06:54
What we really wanted was a full human body.
130
414160
2000
์šฐ๋ฆฌ๊ฐ€ ์›ํ–ˆ๋˜ ๊ฒƒ์€ ์‹ค์ œ ์‚ฌ๋žŒ๊ฐ™์€ ์บ๋ฆญํ„ฐ ์˜€๊ธฐ ๋•Œ๋ฌธ์—
06:56
So we started the company.
131
416160
1000
๋” ์—ฐ๊ตฌํ•˜๊ธฐ ์œ„ํ•ด ํšŒ์‚ฌ๋ฅผ ์„ค๋ฆฝํ–ˆ์Šต๋‹ˆ๋‹ค.
06:57
We hired a team of physicists, software engineers and biologists
132
417160
5000
๋ฌผ๋ฆฌํ•™์ž, ์†Œํ”„ํŠธ์›จ์–ด ์—”์ง€๋‹ˆ์–ด, ์ƒ๋ฌผํ•™์ž๋กœ ๊ตฌ์„ฑ๋œ ํŒ€์„
07:02
to work on this, and the first thing we had to work on
133
422160
3000
๊ณ ์šฉํ•˜๊ณ  ์ฒ˜์Œ์œผ๋กœ ํ•œ๊ฒƒ์ด ๋ฐ”๋กœ
07:05
was to create the human body, basically.
134
425160
4000
์ธ๊ฐ„์˜ ๋ชธ์„ ํ‰๋‚ด๋‚ด๋Š” ๊ฒƒ์ด์—ˆ์Šต๋‹ˆ๋‹ค.
07:09
It's got to be relatively fast, so you can run it on a normal machine,
135
429160
3000
์ด์ œ ๊ฝค ๋นจ๋ผ์„œ ์ผ๋ฐ˜ ์ปดํ“จํ„ฐ์—์„œ๋„ ์ž‘๋™์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š”๋ฐ
07:12
but it's got to be accurate enough, so it looks good enough, basically.
136
432160
3000
์‹ค์ œ์ฒ˜๋Ÿผ ๋ณด์ผ ์ •๋„๋กœ ์ •ํ™•ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
07:15
So we put quite a bit of biomechanical knowledge into this thing,
137
435160
3000
๊ทธ๋ž˜์„œ ์šฐ๋ฆฌ๋Š” ์ƒ๋‹นํ•œ ์ˆ˜์ค€์˜ ์ƒ๋ฌผ์—ญํ•™์  ์ง€์‹๋„ ์Ÿ์•„ ๋„ฃ์–ด
07:18
and tried to make it as realistic as possible.
138
438160
4000
๊ฐ€๋Šฅํ•œ ์‹ค์ œ์ฒ˜๋Ÿผ ๋ณด์ด๋„๋ก ํ–ˆ์Šต๋‹ˆ๋‹ค.
07:22
What you see here on the screen right now
139
442160
2000
์ง€๊ธˆ ์Šคํฌ๋ฆฐ์— ๋ณด์ด๋Š” ๊ฒƒ์€
07:24
is a very simple visualization of that body.
140
444160
2000
๊ทธ๋ ‡๊ฒŒ ๋งŒ๋“  ์‚ฌ๋žŒ์˜ ๊ฐ„๋‹จํ•œ ์˜์ƒ์ž…๋‹ˆ๋‹ค.
07:26
I should add that it's very simple to add things like hair, clothes, etc.,
141
446160
4000
๋จธ๋ฆฌ์นด๋ฝ์ด๋‚˜ ์˜ท์„ ์ž…ํžˆ๋Š” ๊ฒƒ์€ ๊ฐ„๋‹จํ•˜์ง€๋งŒ
07:30
but what we've done here is use a very simple visualization,
142
450160
3000
์—ฌ๊ธฐ์„œ๋Š” ์—ฌ๋Ÿฌ๋ถ„์ด ์›€์ง์ž„์— ์ง‘์ค‘ํ•  ์ˆ˜ ์žˆ๋„๋ก
07:33
so you can concentrate on the movement.
143
453160
2000
์ตœ๋Œ€ํ•œ ๋‹จ์ˆœํ™” ํ•˜์˜€์Šต๋‹ˆ๋‹ค.
07:35
Now, what I'm going to do right now, in a moment,
144
455160
3000
์ด์ œ ์ œ๊ฐ€ ํ•  ๊ฒƒ์€
07:38
is just push this character a tiny bit and we'll see what happens.
145
458160
3000
์บ๋ฆญํ„ฐ๋ฅผ ์กฐ๊ธˆ ๋ฐ€๋ฉด, ์–ด๋–ค์ผ์ด ๋ฒŒ์–ด์ง€๋Š”์ง€ ๋ณด์‹œ๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
07:46
Nothing really interesting, basically.
146
466160
2000
๋ณ„๋‹ค๋ฅธ ๊ฒƒ์€ ์—†์ฃ 
07:48
It falls over, but it falls over like a rag doll, basically.
147
468160
3000
์ฒœ์œผ๋กœ ๋งŒ๋“  ์ธํ˜•์ฒ˜๋Ÿผ ๋„˜์–ด์ง€์ฃ 
07:51
The reason for that is that there's no intelligence in it.
148
471160
3000
๊ทธ ์ด์œ ๋Š” ์•„์ง ์ง€๋Šฅ์ด ์—†์–ด์„œ ์ž…๋‹ˆ๋‹ค.
07:54
It becomes interesting when you put artificial intelligence into it.
149
474160
4000
์ธ๊ณต์ง€๋Šฅ์„ ์ถ”๊ฐ€ํ•˜๋ฉด ์žฌ๋ฏธ์žˆ์–ด์ง€๋Š”๋ฐ์š”
07:58
So, this character now has motor skills in the upper body --
150
478160
4000
์ด์ œ ๋ชธ ์œ—์ชฝ์— ์šด๋™๊ธฐ๋Šฅ์„ ๊ฐ–๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
08:02
nothing in the legs yet, in this particular one.
151
482160
2000
์•„์ง ๋‹ค๋ฆฌ์—๋Š” ์—†์ง€๋งŒ.
08:04
But what it will do -- I'm going to push it again.
152
484160
3000
์ œ๊ฐ€ ๋‹ค์‹œ ๋ฐ€๋ฉด
08:07
It will realize autonomously that it's being pushed.
153
487160
2000
๋ˆ„๊ตฐ๊ฐ€ ๋ฐ€๊ณ ์žˆ๋‹ค๋Š” ๊ฒƒ์„ ์ž๋™์ ์œผ๋กœ ๊ฐ์ง€ํ•˜๊ณ 
08:09
It's going to stick out its hands.
154
489160
2000
์†์„ ์•ž์œผ๋กœ ๋ป—์œผ๋ฉด์„œ
08:11
It's going to turn around into the fall, and try and catch the fall.
155
491160
3000
๋Œ์•„์„œ ๋„˜์–ด์ง€๋ ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
08:20
So that's what you see here.
156
500160
2000
์ด๋ ‡๊ฒŒ ๋ง์ด์ฃ 
08:22
Now, it gets really interesting
157
502160
2000
์ด์ œ ์žฌ๋ฏธ์žˆ์–ด ์ง€๋Š”๋ฐ์š”
08:24
if you then add the AI for the lower part of the body as well.
158
504160
4000
์ธ๊ณต์ง€๋Šฅ์„ ๋‹ค๋ฆฌ์—๋„ ์ถ”๊ฐ€ํ•˜๋ฉด
08:28
So here, we've got the same character.
159
508160
2000
์—ฌ๊ธฐ ๋™์ผํ•œ ์บ๋ฆญํ„ฐ์—๊ฐ€ ์žˆ๋Š”๋ฐ,
08:30
I'm going to push it a bit harder now,
160
510160
2000
์กฐ๊ธˆ ๋ฐ€์–ด ๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค.
08:32
harder than I just pushed Chris.
161
512160
2000
Chris ๋ฐ€์—ˆ๋˜ ๊ฒƒ ๋ณด๋‹ค ์ข€๋” ์„ธ๊ฒŒ
08:34
But what you'll see is -- it's going to receive a push now from the left.
162
514160
4000
๊ทธ๋Ÿฌ๋ฉด ์บ๋ฆญํ„ฐ๋Š” ์™ผ์ชฝ์—์„œ ๋ฏธ๋Š” ๊ฒƒ์„ ๋Š๋ผ๊ณ 
08:41
What you see is it takes steps backwards,
163
521160
2000
๋’ท๊ฑธ์Œ ์น˜๋ฉด์„œ
08:43
it tries to counter-balance,
164
523160
2000
๊ท ํ˜•์„ ์žก์œผ๋กœ๊ณ  ํ•˜๊ณ 
08:45
it tries to look at the place where it thinks it's going to land.
165
525160
4000
๋„˜์–ด์งˆ ์žฅ์†Œ๋ฅผ ๋ณด๊ฒŒ๋˜์ฃ .
08:49
I'll show you this again.
166
529160
2000
๋‹ค์‹œ ๋ณด์—ฌ๋“œ๋ฆฌ๊ฒ ์Šต๋‹ˆ๋‹ค.
08:51
And then, finally hits the floor.
167
531160
3000
๊ทธ๋ฆฌ๊ณ  ๊ฒฐ๊ตญ์€ ๋ฐ”๋‹ฅ์— ๋„˜์–ด์ง€์ฃ 
08:55
Now, this becomes really exciting
168
535160
3000
์ด์ œ๋ถ€ํ„ฐ ์ ์  ์žฌ๋ฏธ์žˆ์–ด ์ง€๋Š”๋ฐ์š”
08:58
when you push that character in different directions, again, just as I've done.
169
538160
5000
์ œ๊ฐ€ ํ–ˆ๋˜๊ฒƒ์ฒ˜๋Ÿผ ์ข€ ๋‹ค๋ฅธ ๋ฐฉํ–ฅ์—์„œ ๋ฐ€๊ฒŒ๋˜๋ฉด ์žฌ๋ฏธ์žˆ๋Š” ๊ฒƒ์„ ๋ณผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
09:03
That's something that you cannot do right now.
170
543160
4000
์ด๋Ÿฐ๊ฒƒ์€ ์—ฌ๋Ÿฌ๋ถ„์€ ์ง€๊ธˆ ํ•  ์ˆ˜ ์—†๋Š” ๊ฒƒ์ด์ฃ 
09:07
At the moment, you only have empty computer graphics in games.
171
547160
3000
์ง€๊ธˆ ๊ฒŒ์ž„์†์˜ ์บ๋ฆญํ„ฐ๋Š” ์†๋นˆ ๊ทธ๋ž˜ํ”ฝ์— ๋ถˆ๊ณผํ•˜๊ธฐ ๋•Œ๋ฌธ์ž…๋‹ˆ๋‹ค
09:10
What this is now is a real simulation. That's what I want to show you now.
172
550160
3000
์ด๊ฒŒ ๋ฐ”๋กœ ์ œ๊ฐ€ ๋ณด์—ฌ๋“œ๋ฆฌ๊ณ ์ž ํ•˜๋Š” ์ง„์ •ํ•œ ์‚ฌ๋žŒ์„ ํ‰๋‚ด๋‚ธ ์บ๋ฆญํ„ฐ ์ธ๋ฐ
09:13
So, here's the same character with the same behavior I've just shown you,
173
553160
3000
์•„๊นŒ๋ณด์‹  ๊ฒƒ๊ณผ ๊ฐ™์€ ํ–‰๋™์„ ํ•ฉ๋‹ˆ๋‹ค.
09:16
but now I'm just going to push it from different directions.
174
556160
2000
๊ทธ๋Ÿฐ๋ฐ ์ œ๊ฐ€ ๋‹ค๋ฅธ ๋ฐฉํ–ฅ์—์„œ ๋ฐ€์–ด๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค.
09:18
First, starting with a push from the right.
175
558160
2000
์šฐ์„  ์šฐ์ธก์—์„œ ๋ฐ€์–ด ๋ด…๋‹ˆ๋‹ค.
09:23
This is all slow motion, by the way, so we can see what's going on.
176
563160
3000
์ฐธ๊ณ ๋กœ ์ด๊ฑด ๋Š๋ฆฐ๋™์ž‘์ด๋ผ ์ž์„ธํžˆ ๋ณด์‹ค ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
09:26
Now, the angle will have changed a tiny bit,
177
566160
3000
๋ฐฉํ–ฅ์„ ์กฐ๊ธˆ ๋ฐ”๊พธ์–ด ๋ฐ€๋ฉด
09:29
so you can see that the reaction is different.
178
569160
4000
๋ฐ˜์‘์ด ๋‹ค๋ฅธ ๊ฒƒ์„ ๋ณด์‹ค ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
09:33
Again, a push, now this time from the front.
179
573160
3000
์ด๋ฒˆ์—๋Š” ์•ž์—์„œ ๋ฐ€๋ฉด
09:37
And you see it falls differently.
180
577160
2000
๋‹ค๋ฅด๊ฒŒ ๋„˜์–ด์ง€์ฃ ?
09:39
And now from the left --
181
579160
2000
์ขŒ์ธก์—์„œ ๋ฐ€๋ฉด
09:43
and it falls differently.
182
583160
2000
์—ญ์‹œ ๋‹ค๋ฅด๊ฒŒ ๋„˜์–ด์ง‘๋‹ˆ๋‹ค.
09:45
That was really exciting for us to see that.
183
585160
2000
์ด๊ฑธ ์ฒ˜์Œ ๋ณด์•˜์„๋•Œ
09:47
That was the first time we've seen that.
184
587160
2000
์•„์ฃผ ํฅ๋ถ„๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
09:49
This is the first time the public sees this as well,
185
589160
2000
์ด ์žฅ๋ฉด์€ ์—ฌํƒœ๊ฒƒ ์–ด๋””์—๋„ ๋ณด์—ฌ์ค€ ์ ์ด ์—†๋Š” ๊ฒƒ์œผ๋กœ ์—ฌ๋Ÿฌ๋ถ„์€
09:51
because we have been in stealth mode.
186
591160
2000
์„ธ์ƒ์—์„œ ์ฒ˜์Œ์œผ๋กœ
09:53
I haven't shown this to anybody yet.
187
593160
2000
์ง€๊ธˆ ์ด ์žฅ๋ฉด์„ ๋ณด๊ณ ๊ณ„์‹ญ๋‹ˆ๋‹ค.
09:55
Now, just a fun thing:
188
595160
2000
์ด์ œ ์žฌ๋ฏธ์žˆ๋Š” ์ผ์€
09:57
what happens if you put that character --
189
597160
2000
์ด ์บ๋ฆญํ„ฐ์—๋‹ค๊ฐ€
09:59
this is now a wooden version of it, but it's got the same AI in it --
190
599160
2000
ํ˜„๋Œ€์ ์ธ ์ธ๊ณต์ง€๋Šฅ์„ ์ถ”๊ฐ€ํ•˜๊ณ 
10:01
but if you put that character on a slippery surface, like ice.
191
601160
2000
์–ผ์Œ์ฒ˜๋Ÿผ ๋ฏธ๋„๋Ÿฌ์šด ๋ฐ”๋‹ฅ์— ์˜ฌ๋ ค ๋†“์œผ๋ฉด
10:03
We just did that for a laugh, just to see what happens.
192
603160
3000
์žฅ๋‚œ์‚ผ์•„ ํ•ด๋ณธ๊ฑด๋ฐ ์•„์ฃผ ์žฌ๋ฏธ์žˆ์Šต๋‹ˆ๋‹ค.
10:06
(Laughter)
193
606160
1000
(์›ƒ์Œ)
10:07
And this is what happens.
194
607160
2000
์žฌ๋ฏธ์žˆ์ฃ ?
10:09
(Laughter)
195
609160
3000
(์›ƒ์Œ)
10:12
(Applause)
196
612160
3000
(๋ฐ•์ˆ˜)
10:15
It's nothing we had to do about this.
197
615160
2000
์šฐ๋ฆฌ๊ฐ€ ์ด๋Ÿฐ ๋™์ž‘์„ ๋งŒ๋“ ๊ฒƒ์€ ์•„๋‹™๋‹ˆ๋‹ค.
10:17
We just took this character that I just talked about,
198
617160
2000
๋‹จ์ง€ ์„ค๋ช…๋“œ๋ฆฐ ์บ๋ฆญํ„ฐ๋ฅผ ๋งŒ๋“ค์–ด์„œ
10:19
put it on a slippery surface, and this is what you get out of it.
199
619160
3000
๋ฏธ๋„๋Ÿฌ์šด ๋ฐ”๋‹ฅ์— ์˜ฌ๋ ค๋†“์€ ๊ฒƒ ๋ถ„์ธ๋ฐ ์ด๋ ‡๊ฒŒ ์›€์ง์ž…๋‹ˆ๋‹ค.
10:22
And that's a really fascinating thing about this approach.
200
622160
3000
์•„์ฃผ ์‹ ๋‚˜๋Š” ์ผ์ด์ฃ 
10:26
Now, when we went to film studios and games developers
201
626160
3000
์šฐ๋ฆฌ๊ฐ€ ์ด ์บ๋ฆญํ„ฐ๋ฅผ ๊ฐ€์ง€๊ณ  ์˜ํ™” ์ŠคํŠœ๋””์˜ค์™€
10:29
and showed them that technology, we got a very good response.
202
629160
3000
๊ฒŒ์ž„๊ฐœ๋ฐœ์ž๋“ค์„ ์ฐพ์•„๊ฐ”์„๋•Œ ๋ฐ˜์‘์€ ๋งค์šฐ ๋œจ๊ฑฐ์› ์Šต๋‹ˆ๋‹ค.
10:32
And what they said was, the first thing they need immediately is virtual stuntmen.
203
632160
4000
๋‹น์žฅ ๊ฐ€์ƒ์˜ ์Šคํ„ดํŠธ๋งจ์ด ํ•„์š”ํ–ˆ๋‹ค๊ณ  ํ•˜๋”๊ตฐ์š”
10:36
Because stunts are obviously very dangerous, they're very expensive,
204
636160
4000
์œ„ํ—˜ํ•œ ์ผ์„ ํ•˜๋‹ค๋ณด๋‹ˆ ๋งค์šฐ ๋น„์‹ผ๋ฐ๋‹ค๊ฐ€
10:40
and there are a lot of stunt scenes that you cannot do, obviously,
205
640160
2000
์Šคํ„ดํŠธ๋งจ์ด ํ•  ์ˆ˜ ์—†๋Š” ๋งŽ์€ ์žฅ๋ฉด๋“ค๋„ ๋งŽ์ด ์žˆ์–ด
10:42
because you can't really allow the stuntman to be seriously hurt.
206
642160
3000
๋‹ค์น˜๊ฒŒ ํ• ์ˆ˜ ๋Š” ์—†์œผ๋‹ˆ ๊ณ ๋ฏผ์ด๋ผ๋”๊ตฐ์š”
10:45
So, they wanted to have a digital version of a stuntman
207
645160
3000
๊ทธ๋ž˜์„œ ๊ฐ€์ƒ์˜ ์Šคํ„ดํŠธ๋งจ์„ ์ฐพ๊ณ  ์žˆ์—ˆ๋Š”๋ฐ
10:48
and that's what we've been working on for the past few months.
208
648160
2000
๊ทธ๊ฑด ๋ฐ”๋กœ ์šฐ๋ฆฌ๊ฐ€ ๋ช‡๋‹ฌ๊ฐ„ ์ž‘์—…ํ•ด์˜จ ๊ฒƒ์ด๋ฉฐ
10:50
And that's our first product that we're going to release in a couple of weeks.
209
650160
5000
๋ช‡์ฃผ ํ›„์— ์‹œ์žฅ์— ์ถœ์‹œํ•  ์ฒซ์ œํ’ˆ์ด๊ธฐ๋„ ํ•ฉ๋‹ˆ๋‹ค.
10:55
So, here are just a few very simple scenes of the guy just being kicked.
210
655160
5000
์—ฌ๋Ÿฌ๋ถ„์€ ์ตœ๊ทผ ์™„์„ฑ๋œ ๋ช‡ ๊ฐœ์˜ ๊ฐ„๋‹จํ•œ ์žฅ๋ฉด๋“ค์„ ๋ณด๊ณ  ๊ณ„์‹ญ๋‹ˆ๋‹ค.
11:00
That's what people want. That's what we're giving them.
211
660160
2000
์ด๊ฒƒ์ด ๋ฐ”๋กœ ์„ธ์ƒ์ด ์›ํ•˜๋Š” ๊ฒƒ์ด๋ฉฐ ์šฐ๋ฆฌ๊ฐ€ ์ฃผ๋ ค๊ณ  ํ•˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
11:02
(Laughter)
212
662160
7000
(์›ƒ์Œ)
11:09
You can see, it's always reacting.
213
669160
2000
๋Š๋ผ์‹œ๊ฒ ์ง€๋งŒ ์ƒ๋‹นํžˆ ํ˜„์‹ค์ ์ด์ฃ 
11:11
This is not a dead body. This is a body who basically, in this particular case,
214
671160
4000
๋”ฑ๋”ฑํ•œ ์บ๋ฆญํ„ฐ๊ฐ€ ์•„๋‹ˆ๋ผ, ๋ณด์‹œ๋‹ค ์‹œํ”ผ
11:15
feels the force and tries to protect its head.
215
675160
2000
์™ธ๋ถ€์˜ ํž˜์„ ๋Š๋ผ๊ณ  ์ด์— ๋Œ€์‘ํ•ด์„œ ๋จธ๋ฆฌ๋ฅผ ๋ณดํ˜ธํ•˜๋ ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
11:17
Only, I think it's quite a big blow again.
216
677160
2000
๊ต‰์žฅํ•œ ํž˜์ด ๊ฐ€ํ•ด์ง‘๋‹ˆ๋‹ค.
11:19
You feel kind of sorry for that thing,
217
679160
2000
์—ฌ๋Ÿฌ๋ถ„์€ ๋ถˆ์Œํ•˜๊ฒŒ ๋Š๋ผ์‹œ๊ฒ ์ง€๋งŒ,
11:21
and we've seen it so many times now that
218
681160
2000
์šฐ๋ฆฌ๋Š” ๋„ˆ๋ฌด ๋งŽ์ด ๋ด์„œ
11:23
we don't really care any more.
219
683160
2000
๋ณ„๋กœ ๊ฑฑ์ •์€ ์•ˆ๋ฉ๋‹ˆ๋‹ค.
11:25
(Laughter)
220
685160
1000
(์›ƒ์Œ)
11:26
There are much worse videos than this, by the way, which I have taken out, but ...
221
686160
4000
์ด๊ฑฐ๋ณด๋‹ค ํ›จ์”ฌ ๋”ํ•œ ์žฅ๋ฉด๋„ ์žˆ์ง€๋งŒ ์—ฌ๊ธฐ์„œ๋Š” ์ œ์™ธํ•˜์˜€์Šต๋‹ˆ๋‹ค.
11:31
Now, here's another one.
222
691160
2000
ํ•œ๊ฐ€์ง€ ๋” ๋ณด์—ฌ๋“œ๋ฆด๊ฒŒ ์žˆ๋Š”๋ฐ
11:33
What people wanted as a behavior was to have an explosion,
223
693160
4000
์‚ฌ๋žŒ๋“ค์€ ๊ฐ•๋ ฅํ•œ ํž˜์ด ๊ฐ€ํ•ด์ง€๋Š” ํญ๋ฐœ์ด ์žˆ์„๋•Œ
11:37
a strong force applied to the character,
224
697160
2000
์บ๋ฆญํ„ฐ๊ฐ€ ํž˜์— ๋Œ€ํ•ด
11:39
and have the character react to it in midair.
225
699160
2000
์–ด๋–ป๊ฒŒ ๋ฐ˜์‘ํ•˜๋Š”์ง€๋ฅผ ๋ณด๊ณ ์‹ถ์–ด ํ–ˆ์Šต๋‹ˆ๋‹ค.
11:41
So that you don't have a character that looks limp,
226
701160
2000
๊ทธ๋ž˜์„œ ํญ๋ฐœ์ด ์žˆ์œผ๋ฉด ์ถ•์ฒ˜์ง€๋Š” ๊ฒƒ์ด ์•„๋‹Œ
11:43
but actually a character that you can use in an action film straight away,
227
703160
3000
์‹ค์ œ ๊ณต์ค‘์—์„œ ์‚ด์•„์žˆ๋Š” ๋“ฏํ•œ ์บ๋ฆญํ„ฐ๋ฅผ ํ™œ์šฉํ•ด์„œ
11:46
that looks kind of alive in midair as well.
228
706160
2000
๋ฐ”๋กœ ์•ก์…˜์˜ํ™”์— ํ™œ์šฉํ• ์ˆ˜ ์žˆ๊ธฐ๋ฅผ ์›ํ–ˆ์Šต๋‹ˆ๋‹ค.
11:48
So this character is going to be hit by a force,
229
708160
2000
์—ฌ๊ธฐ ์ด ์บ๋ฆญํ„ฐ๋Š” ์–ด๋–ค ํž˜์— ์˜ํ•ด ๋‚ ์•„๊ฐ‘๋‹ˆ๋‹ค.
11:50
it's going to realize it's in the air,
230
710160
2000
๋ชธ์ด ๊ณต์ค‘์— ๋– ์žˆ๋Š” ๊ฒƒ์„ ๋Š๋ผ๊ฒŒ ๋˜๊ณ 
11:52
and it's going to try and, well,
231
712160
3000
๋–จ์–ด์ง€๋Š” ๋ฐฉํ–ฅ์œผ๋กœ
11:55
stick out its arm in the direction where it's landing.
232
715160
2000
ํŒ”์„ ๋ป—๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
11:59
That's one angle; here's another angle.
233
719160
3000
์ด๋ฒˆ์—๋Š” ๋‹ค๋ฅธ ๊ฐ๋„์ž…๋‹ˆ๋‹ค.
12:02
We now think that the realism we're achieving with this
234
722160
2000
์ด ์บ๋ฆญํ„ฐ๋ฅผ ํ†ตํ•ด ์‹คํ˜„๋˜๋Š” ํ˜„์‹ค๊ฐ์€
12:04
is good enough to be used in films.
235
724160
2000
์ถฉ๋ถ„ํžˆ ์˜ํ™”์— ์ ์šฉ๋  ์ˆ˜ ์žˆ์„ ๋“ฏ ํ•ฉ๋‹ˆ๋‹ค.
12:06
And let's just have a look at a slightly different visualization.
236
726160
3000
์ข€ ๋‹ค๋ฅธ ์˜์ƒ์„ ํ•œ๋ฒˆ ๋ณด์‹œ์ฃ 
12:09
This is something I just got last night
237
729160
2000
์ด๊ฑด ์ œ๊ฐ€ ๋ฐ”๋กœ ์–ด์ œ
12:11
from an animation studio in London, who are using our software
238
731160
3000
๋Ÿฐ๋˜์— ์žˆ๋Š” ์ŠคํŠœ๋””์˜ค์—์„œ ์šฐ๋ฆฌ๊ฐ€ ๋งŒ๋“  ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ
12:14
and experimenting with it right now.
239
734160
2000
ํ™œ์šฉํ•ด์„œ ๋งŒ๋“  ๊ฒ๋‹ˆ๋‹ค.
12:16
So this is exactly the same behavior that you saw,
240
736160
3000
๋ฐฉ๊ธˆ ๋ณด์‹ ๊ฒƒ๊ณผ ๋™์ผํ•œ ๊ฒƒ์ด์ง€๋งŒ
12:19
but in a slightly better rendered version.
241
739160
4000
์ข€๋” ํ–ฅ์ƒ๋œ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
12:23
So if you look at the character carefully,
242
743160
3000
์บ๋ฆญํ„ฐ๋ฅผ ์ž์„ธํžˆ ๋ณด์‹œ๋ฉด
12:26
you see there are lots of body movements going on,
243
746160
2000
๋ฌด์ˆ˜ํžˆ ๋งŽ์€ ๋™์ž‘๋“ค์ด ํฌํ•จ๋œ ๊ฒƒ์„ ํ™•์ธํ•˜์‹ค ์ˆ˜ ์žˆ๋Š”๋ฐ
12:28
none of which you have to animate like in the old days.
244
748160
2000
์˜ˆ์ „์—๋Š” ์ด๋Ÿฐ๋™์ž‘์ด ๋ถˆ๊ฐ€๋Šฅํ–ˆ์Šต๋‹ˆ๋‹ค.
12:30
Animators had to actually animate them.
245
750160
2000
์ œ์ž‘์ž๋Š” ์‹ค์ œ๋กœ ์‚ฌ๋žŒ์„ ํ‰๋‚ด๋‚ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
12:32
This is all happening automatically in the simulation.
246
752160
2000
์‹œ๋ฎฌ๋ ˆ์ด์…˜์—์„œ ์ž๋™์ ์œผ๋กœ ์ผ์–ด๋‚˜๋Š” ๊ฒƒ๋“ค์ด์ฃ .
12:34
This is a slightly different angle,
247
754160
2000
์ด๋ฒˆ์—๋Š” ์ข€ ๋‹ค๋ฅธ ๊ฐ๋„์—์„œ์˜ ๋ชจ์Šต์ธ๋ฐ
12:39
and again a slow motion version of this.
248
759160
2000
๋Š๋ฆฐ๋™์ž‘์œผ๋กœ ๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค.
12:41
This is incredibly quick. This is happening in real time.
249
761160
4000
์ •๋ง ๋น ๋ฅด์ฃ ? ์‹ค์‹œ๊ฐ„์œผ๋กœ ์ผ์–ด๋‚˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค
12:45
You can run this simulation in real time, in front of your eyes,
250
765160
2000
๋ˆˆ์•ž์—์„œ ์‹ค์ œ์™€ ๊ฐ™์€ ์žฅ๋ฉด์„ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ํ•˜๊ณ ,
12:47
change it, if you want to, and you get the animation straight out of it.
251
767160
3000
๋ฐ”๊พธ๊ณ  ์‹ถ์€๊ฒŒ ์žˆ๋‹ค๋ฉด ๋ฐ”๋กœ ์›ํ•˜๋Š” ์˜์ƒ์„ ์–ป์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
12:50
At the moment, doing something like this by hand
252
770160
2000
ํ˜„์žฌ ์ด๊ฑธ ์ˆ˜์ž‘์—…์œผ๋กœ ํ•œ๋‹ค๋ฉด
12:52
would take you probably a couple of days.
253
772160
2000
์•„๋งˆ ๋ฉฐ์น ์€ ์กฑํžˆ ๊ฑธ๋ฆด ๊ฒ๋‹ˆ๋‹ค.
12:55
This is another behavior they requested.
254
775160
3000
์ด ํ–‰๋™๋„ ๊ตฌํ˜„ํ•˜๊ณ  ์‹ถ๋‹ค๊ณ  ํ•˜๋”๊ตฐ์š”
12:58
I'm not quite sure why, but we've done it anyway.
255
778160
2000
์™œ ํ•„์š”ํ•œ์ง€๋Š” ๋ชจ๋ฅด์ง€๋งŒ ํ•œ๋ฒˆ ํ•ด๋ณด์•˜์Šต๋‹ˆ๋‹ค.
13:00
It's a very simple behavior that shows you the power of this approach.
256
780160
2000
์ด ์ ‘๊ทผ๋ฐฉ๋ฒ•์ด ์–ผ๋งˆ๋‚˜ ๊ฐ•๋ ฅํ•œ์ง€ ์•Œ์ˆ˜ ์žˆ๋Š” ๊ฐ„๋‹จํ•œ ํ–‰๋™์„ ํ•ฉ๋‹ˆ๋‹ค.
13:02
In this case, the character's hands
257
782160
2000
์—ฌ๊ธฐ์„œ ์บ๋ฆญํ„ฐ์˜ ์†์€
13:04
are fixed to a particular point in space,
258
784160
2000
๊ณต์ค‘ ์–ด๋”˜๊ฐ€์— ๊ณ ์ •๋˜์–ด ์žˆ๊ณ 
13:06
and all we've told the character to do is to struggle.
259
786160
3000
๋ฐœ๋ฒ„๋‘ฅ ์ฒ˜๋ณด๋ผ๊ณ  ๋ช…๋ นํ–ˆ๋”๋‹ˆ
13:09
And it looks organic. It looks realistic.
260
789160
3000
๋งˆ์น˜ ์‚ด์•„์žˆ๋Š” ์‚ฌ๋žŒ๊ฐ™์Šต๋‹ˆ๋‹ค.
13:12
You feel kind of sorry for the guy.
261
792160
2000
๋ถˆ์Œํ•ด ๋ณด์ด์‹œ๋‚˜์š”?
13:14
It's even worse -- and that is another video I just got last night --
262
794160
3000
์ด๊ฑด ์ œ๊ฐ€ ์–ด์ œ ์ž‘์—…ํ•œ ๋˜ ๋‹ค๋ฅธ ์˜์ƒ์ธ๋ฐ ์ข€ ๋” ํ˜„์‹ค์ ์œผ๋กœ
13:17
if you render that a bit more realistically.
263
797160
2000
๋งŒ๋“ค๋ฉด ๋” ๋ถˆ์Œํ•ด ๋ณด์ž…๋‹ˆ๋‹ค.
13:23
Now, I'm showing this to you just to show you
264
803160
2000
์ œ๊ฐ€ ์ด๊ฑธ ๋ณด์—ฌ๋“œ๋ฆฌ๋Š” ์ด์œ ๋Š”
13:25
how organic it actually can feel, how realistic it can look.
265
805160
2000
์–ผ๋งˆ๋‚˜ ์‚ด์•„์žˆ๊ณ  ํ˜„์‹ค๊ฐ ์žˆ์–ด ๋ณด์ด๋Š”์ง€ ๋ณด์—ฌ๋“œ๋ฆฌ๊ธฐ ์œ„ํ•ด์„œ์ž…๋‹ˆ๋‹ค.
13:27
And this is all a physical simulation of the body,
266
807160
3000
์ด๊ฑด ์‚ฌ๋žŒ์˜ ๋ชธ์„ ์‹ค์ œ์™€ ๊ฐ™์ด ํ‰๋‚ด๋‚ด๊ธฐ ์ž…๋‹ˆ๋‹ค.
13:30
using AI to drive virtual muscles in that body.
267
810160
3000
์ธ๊ณต์ง€๋Šฅ์ด ๊ฐ€์ƒ์˜ ๊ทผ์œก์„ ์›€์ง์ด๊ฒŒ ๋ผ์ฃ .
13:35
Now, one thing which we did for a laugh was
268
815160
3000
์žฅ๋‚œ์‚ผ์•„ ๋งŒ๋“  ์ข€๋” ๋ณต์žกํ•œ
13:38
to create a slightly more complex stunt scene,
269
818160
2000
์Šคํ„ดํŠธ ์žฅ๋ฉด์„ ๋ณด์—ฌ๋“œ๋ฆฌ๊ฒ ์Šต๋‹ˆ๋‹ค.
13:40
and one of the most famous stunts is the one where James Bond
270
820160
3000
์œ ๋ช…ํ•œ ์Šคํ„ดํŠธ ์žฅ๋ฉด์ธ๋ฐ ์ œ์ž„์Šค ๋ณธ๋“œ๊ฐ€ ์Šค์œ„์Šค์— ์žˆ๋Š”
13:43
jumps off a dam in Switzerland and then is caught by a bungee.
271
823160
4000
๋Œ์—์„œ ๋›ฐ์–ด๋‚ด๋ฆฌ๊ณ  ๋งˆ์ง€๋ง‰์— ๋ฒˆ์ง€์ค„์— ๊ฑธ๋ฆฝ๋‹ˆ๋‹ค.
13:48
Got a very short clip here.
272
828160
2000
์—ฌ๊ธฐ์žˆ๊ตฐ์š”
13:54
Yes, you can just about see it here.
273
834160
2000
ํ•œ๋ฒˆ ๊ฐ์ƒํ•ด ๋ณด์‹œ์ฃ .
13:56
In this case, they were using a real stunt man. It was a very dangerous stunt.
274
836160
3000
์—ฌ๊ธฐ์„œ๋Š” ์‹ค์ œ ์Šคํ„ดํŠธ๋งจ์ด ์ ํ”„ํ–ˆ๋Š”๋ฐ ์ •๋ง ์œ„ํ—˜ํ•œ ์ผ์ด์ฃ 
13:59
It was just voted, I think in the Sunday Times, as one of the most impressive stunts.
275
839160
3000
์•„๋งˆ Sunday Times ์˜€๋˜๊ฑฐ ๊ฐ™์€๋ฐ์š” ๊ฐ€์žฅ ๊ธฐ์–ต์— ๋‚จ๋Š” ์žฅ๋ฉด์œผ๋กœ ๋ฝ‘ํ˜”๋‹ค๊ณ  ํ•˜๋„ค์š”
14:02
Now, we've just tried and -- looked at our character and asked ourselves,
276
842160
3000
์ด์žฅ๋ฉด์„ ์šฐ๋ฆฌ๊ฐ€ ๋งŒ๋“  ์บ๋ฆญํ„ฐ๋ฅผ ํ†ตํ•ด ๋งŒ๋“ค์–ด ๋ณด๊ธฐ๋กœ ํ–ˆ์Šต๋‹ˆ๋‹ค.
14:05
"Can we do that ourselves as well?"
277
845160
2000
๊ณผ์—ฐ ํ•  ์ˆ˜ ์žˆ์„๊นŒ ๊ถ๊ธˆํ–ˆ์Šต๋‹ˆ๋‹ค.
14:07
Can we use the physical simulation of the character,
278
847160
2000
์บ๋ฆญํ„ฐ ์‹ ์ฒด ์‹œ๋ฎฌ๋ ˆ์ด์…˜์„ ์œ„ํ•ด
14:09
use artificial intelligence,
279
849160
2000
์ธ๊ณต์ง€๋Šฅ์„
14:11
put that artificial intelligence into the character,
280
851160
2000
์บ๋ฆญํ„ฐ์— ์ถ”๊ฐ€ํ•˜๊ณ 
14:13
drive virtual muscles, simulate the way he jumps off the dam,
281
853160
4000
๊ฐ€์ƒ์˜ ๊ทผ์œก๋„ ์ถ”๊ฐ€ํ•ด์„œ ๋ณธ๋“œ๊ฐ€ ๋Œ์—์„œ ์ ํ”„ํ•˜๊ณ 
14:17
and then skydive afterwards,
282
857160
2000
๊ณต์ค‘์—์„œ ๋‹ค์ด๋น™ํ•ด์„œ ๋‚ด๋ ค์˜ค๋‹ค๊ฐ€
14:19
and have him caught by a bungee afterwards?
283
859160
2000
๋ฒˆ์ง€์ค„์— ๊ฑธ๋ฆฌ๋Š” ์žฅ๋ฉด์„ ํ‰๋‚ด๋‚ผ ์ˆ˜ ์žˆ์„๊นŒ?
14:21
We did that. It took about altogether just two hours,
284
861160
3000
2์‹œ๊ฐ„ ์ •๋„ ๊ฑธ๋ ค์„œ
14:24
pretty much, to create the simulation.
285
864160
2000
๊ทธ ์žฅ๋ฉด์„ ํ‰๋‚ด๋‚ด์–ด ๋ณด์•˜์Šต๋‹ˆ๋‹ค.
14:26
And that's what it looks like, here.
286
866160
2000
๊ทธ๋žฌ๋”๋‹ˆ ์ด๋Ÿฐ ๊ฒฐ๊ณผ๊ฐ€ ๋‚˜์™”์Šต๋‹ˆ๋‹ค.
14:37
Now, this could do with a bit more work. It's still very early stages,
287
877160
3000
์ข€๋” ์ž‘์—…์„ ํ•  ์ˆ˜ ์žˆ์ง€๋งŒ, ์ด๊ฑด ์•„์ฃผ ์ดˆ๊ธฐ๋‹จ๊ณ„๋ผ์„œ
14:40
and we pretty much just did this for a laugh,
288
880160
2000
์žฅ๋‚œ์‚ผ์•„์„œ ์šฐ๋ฆฌ๊ฐ€ ๋ฌด์—‡์„ ์–ป์–ด๋‚ผ์ˆ˜ ์žˆ๋Š”์ง€
14:42
just to see what we'd get out of it.
289
882160
2000
๋งŒ๋“ค์–ด ๋ณธ ๊ฒƒ์ž…๋‹ˆ๋‹ค.
14:44
But what we found over the past few months
290
884160
2000
๊ทธ๋Ÿฐ๋ฐ ์šฐ๋ฆฌ๊ฐ€ ์ง€๋‚œ ๋ช‡๋‹ฌ๊ฐ„ ๋ฐœ๊ฒฌํ•œ ๊ฒƒ์€
14:46
is that this approach -- that we're pretty much standard upon --
291
886160
3000
์ƒ๋‹นํžˆ ํ‘œ์ค€ํ™”๋œ ์ด๋ฐฉ๋ฒ•์ด ๋†€๋ผ์šธ ์ •๋„๋กœ
14:49
is incredibly powerful.
292
889160
2000
๊ฐ•๋ ฅํ•˜๋‹ค๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
14:51
We are ourselves surprised what you actually get out of the simulations.
293
891160
4000
์šฐ๋ฆฌ๋“ค ์ž์‹ ๋„ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ๊ฒฐ๊ณผ๋ฅผ ๋ณด๊ณ  ๋†€๋ž์Šต๋‹ˆ๋‹ค.
14:55
There's very often very surprising behavior that you didn't predict before.
294
895160
4000
์˜ˆ์ „์—๋Š” ์ƒ์ƒํ•  ์ˆ˜ ์—†์—ˆ๋˜ ๋†€๋ผ์šด ํ–‰๋™๋“ค์ด ๊ฐ€๋Šฅํ•ฉ๋‹ˆ๋‹ค.
14:59
There's so many things we can do with this right now.
295
899160
2000
์ด๊ฑธ ๊ฐ€์ง€๊ณ  ์•„์ฃผ ๋งŽ์€ ์ผ๋“ค์„ ๋‹น์žฅ ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
15:01
The first thing, as I said, is going to be virtual stuntmen.
296
901160
3000
์šฐ์„ ์€ ๊ฐ€์ƒ์˜ ์Šคํ„ดํŠธ๋งจ์„ ๋งŒ๋“œ๋Š” ๊ฒ๋‹ˆ๋‹ค.
15:04
Several studios are using this software now to produce virtual stuntmen,
297
904160
4000
์—ฌ๋Ÿฌ ์ŠคํŠœ๋””์˜ค์—์„œ ๊ฐ€์ƒ์˜ ์Šคํ„ดํŠธ๋งจ์„ ๋งŒ๋“ค๊ธฐ ์œ„ํ•ด ์šฐ๋ฆฌ์˜ ์†Œํ”„ํŠธ์›จ์–ด๋ฅผ ์‚ฌ์šฉํ•˜๊ณ  ์žˆ์œผ๋ฉฐ
15:08
and they're going to hit the screen quite soon, actually,
298
908160
2000
์กฐ๋งŒ๊ฐ„ ๋ช‡๋ช‡ ๋Œ€ํ˜• ์ œ์ž‘์ž๋“ค์ด ์ด๋ฅผ ์‹ค์ œ ์˜ํ™”์— ํ™œ์šฉํ•˜์—ฌ
15:10
for some major productions.
299
910160
2000
๋ฐฉ์˜ํ•  ์˜ˆ์ •์ž…๋‹ˆ๋‹ค.
15:12
The second thing is video games.
300
912160
3000
๋‘๋ฒˆ์งธ๋กœ ํ™œ์šฉ๊ฐ€๋Šฅํ•œ ๋ถ„์•ผ๋Š” ๋น„๋””์˜ค ๊ฒŒ์ž„์ž…๋‹ˆ๋‹ค.
15:15
With this technology, video games will look different and they will feel very different.
301
915160
4000
์ด ๊ธฐ์ˆ ์€ ๋น„๋””์˜ค๊ฒŒ์ž„์„ ๋‹ค๋ฅด๊ฒŒ ๋ณด์ด๊ณ  ๋Š๊ปด์ง€๋„๋ก ๋งŒ๋“ญ๋‹ˆ๋‹ค.
15:19
For the first time, you'll have actors that really feel very interactive,
302
919160
3000
์—ฌ๋Ÿฌ๋ถ„์€ ์ด์ œ ์ฒ˜์Œ์œผ๋กœ ์‹ค์ œ์™€ ๊ฐ™์ด ๋ฐ˜์‘ํ•˜๊ณ 
15:22
that have real bodies that really react.
303
922160
2000
์‹ค์ œ ์œก์ฒด๋ฅผ ๊ฐ€์ง„ ์บ๋ฆญํ„ฐ๋ฅผ ๊ฐ€์ง€๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
15:24
I think that's going to be incredibly exciting.
304
924160
3000
์ •๋ง ๋Œ€๋‹จํ•œ ์ผ์ด์ฃ ?
15:27
Probably starting with sports games,
305
927160
2000
์•„๋งˆ ์Šคํฌ์ธ  ๊ฒŒ์ž„์ฒ˜๋Ÿผ ์ƒํ˜ธ์ž‘์šฉ์ด ๋งŽ์€
15:29
which are going to become much more interactive.
306
929160
2000
๋ถ€๋ถ„๋ถ€ํ„ฐ ์‹œ์ž‘ํ•  ๊ฒƒ์ž…๋‹ˆ๋‹ค.
15:31
But I particularly am really excited
307
931160
1000
์ œ๊ฐ€ ํŠนํžˆ ์˜จ๋ผ์ธ์—์„œ ์ด ๊ธฐ์ˆ ์„ ์‚ฌ์šฉํ•  ๊ฒฝ์šฐ
15:32
about using this technology in online worlds,
308
932160
3000
ํฅ๋ถ„๋˜๋Š” ๊ฒƒ์€, ์˜ˆ๋ฅผ ๋“ค์–ด,
15:35
like there, for example, that Tom Melcher has shown us.
309
935160
3000
Tom Melcher๊ฐ€ ์šฐ๋ฆฌ์—๊ฒŒ ๋ณด์—ฌ์ค€ ๊ฒƒ ์ž…๋‹ˆ๋‹ค.
15:38
The degree of interactivity you're going to get
310
938160
2000
์ƒํ˜ธ์ž‘์šฉ์˜ ์ •๋„๋Š” ์—ฌ์ง€๊นŒ์ง€์˜ ๊ฒƒ๊ณผ๋Š”
15:40
is totally different, I think, from what you're getting right now.
311
940160
3000
์™„์ „ํžˆ ๋‹ค๋ฅธ ์ˆ˜์ค€์ž…๋‹ˆ๋‹ค.
15:44
A third thing we are looking at and very interested in is simulation.
312
944160
4000
์„ธ๋ฒˆ์งธ๋กœ ์šฐ๋ฆฌ๊ฐ€ ๋ณด๊ณ ์žˆ๋Š” ๊ฒƒ์€ ์•„์ฃผํฅ๋ฏธ๋กญ์Šต๋‹ˆ๋‹ค.
15:49
We've been approached by several simulation companies,
313
949160
2000
์—ฌ๋Ÿฌ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ํšŒ์‚ฌ๋“ค์ด ์ œ์•ˆ์„ ํ•ด์™”์ง€๋งŒ
15:51
but one project we're particularly excited about, which we're starting next month,
314
951160
3000
ํŠนํžˆ ๊ด€์‹ฌ์„ ๋„๋Š” ํ•œ๊ฐ€์ง€๊ฐ€ ์žˆ๋Š”๋ฐ, ๋ฐ”๋กœ ๋‹ค์Œ๋‹ฌ ๋ถ€ํ„ฐ ์‹œ์ž‘ํ•ฉ๋‹ˆ๋‹ค.
15:54
is to use our technology -- and in particular, the walking technology --
315
954160
4000
๊ทธ๊ฑด ์šฐ๋ฆฌ์˜ ๊ธฐ์ˆ  ๊ทธ ์ค‘์—์„œ๋„ ํŠนํžˆ ๊ฑท๋Š” ๊ธฐ์ˆ ์„
15:58
to help aid surgeons who work on children with cerebral palsy,
316
958160
4000
์™ธ๊ณผ์˜์‚ฌ๋“ค์ด ๋‡Œ์„ฑ๋งˆ๋น„๊ฐ€ ์žˆ๋Š” ์•„์ด๋“ค์˜ ์ˆ˜์ˆ  ๊ฒฐ๊ณผ๊ฐ€
16:02
to predict the outcome of operations on these children.
317
962160
3000
์–ด๋–จ์ง€ ์˜ˆ์ธกํ•  ์ˆ˜ ์žˆ๋„๋ก ๋„์™€์ฃผ๋ฐ ์‚ฌ์šฉํ•˜๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
16:05
As you probably know,
318
965160
2000
์•„๋งˆ ์•„์‹œ๊ฒ ์ง€๋งŒ
16:07
it's very difficult to predict what the outcome of an operation is
319
967160
3000
๊ฑธ์Œ๊ฑธ์ด๊ฐ€ ์–ด๋–ป๊ฒŒ ๋ฐ”๋€”์ง€ ์ˆ˜์ˆ  ๊ฒฐ๊ณผ๋ฅผ ์˜ˆ์ธกํ•œ๋‹ค๋Š” ๊ฒƒ์€ ์ •๋ง์ด์ง€
16:10
if you try and correct the gait.
320
970160
2000
์–ด๋ ค์šด ์ผ์ž…๋‹ˆ๋‹ค.
16:12
The classic quote is, I think, it's unpredictable at best,
321
972160
3000
์—ด๊ธธ ๋ฌผ์†์€ ์•Œ์•„๋„ ํ•œ๊ธธ ์‚ฌ๋žŒ์†์€ ๋ชจ๋ฅธ๋‹ค๋Š” ์†๋‹ด์ด ์žˆ์ง€๋งŒ, ์ด ๊ฒฐ๊ณผ ์—ญ์‹œ
16:15
is what people think right now, is the outcome.
322
975160
3000
์•„๋ฌด๋„ ์•Œ ์ˆ˜ ์—†์Šต๋‹ˆ๋‹ค.
16:18
Now, what we want to do with our software is allow our surgeons to have a tool.
323
978160
4000
์šฐ๋ฆฌ์˜ ์†Œํ”„ํŠธ์›จ์–ด๊ฐ€ ์™ธ๊ณผ์˜์‚ฌ๋“ค์—๊ฒŒ ์ข‹์€ ์ˆ˜๋‹จ์ด ๋˜์—ˆ์œผ๋ฉด ํ•ฉ๋‹ˆ๋‹ค.
16:22
We're going to simulate the gait of a particular child
324
982160
3000
์šฐ๋ฆฌ๋Š” ํ•œ ์•„์ด์˜ ๊ฑธ์Œ๊ฑธ์ด๋ฅผ ํ‰๋‚ด๋‚ด๊ณ ,
16:25
and the surgeon can then work on that simulation
325
985160
3000
์˜์‚ฌ์„ ์ƒ๋‹˜๋„ ์ฐธ์—ฌํ•ด์„œ
16:28
and try out different ways to improve that gait,
326
988160
2000
๊ฑธ์Œ๊ฑธ์ด๋ฅผ ๊ต์ •ํ•˜๊ธฐ ์œ„ํ•ด ์—ฌ๋Ÿฌ ๋ฐฉ๋ฒ•๋“ค์„ ์‹œ๋„ํ•œ ํ›„์—
16:30
before he actually commits to an actual surgery.
327
990160
3000
์˜์‚ฌ์„ ์ƒ๋‹˜์ด ์‹ค์ œ ์ˆ˜์ˆ ์— ๋“ค์–ด๊ฐˆ๊ฒ๋‹ˆ๋‹ค.
16:33
That's one project we're particularly excited about,
328
993160
2000
์ด๊ฑด ์ •๋ง ํฅ๋ถ„๋˜๋Š” ์ผ์ธ๋ฐ์š”
16:35
and that's going to start next month.
329
995160
2000
๋ฐ”๋กœ ๋‹ค์Œ๋‹ฌ์— ์‹œ์ž‘ํ•˜๋ คํ•ฉ๋‹ˆ๋‹ค.
16:39
Just finally, this is only just the beginning.
330
999160
3000
๋งˆ์ง€๋ง‰์œผ๋กœ, ์—ฌ๊ธฐ๊นŒ์ง€๋Š” ์‹œ์ž‘์— ๋ถˆ๊ณผํ•ฉ๋‹ˆ๋‹ค.
16:42
We can only do several behaviors right now.
331
1002160
2000
์—ฌ๋Ÿฌ๊ฐ€์ง€ ๊ฒƒ๋“ค์„ ๋‹น์žฅ์ด๋ผ๋„ ํ•  ์ˆ˜ ์žˆ์ฃ .
16:44
The AI isn't good enough to simulate a full human body.
332
1004160
3000
์ธ๊ณต์ง€๋Šฅ๋งŒ์œผ๋กœ ์‚ฌ๋žŒ์„ ํ‰๋‚ด๋‚ด๋Š” ๊ฒƒ์€ ์–ด๋ ต์Šต๋‹ˆ๋‹ค.
16:47
The body yes, but not all the motor skills that we have.
333
1007160
3000
ํŠนํžˆ ๋ชจ๋“  ์šด๋™๊ธฐ๋Šฅ์„ ๊ตฌํ˜„ํ•˜๊ธฐ๋Š” ๋”์šฑ ์–ด๋ ต์ฃ .
16:50
And, I think, we're only there if we can have something like ballet dancing.
334
1010160
3000
๋ฐœ๋ ˆ์ถค์ถ”๋Š” ๊ฒƒ์„ ํ‰๋‚ด๋‚ด๋ ค๋ฉด ๋ชจ๋“  ์šด๋™๊ธฐ๋Šฅ์„ ๊ตฌํ˜„ํ•ด์•ผํ•˜๋Š”๋ฐ
16:53
Right now, we don't have that
335
1013160
2000
์•„์ง ๊ฑฐ๊ธฐ๊นŒ์ง€๋Š” ์•„๋‹™๋‹ˆ๋‹ค.
16:55
but I'm very sure that we will be able to do that at some stage.
336
1015160
2000
ํ•˜์ง€๋งŒ ์–ธ์  ๊ฐ€๋Š” ๊ฐ€๋Šฅํ•˜๋ฆฌ๋ผ ๋ด…๋‹ˆ๋‹ค.
16:57
We do have one unintentional dancer actually,
337
1017160
3000
๋งˆ์ง€๋ง‰์œผ๋กœ ๋ณด์—ฌ๋“œ๋ฆด ๊ฒƒ์€ ์ถค์ถ”๋Š” ์‚ฌ๋žŒ์ธ๋ฐ์š”
17:00
the last thing I'm going to show you.
338
1020160
2000
์˜๋„ํ•œ๊ฑด ์•„๋‹™๋‹ˆ๋‹ค.
17:02
This was an AI contour that was produced and evolved --
339
1022160
3000
์ด๊ฑด ์ธ๊ณต์ง€๋Šฅ์„ ๊ฐ–์ถ˜ ๋ชจ์Šต์ธ๋ฐ
17:05
half-evolved, I should say -- to produce balance, basically.
340
1025160
3000
๊ท ํ˜•๊ฐ์„ ๊ฐ–์ถ”๊ธฐ ์œ„ํ•ด ๋ฐ˜์ฏค ์ง„ํ™”๋œ ์ˆ˜์ค€์ž…๋‹ˆ๋‹ค.
17:08
So, you kick the guy and the guy's supposed to counter-balance.
341
1028160
3000
์ด์‚ฌ๋žŒ์„ ๊ฑท์–ด์ฐจ๋ฉด ๊ท ํ˜•์„ ์žก๋„๋ก ๋˜์–ด ์žˆ์ฃ .
17:11
That's what we thought was going to come out of this.
342
1031160
3000
๊ทธ๊ฒŒ ์šฐ๋ฆฌ๊ฐ€ ์›ํ•˜๋˜ ๋ชจ์Šต์ธ๋ฐ
17:14
But this is what emerged out of it, in the end.
343
1034160
2000
์‹ค์ œ๋Š” ์ด๋ ‡์Šต๋‹ˆ๋‹ค.
17:17
(Music)
344
1037160
10000
(์Œ์•…)
17:27
Bizarrely, this thing doesn't have a head. I'm not quite sure why.
345
1047160
3000
์ด์ƒํ•˜๊ฒŒ๋„ ๋จธ๋ฆฌ๊ฐ€ ์—†๋Š”๋ฐ ์™œ๊ทธ๋Ÿฐ์ง€๋Š” ์ €๋„ ์ž˜ ๋ชจ๋ฆ…๋‹ˆ๋‹ค.
17:31
So, this was not something we actually put in there.
346
1051160
2000
์šฐ๋ฆฌ๊ฐ€ ๊ทธ๋ ‡๊ฒŒ ๋งŒ๋“ ๊ฑด ์•„๋‹ˆ๋ž€ ์–˜๊ธฐ์ฃ .
17:33
He just started to create that dance himself.
347
1053160
4000
์Šค์Šค๋กœ ์ถค์„ ์ถ”๋Š”๋ฐ
17:37
He's actually a better dancer than I am, I have to say.
348
1057160
3000
์‚ฌ์‹ค ์ €๋ณด๋‹ค ์ž˜์ถ”๋Š” ๊ตฐ์š”.
17:41
And what you see after a while --
349
1061160
2000
์ž ์‹œ๋’ค์—๋Š”
17:43
I think he even goes into a climax right at the end.
350
1063160
2000
์ ˆ์ •์˜ ๋งˆ์ง€๋ง‰์„ ์žฅ์‹ํ•˜๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
17:49
And I think -- there you go.
351
1069160
3000
๋ณด์ด์‹œ์ฃ ?
17:52
(Laughter)
352
1072160
2000
(์›ƒ์Œ)
17:54
So, that all happened automatically. We didn't put that in there.
353
1074160
2000
๋ชจ๋“ ๊ฒŒ ์ž๋™์œผ๋กœ ๋‚˜์˜ค๋Š” ์˜์ƒ์œผ๋กœ ์šฐ๋ฆฌ๊ฐ€ ํ•œ๊ฑด ์—†์Šต๋‹ˆ๋‹ค.
17:56
That's just the simulation creating this itself, basically.
354
1076160
3000
๊ธฐ๋ณธ์ ์œผ๋กœ ์Šค์Šค๋กœ๋ฅผ ์ฐฝ์กฐํ•˜๋Š” ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ์ด์ฃ .
17:59
So it's just --
355
1079160
2000
์•„์ง์€
18:01
(Applause)
356
1081160
1000
(๋ฐ•์ˆ˜)
18:02
Thanks.
357
1082160
2000
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
18:05
Not quite John Travolta yet, but we're working on that as well,
358
1085160
3000
์•„์ง์€ ์กดํŠธ๋ผ๋ณผํƒ€๊นŒ์ง€๋Š” ์•„๋‹ˆ์ง€๋งŒ ์ด๋ฏธ ์ž‘์—… ์ค‘์ด๋ž๋‹ˆ๋‹ค.
18:08
so thanks very much for your time.
359
1088160
2000
์‹œ๊ฐ„ ๋‚ด์ฃผ์…”์„œ ๊ฐ์‚ฌ๋“œ๋ฆฝ๋‹ˆ๋‹ค.
18:10
Thanks.
360
1090160
1000
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
18:11
(Applause)
361
1091160
1000
๋ฐ•์ˆ˜
18:12
CA: Incredible. That was really incredible.
362
1092160
2000
Chris Anderson: ์ •๋ง ๋†€๋ž๊ตฐ์š”.
18:14
TR: Thanks.
363
1094160
1000
Torsten Reil:๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
์ด ์›น์‚ฌ์ดํŠธ ์ •๋ณด

์ด ์‚ฌ์ดํŠธ๋Š” ์˜์–ด ํ•™์Šต์— ์œ ์šฉํ•œ YouTube ๋™์˜์ƒ์„ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ „ ์„ธ๊ณ„ ์ตœ๊ณ ์˜ ์„ ์ƒ๋‹˜๋“ค์ด ๊ฐ€๋ฅด์น˜๋Š” ์˜์–ด ์ˆ˜์—…์„ ๋ณด๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๊ฐ ๋™์˜์ƒ ํŽ˜์ด์ง€์— ํ‘œ์‹œ๋˜๋Š” ์˜์–ด ์ž๋ง‰์„ ๋”๋ธ” ํด๋ฆญํ•˜๋ฉด ๊ทธ๊ณณ์—์„œ ๋™์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค. ๋น„๋””์˜ค ์žฌ์ƒ์— ๋งž์ถฐ ์ž๋ง‰์ด ์Šคํฌ๋กค๋ฉ๋‹ˆ๋‹ค. ์˜๊ฒฌ์ด๋‚˜ ์š”์ฒญ์ด ์žˆ๋Š” ๊ฒฝ์šฐ ์ด ๋ฌธ์˜ ์–‘์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฌธ์˜ํ•˜์‹ญ์‹œ์˜ค.

https://forms.gle/WvT1wiN1qDtmnspy7