1
00:00:08,960 --> 00:00:11,919
Hello, and welcome to the Physics World weekly
2
00:00:11,919 --> 00:00:12,419
podcast,
3
00:00:12,884 --> 00:00:15,945
the last episode of 2025.
4
00:00:16,484 --> 00:00:18,664
And it features Pat Hanrahan,
5
00:00:19,125 --> 00:00:19,945
who studied
6
00:00:20,245 --> 00:00:22,585
nuclear engineering and biophysics
7
00:00:23,285 --> 00:00:25,704
before becoming a founding employee
8
00:00:26,164 --> 00:00:26,984
of Pixar
9
00:00:27,449 --> 00:00:28,669
Animation Studios
10
00:00:29,210 --> 00:00:30,589
in the nineteen eighties.
11
00:00:31,289 --> 00:00:34,429
That conversation is coming up after this message
12
00:00:34,489 --> 00:00:35,469
about a webinar
13
00:00:35,850 --> 00:00:36,909
that we're presenting
14
00:00:37,289 --> 00:00:39,229
with Leo Cancer Care.
15
00:00:39,905 --> 00:00:41,364
It's called unlocking
16
00:00:41,905 --> 00:00:45,284
novel radiation beams for cancer treatment
17
00:00:45,664 --> 00:00:47,844
with upright patient positioning,
18
00:00:48,384 --> 00:00:51,045
and it will feature a panel of leading
19
00:00:51,104 --> 00:00:51,604
clinicians
20
00:00:51,984 --> 00:00:52,804
and experts.
21
00:00:53,789 --> 00:00:56,770
They will discuss how the shift in patient
22
00:00:56,909 --> 00:00:57,409
positioning
23
00:00:57,869 --> 00:00:59,570
is enabling the exploration
24
00:01:00,109 --> 00:01:02,049
of new treatment geometries
25
00:01:02,829 --> 00:01:03,570
and supporting
26
00:01:03,949 --> 00:01:04,689
the development
27
00:01:05,150 --> 00:01:05,969
of advanced
28
00:01:06,564 --> 00:01:08,344
future cancer therapies.
29
00:01:09,284 --> 00:01:12,424
The live event is aimed at medical physicists,
30
00:01:13,284 --> 00:01:14,584
radiation oncologists,
31
00:01:15,685 --> 00:01:16,185
radiotherapy
32
00:01:16,805 --> 00:01:17,305
planners,
33
00:01:17,685 --> 00:01:18,584
and researchers
34
00:01:18,965 --> 00:01:19,864
in the field.
35
00:01:20,700 --> 00:01:24,780
Participants will learn about this paradigm shift in
36
00:01:24,780 --> 00:01:25,280
radiation
37
00:01:25,579 --> 00:01:26,079
therapy,
38
00:01:26,939 --> 00:01:28,239
gain an understanding
39
00:01:28,619 --> 00:01:30,560
of emerging and novel
40
00:01:31,019 --> 00:01:31,519
radiation
41
00:01:31,819 --> 00:01:33,040
beam modalities,
42
00:01:33,944 --> 00:01:34,765
and discover
43
00:01:35,064 --> 00:01:36,045
how foundational
44
00:01:36,825 --> 00:01:37,325
radiation
45
00:01:37,704 --> 00:01:38,204
physics
46
00:01:38,825 --> 00:01:39,325
translates
47
00:01:39,784 --> 00:01:41,805
into real world impact
48
00:01:42,105 --> 00:01:43,484
in cancer care.
49
00:01:44,265 --> 00:01:47,179
The webinar is on the February 17
50
00:01:47,640 --> 00:01:48,780
at 4PM
51
00:01:49,239 --> 00:01:49,739
GMT.
52
00:01:50,520 --> 00:01:52,380
That's 11AM
53
00:01:52,680 --> 00:01:54,299
Eastern Standard Time.
54
00:01:54,920 --> 00:01:57,420
You can register for this free event
55
00:01:57,719 --> 00:01:59,579
on the Physics World website.
56
00:02:02,165 --> 00:02:05,765
Earlier this year, Physics World's Margaret Harris attended
57
00:02:05,765 --> 00:02:06,584
the Heidelberg
58
00:02:07,045 --> 00:02:07,545
Laureate
59
00:02:07,844 --> 00:02:09,305
Forum in Germany.
60
00:02:10,004 --> 00:02:14,104
This international event brings together early career researchers
61
00:02:14,689 --> 00:02:17,349
with some winners of the most prestigious
62
00:02:17,650 --> 00:02:19,189
awards in mathematics
63
00:02:19,569 --> 00:02:21,030
and computer science.
64
00:02:21,650 --> 00:02:24,610
There, she spoke to a laureate who's also
65
00:02:24,610 --> 00:02:28,389
won three academy awards. Here's that conversation.
66
00:02:36,794 --> 00:02:39,114
So I'm talking to Pat Hanrahan, who won
67
00:02:39,114 --> 00:02:41,995
the Association for Computing Machineries AM Turing Award
68
00:02:41,995 --> 00:02:42,974
in 2019
69
00:02:43,310 --> 00:02:45,870
for contributions to three d computer graphics or
70
00:02:45,870 --> 00:02:46,370
CGI.
71
00:02:47,069 --> 00:02:48,509
So you're best known for that type of
72
00:02:48,509 --> 00:02:50,269
work, but you actually started out, I think,
73
00:02:50,269 --> 00:02:52,989
in nuclear engineering Yes. And then you got
74
00:02:52,989 --> 00:02:55,229
a PhD in biophysics. Yes. So how did
75
00:02:55,229 --> 00:02:56,935
you get into those subjects? Tell me about
76
00:02:56,935 --> 00:02:58,935
nuclear engineering first. Well, when I was in
77
00:02:58,935 --> 00:03:00,615
school, I I really felt I wanted to
78
00:03:00,615 --> 00:03:03,094
become a physicist. Anyway, I like math too,
79
00:03:03,094 --> 00:03:04,935
and I like science in general, but I
80
00:03:04,935 --> 00:03:07,415
think physics was the field that was most
81
00:03:07,415 --> 00:03:09,270
interesting to me. And so, yeah, when I
82
00:03:09,270 --> 00:03:11,030
went to the University of Wisconsin actually, it
83
00:03:11,030 --> 00:03:12,790
was it was it was doubly nice because
84
00:03:12,790 --> 00:03:14,469
they had this really nice, what they call,
85
00:03:14,469 --> 00:03:15,370
honors program,
86
00:03:15,750 --> 00:03:17,270
where you have a lot of freedom in
87
00:03:17,270 --> 00:03:19,270
what classes you took and all that. So,
88
00:03:19,270 --> 00:03:21,349
you know, not just take physics, but do
89
00:03:21,349 --> 00:03:24,064
other things as well. Take chemistry, biology, other
90
00:03:24,064 --> 00:03:24,564
sciences.
91
00:03:24,864 --> 00:03:26,625
So yeah. So I was always interested in
92
00:03:26,625 --> 00:03:28,704
physic. I love my intro physics course. I
93
00:03:28,704 --> 00:03:29,364
can still
94
00:03:29,664 --> 00:03:32,144
remember the lectures and, you know, going through
95
00:03:32,144 --> 00:03:33,044
it all. And,
96
00:03:33,344 --> 00:03:35,104
yeah. So I just loved physics, you know,
97
00:03:35,104 --> 00:03:36,884
all the way through the undergraduate
98
00:03:37,185 --> 00:03:38,084
physics, mechanics,
99
00:03:38,730 --> 00:03:39,230
electrodynamics,
100
00:03:40,169 --> 00:03:42,569
quantum mechanics. I just loved all those courses,
101
00:03:42,569 --> 00:03:43,550
my favorite courses.
102
00:03:43,849 --> 00:03:46,169
But why nuclear engineering specifically? Why not? Well,
103
00:03:46,169 --> 00:03:47,610
I think it was mostly because of this
104
00:03:47,610 --> 00:03:50,090
honors program. I was actually I actually and
105
00:03:50,090 --> 00:03:51,770
I when I went there, the chairman of
106
00:03:51,770 --> 00:03:52,349
the department,
107
00:03:52,650 --> 00:03:53,790
offered me a job,
108
00:03:54,665 --> 00:03:56,344
and he was the chair of the nuclear
109
00:03:56,344 --> 00:03:58,745
regulatory commission at the time. So I ended
110
00:03:58,745 --> 00:04:00,665
up, you know, in in this department. But,
111
00:04:00,665 --> 00:04:02,025
you know, I was he was working for
112
00:04:02,025 --> 00:04:03,784
the nuclear regulatory commission. This was, you know,
113
00:04:03,784 --> 00:04:05,165
like, in the late seventies.
114
00:04:05,705 --> 00:04:08,025
You know, nuclear power was having you know,
115
00:04:08,025 --> 00:04:10,120
people were pretty down on that. Right? And
116
00:04:10,120 --> 00:04:11,400
when I was reading about all this stuff,
117
00:04:11,400 --> 00:04:12,919
I got pretty down on it too because
118
00:04:12,919 --> 00:04:14,680
they weren't really addressing a lot of the
119
00:04:14,680 --> 00:04:16,439
currents. I think that's maybe changing a little
120
00:04:16,439 --> 00:04:17,740
bit, but I was definitely
121
00:04:18,279 --> 00:04:20,620
not that interested especially at towards the end.
122
00:04:21,160 --> 00:04:23,560
So I wanted to work on other areas
123
00:04:23,560 --> 00:04:24,220
of physics.
124
00:04:24,654 --> 00:04:26,415
I did a lot of theoretical physics. You
125
00:04:26,415 --> 00:04:28,095
know? Like, when I was a graduate student,
126
00:04:28,095 --> 00:04:29,935
I I had to take, like, I don't
127
00:04:29,935 --> 00:04:32,415
know, five courses in quantum mechanics, you know,
128
00:04:32,415 --> 00:04:34,574
quantum field theory. You know? I was really
129
00:04:34,574 --> 00:04:37,079
headed towards the theoretical physics thing, but I
130
00:04:37,079 --> 00:04:38,519
don't know what you I I the the
131
00:04:38,519 --> 00:04:41,000
low level of physics just wasn't really exciting
132
00:04:41,000 --> 00:04:42,439
me. So I I wanted to be a
133
00:04:42,439 --> 00:04:43,899
physicist, but I
134
00:04:44,360 --> 00:04:46,279
wanted to work on sort of higher level
135
00:04:46,279 --> 00:04:46,779
problems.
136
00:04:47,560 --> 00:04:50,439
So I I admired, like, Francis Crick, and
137
00:04:50,439 --> 00:04:51,660
he was, like, a theoretical
138
00:04:52,524 --> 00:04:53,904
biologist in some sense.
139
00:04:54,285 --> 00:04:55,725
And I thought, well, maybe I could become,
140
00:04:55,725 --> 00:04:57,264
like, a theoretical biologist.
141
00:04:57,964 --> 00:04:59,584
And then I met this person,
142
00:05:00,045 --> 00:05:01,964
and I was, he had was one of
143
00:05:01,964 --> 00:05:04,865
the founders of the computational neuroscience program.
144
00:05:05,319 --> 00:05:07,800
Or well, I'm sorry. Not computational neuroscience. That
145
00:05:07,800 --> 00:05:11,160
didn't exist. Neuroscience program. And, he was working
146
00:05:11,160 --> 00:05:13,319
on neurobiology of nematodes. He had worked for
147
00:05:13,319 --> 00:05:14,139
Sydney Brenner.
148
00:05:14,600 --> 00:05:16,279
And so I thought, well, I could I
149
00:05:16,279 --> 00:05:17,560
could do you know, he and they weren't
150
00:05:17,560 --> 00:05:19,319
being very quantitative, and I thought I could,
151
00:05:19,319 --> 00:05:21,555
like, apply what I know about physics to
152
00:05:21,555 --> 00:05:23,314
help them, you know, build up models of
153
00:05:23,314 --> 00:05:25,154
the nervous system. And I I started doing
154
00:05:25,154 --> 00:05:27,715
some simulations and things like that. Yeah. And
155
00:05:27,715 --> 00:05:28,995
is that about the time that you got
156
00:05:28,995 --> 00:05:31,314
first involved in Pixar, the project that became
157
00:05:31,314 --> 00:05:33,395
the first Toy Story film? No. That took
158
00:05:33,395 --> 00:05:34,675
a little while. I mean, you know, you
159
00:05:34,675 --> 00:05:37,170
have to realize computer graphics didn't exist then.
160
00:05:37,170 --> 00:05:39,410
It really didn't exist. I mean, you know,
161
00:05:39,410 --> 00:05:41,649
it just did not exist. My advice so
162
00:05:41,649 --> 00:05:43,170
it turned out it's sort of funny story.
163
00:05:43,170 --> 00:05:46,069
Sydney Brenner, you know, Nobel laureate in biology.
164
00:05:46,610 --> 00:05:48,129
There was he had two a bunch of
165
00:05:48,129 --> 00:05:51,165
colleagues and students, Francis Crick. There was another
166
00:05:51,165 --> 00:05:53,245
guy named John White and another guy named
167
00:05:53,245 --> 00:05:54,064
John Sulston.
168
00:05:54,444 --> 00:05:56,145
And John White is famous for
169
00:05:56,524 --> 00:05:57,024
determining,
170
00:05:57,645 --> 00:05:59,725
the structure of the nervous system of c
171
00:05:59,725 --> 00:06:01,245
l. So they all worked on C. Elegans,
172
00:06:01,245 --> 00:06:03,185
which is a a free living nematode.
173
00:06:03,680 --> 00:06:06,240
So, John White had, like, figured out the
174
00:06:06,240 --> 00:06:08,879
entire wiring grant by sectioning the words and
175
00:06:08,879 --> 00:06:11,360
worm and tracing all the neurons. And,
176
00:06:11,839 --> 00:06:14,019
John Solston had, worked on
177
00:06:14,560 --> 00:06:17,680
developmental biology, and he recorded, you know, how
178
00:06:17,680 --> 00:06:18,339
the cells
179
00:06:18,814 --> 00:06:20,574
divided, you know, from an egg all the
180
00:06:20,574 --> 00:06:22,435
way to a full grown one.
181
00:06:22,975 --> 00:06:24,975
And, anyways, the three of them all got
182
00:06:24,975 --> 00:06:25,954
really the same computers.
183
00:06:27,294 --> 00:06:27,794
And,
184
00:06:28,334 --> 00:06:30,495
and so, you know, my my adviser was
185
00:06:30,495 --> 00:06:32,014
in that group, and he was trying to
186
00:06:32,014 --> 00:06:33,555
convince me to work on computers.
187
00:06:34,129 --> 00:06:36,370
And I didn't have any background in computers.
188
00:06:36,370 --> 00:06:37,829
So I started simulating,
189
00:06:38,370 --> 00:06:39,589
this nervous system,
190
00:06:40,129 --> 00:06:42,209
and I had you, write these programs that
191
00:06:42,209 --> 00:06:43,829
simulated. I was, like, solving,
192
00:06:44,209 --> 00:06:47,169
like, basically cable equations for neurons because these
193
00:06:47,169 --> 00:06:49,189
neurons are, like, you know, electrical cables.
194
00:06:49,995 --> 00:06:52,074
But I needed to do visualization. I had
195
00:06:52,074 --> 00:06:53,435
to be able to see my result. Like,
196
00:06:53,435 --> 00:06:55,915
you just so my roommate introduced me to
197
00:06:55,915 --> 00:06:57,275
computer graphics. He said, oh, by the way,
198
00:06:57,275 --> 00:06:58,975
do you know about this graphic stuff?
199
00:06:59,355 --> 00:07:00,975
You know, I just sort of got enchanted
200
00:07:01,115 --> 00:07:02,235
by what I could do with it because
201
00:07:02,235 --> 00:07:03,835
I could take all my theories, like, whether
202
00:07:03,835 --> 00:07:06,580
it was simulation of neurons or whatever I
203
00:07:06,580 --> 00:07:08,819
was interested in and see it, you know,
204
00:07:08,819 --> 00:07:11,080
make it really concrete. And it just became
205
00:07:11,779 --> 00:07:13,460
even more fun. You know? Like, the the
206
00:07:13,460 --> 00:07:14,439
physics was suddenly
207
00:07:14,740 --> 00:07:16,500
infinitely more fun because I could see it.
208
00:07:16,500 --> 00:07:18,564
You know? Like, I could do animations,
209
00:07:18,944 --> 00:07:20,305
not much animation, but I could at least
210
00:07:20,305 --> 00:07:21,904
make pictures and so on. So I sort
211
00:07:21,904 --> 00:07:24,145
of got seduced by that idea, and it
212
00:07:24,145 --> 00:07:26,225
it was a wide open field. You know?
213
00:07:26,225 --> 00:07:28,305
Physics and, you know, biology was, you know,
214
00:07:28,305 --> 00:07:30,680
sort of slow paced. This was, like, completely
215
00:07:30,680 --> 00:07:32,759
wide open. So, like like like, the first
216
00:07:32,759 --> 00:07:34,280
week I worked on it, I, like, wrote
217
00:07:34,280 --> 00:07:36,280
a paper. I got accepted in the major
218
00:07:36,280 --> 00:07:37,800
conference. You know? Whereas in physics, you know,
219
00:07:37,800 --> 00:07:39,319
I'd been a grad student for a couple
220
00:07:39,319 --> 00:07:41,560
years, never written a paper. And so then
221
00:07:41,560 --> 00:07:43,080
I decided I had you go learn more
222
00:07:43,080 --> 00:07:44,220
about computer graphics,
223
00:07:44,794 --> 00:07:46,714
but, you know, because there was nobody there.
224
00:07:46,714 --> 00:07:48,954
There was only one or two places in
225
00:07:48,954 --> 00:07:50,794
the world that was doing computer graphics, and
226
00:07:50,794 --> 00:07:53,435
eventually, I ended up at Pixar a couple
227
00:07:53,435 --> 00:07:54,495
years later. Yeah.
228
00:07:55,435 --> 00:07:58,314
And a lot of your research on computer
229
00:07:58,314 --> 00:08:01,720
generated imagery focuses on getting the simulated world
230
00:08:01,720 --> 00:08:02,459
to display
231
00:08:03,079 --> 00:08:05,000
things in the way it's grounded in physics.
232
00:08:05,000 --> 00:08:06,519
Yeah. Yeah. You know, you literally wrote the
233
00:08:06,519 --> 00:08:09,479
book about physically based rendering. Yeah. What are
234
00:08:09,479 --> 00:08:11,639
the benefits of that approach, and what are
235
00:08:11,639 --> 00:08:14,055
some of the main challenges in it? That's
236
00:08:14,055 --> 00:08:15,035
a great question.
237
00:08:15,335 --> 00:08:17,095
You know, I love that question. I mean,
238
00:08:17,095 --> 00:08:18,694
you know, you know, we're we're creating a
239
00:08:18,694 --> 00:08:21,335
virtual world, and the world, you know, has
240
00:08:21,335 --> 00:08:23,495
physics in it. I mean, you know, and
241
00:08:23,495 --> 00:08:25,175
just imagine. I mean, what we did, we're
242
00:08:25,175 --> 00:08:26,694
just trying to create a world that, like,
243
00:08:26,694 --> 00:08:29,120
if somebody looked at it had all the
244
00:08:29,120 --> 00:08:31,439
phenomena that we think of in physics and
245
00:08:31,439 --> 00:08:33,759
even biology as well, but all these phenomena.
246
00:08:33,759 --> 00:08:35,160
So, you know, like, what are some of
247
00:08:35,160 --> 00:08:37,299
the found well, a cloud might move. Right?
248
00:08:37,759 --> 00:08:40,019
Wind will move the leaves of the tree.
249
00:08:40,079 --> 00:08:41,519
You know, you could have an object, a
250
00:08:41,519 --> 00:08:42,339
rock falling.
251
00:08:42,725 --> 00:08:44,725
You know, get out something a building kind
252
00:08:44,725 --> 00:08:47,045
of blowing apart. You could have, you know,
253
00:08:47,045 --> 00:08:50,485
cloth, you know, conforming to somebody's skin. You
254
00:08:50,485 --> 00:08:52,965
have the muscles moving the body. I mean,
255
00:08:52,965 --> 00:08:54,965
there's just physics everywhere in the world. Right?
256
00:08:54,965 --> 00:08:56,570
And so, you know, if we're trying to
257
00:08:56,570 --> 00:08:57,970
simulate all I mean, we're trying to make
258
00:08:57,970 --> 00:08:59,690
a picture of all this stuff. I mean,
259
00:08:59,690 --> 00:09:01,050
how else would you do it? I mean,
260
00:09:01,050 --> 00:09:02,889
you you have to in some sense, you
261
00:09:02,889 --> 00:09:04,889
have to somehow capture the physics of the
262
00:09:04,889 --> 00:09:06,490
everyday world. It's I I call it the
263
00:09:06,490 --> 00:09:08,970
everyday world because it's at the human scale.
264
00:09:08,970 --> 00:09:11,115
You know? It's what we see around us.
265
00:09:11,355 --> 00:09:13,534
And I mostly worked on,
266
00:09:13,914 --> 00:09:16,075
simulating the appearance of it, right, which is
267
00:09:16,075 --> 00:09:17,595
like, you know, you know, why does it
268
00:09:17,595 --> 00:09:19,274
look the way it does? Like, why how
269
00:09:19,274 --> 00:09:22,315
do, like, materials scatter light, and how does
270
00:09:22,315 --> 00:09:25,259
light get transported through the environment, and, you
271
00:09:25,259 --> 00:09:27,899
know, even how cameras simulate cameras and simulate
272
00:09:27,899 --> 00:09:29,820
light sources. So, yeah, so if you look
273
00:09:29,820 --> 00:09:31,420
at the what we actually ended up doing,
274
00:09:31,420 --> 00:09:33,179
if you look at that book, you know,
275
00:09:33,179 --> 00:09:34,080
we write down
276
00:09:34,460 --> 00:09:36,700
a bunch of equations that describe the scattering
277
00:09:36,700 --> 00:09:39,279
of light off surfaces and off infinite media,
278
00:09:39,465 --> 00:09:41,245
and then we just solve this
279
00:09:42,024 --> 00:09:44,264
solve for the lightning distribution in the world.
280
00:09:44,264 --> 00:09:46,345
You know? And it's a just a giant
281
00:09:46,345 --> 00:09:47,725
computational science problem.
282
00:09:48,105 --> 00:09:49,945
But the challenge is you have all these
283
00:09:49,945 --> 00:09:50,925
real world materials,
284
00:09:52,345 --> 00:09:53,884
which are both complex.
285
00:09:54,529 --> 00:09:57,009
You know, they're not just like ideal mirrors
286
00:09:57,009 --> 00:09:58,929
and, you know, one type of cars where
287
00:09:58,929 --> 00:10:00,690
I just, like, opened up an optics book
288
00:10:00,690 --> 00:10:02,290
and said, how much is in this book
289
00:10:02,290 --> 00:10:04,389
that would help us do this? Almost nothing.
290
00:10:04,690 --> 00:10:06,210
You know what I mean? Because, you know,
291
00:10:06,210 --> 00:10:07,889
how does light scatter off skin? How does
292
00:10:07,889 --> 00:10:10,085
light scatter off a rock? How does light
293
00:10:10,085 --> 00:10:12,565
scatter off a cloud? Just not detailed stuff
294
00:10:12,565 --> 00:10:14,644
in there. So we'd often have to come
295
00:10:14,644 --> 00:10:15,625
up with, you know,
296
00:10:16,004 --> 00:10:16,504
approximations
297
00:10:16,884 --> 00:10:17,784
and models
298
00:10:18,164 --> 00:10:20,325
for all these things. And I worked a
299
00:10:20,325 --> 00:10:22,164
lot. Like, I I worked on my papers
300
00:10:22,164 --> 00:10:23,464
as I'm modeling skin.
301
00:10:24,419 --> 00:10:26,339
Hair. How does light scatter from hair? How
302
00:10:26,339 --> 00:10:28,899
does light scatter from cloth? Stuff like that.
303
00:10:28,899 --> 00:10:30,339
But, anyways, when it all is said and
304
00:10:30,339 --> 00:10:31,940
done, you get, you know, like, a billion
305
00:10:31,940 --> 00:10:32,440
objects
306
00:10:32,820 --> 00:10:34,679
each with really complex materials,
307
00:10:35,059 --> 00:10:37,375
and then you have to solve for the
308
00:10:37,375 --> 00:10:38,115
lighting distribution,
309
00:10:38,574 --> 00:10:41,454
which is really complicated because the objects have
310
00:10:41,454 --> 00:10:44,274
complicated shape. So we use Monte Carlo techniques.
311
00:10:44,495 --> 00:10:46,174
You know? In fact, just like what they
312
00:10:46,174 --> 00:10:48,815
did up during the war to simulate the
313
00:10:48,815 --> 00:10:49,794
neutron transport,
314
00:10:50,110 --> 00:10:51,870
And that's actually was the connection back to
315
00:10:51,870 --> 00:10:54,509
my work in nuclear engineering. I had learned
316
00:10:54,509 --> 00:10:57,230
about transport theory and Monte Carlo techniques, you
317
00:10:57,230 --> 00:10:59,309
know, in school. And so, you know, I
318
00:10:59,309 --> 00:11:01,009
I sort of also just sort of gravitated
319
00:11:01,070 --> 00:11:02,590
towards those problems because those are the ones
320
00:11:02,590 --> 00:11:03,570
I knew to sell.
321
00:11:03,904 --> 00:11:06,144
So you actually won, one of your three
322
00:11:06,144 --> 00:11:08,544
Academy Awards, Oscars, in other words Yeah. Is
323
00:11:08,544 --> 00:11:11,585
specifically for developing that software that simulates the
324
00:11:11,585 --> 00:11:13,665
subsurface scattering of light Yes. Yeah. In in
325
00:11:13,665 --> 00:11:16,304
materials that are translucent. Yeah. Yeah. Why is
326
00:11:16,304 --> 00:11:18,144
that so difficult, and why is it so
327
00:11:18,144 --> 00:11:18,644
important
328
00:11:19,340 --> 00:11:21,980
for film and computer graphics? Yeah. Well, okay.
329
00:11:21,980 --> 00:11:23,980
Why is it important? Well, I'm looking at
330
00:11:23,980 --> 00:11:26,080
you, and your skin looks really healthy.
331
00:11:27,019 --> 00:11:28,300
I mean, you know, I mean, you know,
332
00:11:28,300 --> 00:11:30,940
in real people, skin is really important. Right?
333
00:11:30,940 --> 00:11:33,340
It has to look right. And until then,
334
00:11:33,340 --> 00:11:34,399
it looked like plastic.
335
00:11:34,715 --> 00:11:36,154
You know, it looked like, yeah, just were
336
00:11:36,154 --> 00:11:38,235
completely covered in makeup. You know? You know,
337
00:11:38,235 --> 00:11:40,554
makeup and things like that, you know, almost
338
00:11:40,554 --> 00:11:41,915
like pink. You know what I mean? They're
339
00:11:41,915 --> 00:11:43,915
just very matte and, you know, very I
340
00:11:43,915 --> 00:11:45,995
mean, you know, that's why they're matte. They're
341
00:11:45,995 --> 00:11:47,740
not they did not you know, don't have
342
00:11:47,740 --> 00:11:49,180
all the complexity of real skin, but real
343
00:11:49,180 --> 00:11:51,820
skin has oil. It has texture to it.
344
00:11:51,820 --> 00:11:52,480
It has
345
00:11:52,940 --> 00:11:54,700
a bunch of properties that are very different
346
00:11:54,700 --> 00:11:57,519
than these idealized plastic like, paint like material.
347
00:11:57,820 --> 00:11:59,899
So so, obviously, for the movies, they want
348
00:11:59,899 --> 00:12:02,559
people to look, you know, natural and appealing.
349
00:12:03,144 --> 00:12:04,664
So that was one that was sort of
350
00:12:04,664 --> 00:12:06,985
a motivation for it. What was interesting, the
351
00:12:06,985 --> 00:12:08,365
discovery was that
352
00:12:08,825 --> 00:12:09,565
most materials
353
00:12:10,105 --> 00:12:13,225
are actually somewhat translucent, and subsurface scatter is
354
00:12:13,225 --> 00:12:15,929
actually more common than just normal surface scatter.
355
00:12:15,929 --> 00:12:17,690
So what happened in surface scatter, which you
356
00:12:17,690 --> 00:12:19,450
learn about in physics, you know, light ray
357
00:12:19,450 --> 00:12:21,769
hits the surface and just gets reflected off
358
00:12:21,769 --> 00:12:24,170
the interface between, say, the air and the
359
00:12:24,170 --> 00:12:26,809
the material itself. Yeah. Yeah. But in most
360
00:12:26,809 --> 00:12:28,809
materials, they're actually dielectric. And by the way,
361
00:12:28,809 --> 00:12:30,410
that's what happens in the metal. You know?
362
00:12:30,410 --> 00:12:32,334
Like, light can't go into a metal because
363
00:12:32,334 --> 00:12:32,995
it's conductive,
364
00:12:33,534 --> 00:12:34,914
so it has to get reflected.
365
00:12:35,294 --> 00:12:38,095
Okay? That's why metals are shiny. But most
366
00:12:38,095 --> 00:12:40,334
materials are dielectrics. You know? They have an
367
00:12:40,334 --> 00:12:42,815
index of refraction. And so most light goes
368
00:12:42,815 --> 00:12:44,815
into the materials. It doesn't get reflected. You
369
00:12:44,815 --> 00:12:46,589
know? You just look at it physically. And
370
00:12:46,589 --> 00:12:48,829
why are the dialectics? Well, they're rocks. You
371
00:12:48,829 --> 00:12:51,069
know? They're based on water. Like, we have
372
00:12:51,069 --> 00:12:53,009
a lot of water in our human body.
373
00:12:53,069 --> 00:12:54,929
You know? They're made of things like fibers,
374
00:12:55,149 --> 00:12:56,449
anything like biological.
375
00:12:57,149 --> 00:12:59,534
And most minerals are not metals. They're,
376
00:12:59,934 --> 00:13:02,095
you know, they're dielectric. So almost everything around
377
00:13:02,095 --> 00:13:03,634
us is a dielectric,
378
00:13:04,495 --> 00:13:06,815
and what happens is the light goes into
379
00:13:06,815 --> 00:13:09,134
the material and then it scatters around a
380
00:13:09,134 --> 00:13:11,054
little bit, and then it comes up. It
381
00:13:11,054 --> 00:13:13,294
looks like it's reflecting off it. Let's actually
382
00:13:13,294 --> 00:13:15,240
go on and scatter around, come out, And
383
00:13:15,240 --> 00:13:17,019
that's called subsurface reflection.
384
00:13:17,639 --> 00:13:19,320
You know, it's been studied, but it's never
385
00:13:19,320 --> 00:13:21,159
really been studied very much in natural no
386
00:13:21,159 --> 00:13:23,799
normal everyday materials like skin or whatever. So
387
00:13:23,799 --> 00:13:25,899
we we develop you know, like a physicist,
388
00:13:26,200 --> 00:13:28,600
you know, just, you know, we read models
389
00:13:28,600 --> 00:13:30,415
about how you might do that. I used
390
00:13:30,415 --> 00:13:33,075
to read books like people like in atmospheric
391
00:13:33,375 --> 00:13:36,254
sciences study that, meteorology study that, like, if
392
00:13:36,254 --> 00:13:38,035
you observe a planetary atmosphere
393
00:13:38,575 --> 00:13:40,335
from a distance, like, it's like you have
394
00:13:40,335 --> 00:13:43,554
this layer of gas around the planet,
395
00:13:43,879 --> 00:13:45,559
and so it's gonna be sort of like
396
00:13:45,559 --> 00:13:47,159
that. The light goes into the atmosphere and
397
00:13:47,159 --> 00:13:48,839
gets scattered. So there, you know, there's things
398
00:13:48,839 --> 00:13:50,039
known about it, but they were all like
399
00:13:50,039 --> 00:13:52,139
things like that, like, you know, planetary atmosphere.
400
00:13:52,519 --> 00:13:54,360
So we yeah. We we just took a
401
00:13:54,360 --> 00:13:56,200
lot of those ideas that they had developed
402
00:13:56,200 --> 00:13:58,085
and and, you know, sort of, you know,
403
00:13:58,085 --> 00:14:00,264
mod used them to model things like skin.
404
00:14:00,644 --> 00:14:02,165
And in the same thing with hair, what
405
00:14:02,165 --> 00:14:03,625
happens is hair is mostly,
406
00:14:04,245 --> 00:14:06,325
translucent as well. It's like a little optical
407
00:14:06,325 --> 00:14:08,165
fiber. And so light goes into it, bounces
408
00:14:08,165 --> 00:14:10,085
around, and comes out. So these just lead
409
00:14:10,085 --> 00:14:12,264
to very different appearances than
410
00:14:12,629 --> 00:14:15,110
metals and things like that. So yeah. So
411
00:14:15,110 --> 00:14:16,230
I I did a lot of work on
412
00:14:16,230 --> 00:14:18,809
on, you know, just modeling different materials. And
413
00:14:18,870 --> 00:14:21,210
the subsurface scattering works, I mean, since everything
414
00:14:21,429 --> 00:14:23,750
looked like plastic, as soon as we wrote
415
00:14:23,750 --> 00:14:26,149
the paper, I mean, every studio in the
416
00:14:26,149 --> 00:14:27,210
world immediately,
417
00:14:28,414 --> 00:14:30,975
adopted them. And we we published the source
418
00:14:30,975 --> 00:14:34,095
code, and it was relatively simple model, so
419
00:14:34,095 --> 00:14:35,855
they could just, like, take it and put
420
00:14:35,855 --> 00:14:36,995
it into their system.
421
00:14:37,454 --> 00:14:39,294
And so yeah. So they like, within a
422
00:14:39,294 --> 00:14:41,389
year or two, everybody was using it sort
423
00:14:41,389 --> 00:14:43,230
of very quickly. And but, you know, it's
424
00:14:43,309 --> 00:14:44,669
it was I think it was the first
425
00:14:44,669 --> 00:14:47,870
time an academic paper had ever gotten an
426
00:14:47,870 --> 00:14:48,769
Academy Award.
427
00:14:50,829 --> 00:14:52,769
What would you say is the most difficult
428
00:14:53,115 --> 00:14:55,995
CGI problem either you've solved or you've seen
429
00:14:55,995 --> 00:14:58,875
someone else solve by some building on your
430
00:14:58,875 --> 00:14:59,375
work?
431
00:14:59,914 --> 00:15:01,455
Well, I I I think the
432
00:15:01,914 --> 00:15:04,095
I think the hardest thing to solve
433
00:15:04,794 --> 00:15:06,554
was not just, like, how to make, you
434
00:15:06,554 --> 00:15:08,940
know, skin look right. I mean, that's really
435
00:15:08,940 --> 00:15:10,160
insane, hard problem.
436
00:15:10,779 --> 00:15:12,639
It's how to build a system
437
00:15:13,259 --> 00:15:14,639
that would include everything.
438
00:15:15,179 --> 00:15:16,620
You know what I mean? Like, it's like
439
00:15:16,620 --> 00:15:18,620
this it's like sort of grand unified theory.
440
00:15:18,620 --> 00:15:21,519
Right? How can I build one rendering system
441
00:15:21,924 --> 00:15:23,625
that would handle any shape
442
00:15:23,925 --> 00:15:27,144
and any material and any light source,
443
00:15:27,605 --> 00:15:28,985
you know, and any kind of,
444
00:15:29,365 --> 00:15:32,165
you know, participating media and and just build
445
00:15:32,165 --> 00:15:34,325
a system that could do it all? And
446
00:15:34,325 --> 00:15:35,605
the problem is you if you if you
447
00:15:35,605 --> 00:15:37,259
can't do a bunch of stuff, like, say
448
00:15:37,259 --> 00:15:38,720
you couldn't do the atmosphere.
449
00:15:39,179 --> 00:15:40,700
I mean, it's just not gonna be useful.
450
00:15:40,700 --> 00:15:42,620
Right? And if you can't do people, it's
451
00:15:42,620 --> 00:15:44,559
not gonna be useful. You can't do plants.
452
00:15:44,860 --> 00:15:46,460
You know, like plants were a huge problem
453
00:15:46,460 --> 00:15:48,300
because there's so many leaves and they have
454
00:15:48,300 --> 00:15:50,460
such complicated scattering, but you have billions of
455
00:15:50,460 --> 00:15:52,559
them. You know? Billions of leaves and
456
00:15:53,035 --> 00:15:55,274
dandelions and stuff. So, you know, just having
457
00:15:55,274 --> 00:15:57,514
a system that could handle all this stuff
458
00:15:57,514 --> 00:15:59,615
and scale to the complexities,
459
00:16:00,475 --> 00:16:02,955
I mean, it's it's mostly an engineering problem,
460
00:16:02,955 --> 00:16:04,495
but it was it was a conceptualization
461
00:16:05,195 --> 00:16:07,610
problem. Like, how do you break down,
462
00:16:08,070 --> 00:16:09,929
you know, the world in some sense
463
00:16:10,389 --> 00:16:12,870
into all these what are some you know,
464
00:16:12,870 --> 00:16:14,070
you know, we sort of use, like, object
465
00:16:14,070 --> 00:16:16,070
oriented, but what are all the objects you
466
00:16:16,070 --> 00:16:18,389
need, and how are they all interrelated? How
467
00:16:18,389 --> 00:16:20,309
do they all work together? And then so
468
00:16:20,309 --> 00:16:22,605
that you can, like, plug in something new.
469
00:16:22,605 --> 00:16:24,205
Like, when we had that skin model, like,
470
00:16:24,205 --> 00:16:25,565
you just plug it in. You know, the
471
00:16:25,565 --> 00:16:27,804
system had the design. So that was actually
472
00:16:27,804 --> 00:16:29,485
my first academy where we had this system
473
00:16:29,485 --> 00:16:30,705
we called Reyes,
474
00:16:31,165 --> 00:16:33,585
R E Y E S, for the California.
475
00:16:33,725 --> 00:16:35,825
There's this park called Point Reyes.
476
00:16:36,740 --> 00:16:38,339
It's a very beautiful place. It was near
477
00:16:38,339 --> 00:16:41,779
Lucasfilm, so everybody knew about it. But it
478
00:16:41,779 --> 00:16:44,259
it was the acronym for renders everything you
479
00:16:44,259 --> 00:16:46,419
ever saw. So, you know, the first thing
480
00:16:46,500 --> 00:16:48,259
what I mostly worked on at Pixar was
481
00:16:48,259 --> 00:16:50,024
just how to build this system. You know?
482
00:16:50,105 --> 00:16:52,105
How how to build a system that renders
483
00:16:52,105 --> 00:16:54,264
everything you ever saw. And, you know, it's
484
00:16:54,425 --> 00:16:55,404
it was very,
485
00:16:56,024 --> 00:16:58,264
very challenging and non trivial to think of
486
00:16:58,264 --> 00:17:00,264
how to design this thing so it would
487
00:17:00,264 --> 00:17:01,705
work. And it's still being used. I mean,
488
00:17:01,705 --> 00:17:03,144
I did this many, many years ago. But,
489
00:17:03,144 --> 00:17:04,904
you know, once you get the basic framework
490
00:17:04,904 --> 00:17:05,549
in place,
491
00:17:06,110 --> 00:17:07,869
you know, you can just keep enhancing it
492
00:17:07,869 --> 00:17:09,630
because you can add stuff to it and
493
00:17:09,630 --> 00:17:11,070
they it just sort of you know, you
494
00:17:11,070 --> 00:17:13,309
have the right design when something new comes
495
00:17:13,309 --> 00:17:15,070
along and just sort of clicks into place.
496
00:17:15,070 --> 00:17:16,670
You know what I mean? So that that
497
00:17:16,750 --> 00:17:18,190
I think that was the hardest part. I
498
00:17:18,190 --> 00:17:20,194
mean, you know, maybe not quite as mathematical,
499
00:17:20,335 --> 00:17:22,515
but just how to conceive of this thing
500
00:17:22,654 --> 00:17:24,595
in a way that, you know, was both
501
00:17:24,654 --> 00:17:27,214
conceptually elegant, but also worked. You know what
502
00:17:27,214 --> 00:17:29,315
I mean? It could be efficient and and
503
00:17:29,454 --> 00:17:30,815
and so on. So that I think that
504
00:17:30,815 --> 00:17:32,759
was the hardest problem. Yeah. What would you
505
00:17:32,759 --> 00:17:35,559
say then is the the most difficult unsolved
506
00:17:35,559 --> 00:17:37,400
problem? What what can't you do now that
507
00:17:37,400 --> 00:17:38,680
you really like to be able to do
508
00:17:38,680 --> 00:17:39,740
in computer graphics?
509
00:17:40,519 --> 00:17:41,880
Well, I think that I have a couple
510
00:17:41,880 --> 00:17:43,400
of opinions. I mean, you know, on the
511
00:17:43,400 --> 00:17:45,340
on the so we would call this photorealistic
512
00:17:45,720 --> 00:17:47,545
rendering and so on. I mean, I don't
513
00:17:47,625 --> 00:17:49,144
I mean, pretty much, you know, we've been
514
00:17:49,144 --> 00:17:50,825
working on it so long. I mean, I
515
00:17:50,825 --> 00:17:52,424
don't know if you go to movies. Some
516
00:17:52,424 --> 00:17:53,705
of the people do, but, you know, the
517
00:17:53,705 --> 00:17:55,865
movies are pretty convincing at this point. You
518
00:17:55,865 --> 00:17:57,464
know? And even like Toy Story was the
519
00:17:57,464 --> 00:17:59,224
first one that we did. It was a
520
00:17:59,224 --> 00:18:00,365
cartoon movie.
521
00:18:00,664 --> 00:18:02,205
Now, you know, you see movies
522
00:18:02,920 --> 00:18:04,380
routinely that just
523
00:18:04,920 --> 00:18:05,900
is completely indistinguishable
524
00:18:06,519 --> 00:18:07,019
from,
525
00:18:07,320 --> 00:18:09,640
you know, from things like Avatar and, you
526
00:18:09,640 --> 00:18:12,119
know, all these movie. So it's pretty hard.
527
00:18:12,119 --> 00:18:13,240
I mean, I don't know if I can
528
00:18:13,240 --> 00:18:14,059
think of really,
529
00:18:14,599 --> 00:18:17,559
one thing. I find natural scenes, outdoor scenes,
530
00:18:17,559 --> 00:18:19,305
sort of still somewhat unsatisfying.
531
00:18:20,085 --> 00:18:21,765
I mean, remember, they still fake a lot
532
00:18:21,765 --> 00:18:23,465
of stuff. They don't model everything
533
00:18:24,085 --> 00:18:25,924
and so on, so they can avoid, you
534
00:18:25,924 --> 00:18:27,865
know, showing you things that they can't do.
535
00:18:27,924 --> 00:18:30,005
But I find, you know, really if I
536
00:18:30,005 --> 00:18:31,865
take a walk in the forest or something,
537
00:18:32,119 --> 00:18:33,960
you know, there's something about it that I
538
00:18:33,960 --> 00:18:35,880
don't I don't see in the computer, but
539
00:18:35,880 --> 00:18:37,240
I can't quite pin it down. That's the
540
00:18:37,240 --> 00:18:39,160
other problem. What? You're you're looking at this
541
00:18:39,160 --> 00:18:40,940
stuff. Like, you look I'm looking at skin.
542
00:18:41,240 --> 00:18:43,240
Like, what are we not getting about it?
543
00:18:43,240 --> 00:18:45,480
You know? There's something underlying the phenomena of
544
00:18:45,480 --> 00:18:47,734
the scattering. So I don't quite know exactly
545
00:18:47,795 --> 00:18:48,615
for these
546
00:18:48,994 --> 00:18:49,494
natural,
547
00:18:50,195 --> 00:18:53,234
environments what we're getting wrong. But, anyway, I
548
00:18:53,234 --> 00:18:54,994
think that's one. But I actually think, you
549
00:18:54,994 --> 00:18:55,494
know,
550
00:18:55,955 --> 00:18:58,615
a more interesting thing is not just photorealistic
551
00:18:59,154 --> 00:19:01,029
effects, but, you know, how to make it
552
00:19:01,029 --> 00:19:02,710
easy. You know, like, I gave a talk
553
00:19:02,710 --> 00:19:05,589
once about starting with Leonardo, how, you know,
554
00:19:05,589 --> 00:19:08,970
we sort of reproduced what Leonardo had discovered,
555
00:19:09,029 --> 00:19:11,349
right, like perspective and shading and all that
556
00:19:11,349 --> 00:19:13,289
stuff, and we just computerized it.
557
00:19:13,865 --> 00:19:15,465
But then the event art sort of went
558
00:19:15,465 --> 00:19:17,785
in lots of different directions. It wasn't just
559
00:19:17,785 --> 00:19:19,865
about realism. You know? So, like, maybe how
560
00:19:19,865 --> 00:19:21,485
to make other kinds of pictures,
561
00:19:22,105 --> 00:19:24,424
like, either more abstract kind of pictures like
562
00:19:24,424 --> 00:19:26,930
an artist might do, or more functional kind
563
00:19:26,930 --> 00:19:28,390
of pictures, like, in visualizations
564
00:19:28,690 --> 00:19:30,529
of things. You know, just how you know,
565
00:19:30,529 --> 00:19:32,609
because there you're not just simulating the physics.
566
00:19:32,609 --> 00:19:34,130
You have to ask but you still you
567
00:19:34,130 --> 00:19:36,049
know, it might be some formal technique. Let's
568
00:19:36,049 --> 00:19:37,589
say we had some theory of aesthetics
569
00:19:38,265 --> 00:19:40,345
or something like that. Maybe we could formalize
570
00:19:40,345 --> 00:19:42,025
that and have a computer do that kind
571
00:19:42,025 --> 00:19:43,384
of stuff. Yeah. You see some of that
572
00:19:43,384 --> 00:19:45,785
generated by AI right now. But, yeah, just,
573
00:19:45,785 --> 00:19:47,465
you know, other there are different kinds of
574
00:19:47,465 --> 00:19:47,965
picture.
575
00:19:48,585 --> 00:19:50,505
And I think we could make a lot
576
00:19:50,505 --> 00:19:52,904
of other interesting pictures besides just sort of
577
00:19:52,904 --> 00:19:53,404
photographic,
578
00:19:53,750 --> 00:19:55,529
You know, sort of a classic thing, like,
579
00:19:55,830 --> 00:19:57,509
first we once we had a camera, like,
580
00:19:57,509 --> 00:19:58,789
you know, what do we why do we
581
00:19:58,789 --> 00:20:00,710
need to you know, we don't want it
582
00:20:00,710 --> 00:20:02,950
to we have cameras to take realistic pictures.
583
00:20:02,950 --> 00:20:05,109
We don't need, you know, artists to draw
584
00:20:05,109 --> 00:20:07,634
perfectly realistic pictures. Yeah. So you're still looking
585
00:20:07,634 --> 00:20:10,115
for the Pixar version of Picasso. Yeah. Yeah.
586
00:20:10,115 --> 00:20:12,455
Exactly. Yeah. Yeah. Yeah. Exactly. Exactly.
587
00:20:13,075 --> 00:20:14,994
So final question. How if you could just
588
00:20:14,994 --> 00:20:16,275
sum it up, how do you think your
589
00:20:16,275 --> 00:20:18,835
training in physics has helped you in your
590
00:20:18,835 --> 00:20:19,335
career?
591
00:20:19,759 --> 00:20:21,359
Well, I certainly learned a lot of phys
592
00:20:21,519 --> 00:20:23,619
you know, physical phenomena, you know,
593
00:20:24,240 --> 00:20:25,059
like transport,
594
00:20:25,519 --> 00:20:26,339
light optics,
595
00:20:26,960 --> 00:20:28,980
and mechanics, classical mechanics.
596
00:20:29,440 --> 00:20:32,085
You know, lots of, particular things I learned.
597
00:20:32,325 --> 00:20:33,924
But the best thing I learned was how
598
00:20:33,924 --> 00:20:34,825
to solve problems.
599
00:20:36,085 --> 00:20:37,684
I mean, you know, I mean, you just
600
00:20:37,845 --> 00:20:39,365
you know, the problems they gave you to
601
00:20:39,365 --> 00:20:40,804
solving, you just had to solve. And there
602
00:20:40,804 --> 00:20:42,404
was a certain way of solving them. You
603
00:20:42,404 --> 00:20:44,404
know? How do you like, you notice I
604
00:20:44,404 --> 00:20:46,484
use the word modeling and simulation a lot.
605
00:20:46,484 --> 00:20:48,849
You know? Like, you have physical models and,
606
00:20:49,230 --> 00:20:51,230
you know, models are useful, but they're not
607
00:20:51,230 --> 00:20:52,509
the whole thing. You have to, you know,
608
00:20:52,509 --> 00:20:54,049
you have to pick the right model
609
00:20:54,509 --> 00:20:56,269
for, you know, you know, like, start out
610
00:20:56,269 --> 00:20:57,649
with really phenomenological
611
00:20:57,950 --> 00:20:59,470
like models, but they can still be super
612
00:20:59,470 --> 00:21:02,049
useful when you also want more fundamental models.
613
00:21:02,174 --> 00:21:03,855
But there's many models, and you have to
614
00:21:03,855 --> 00:21:05,695
pick the right one. And then and so
615
00:21:05,695 --> 00:21:07,394
you just have to pick the right one.
616
00:21:07,695 --> 00:21:08,975
And, you know, a lot of people, they
617
00:21:08,975 --> 00:21:10,735
they just they don't know how to pick
618
00:21:10,735 --> 00:21:12,414
the right model. You know what I mean?
619
00:21:12,414 --> 00:21:13,775
And then you and then you have to
620
00:21:13,775 --> 00:21:14,755
crank through whatever
621
00:21:15,089 --> 00:21:17,569
it takes to sort of make it solvable.
622
00:21:17,569 --> 00:21:19,169
You know what I mean? Like, you know,
623
00:21:19,169 --> 00:21:21,169
it might involve spherical harmonics or it might
624
00:21:21,169 --> 00:21:22,690
involve a bunch of stuff. And then some
625
00:21:22,690 --> 00:21:24,690
of that stuff sort of just difficult. Right?
626
00:21:24,690 --> 00:21:26,769
You just gotta have the the you have
627
00:21:26,769 --> 00:21:29,410
to be, like, powerful problem solver to actually
628
00:21:29,410 --> 00:21:31,434
turn that into, like, something that works. So
629
00:21:31,434 --> 00:21:32,875
I think that's what I learned in physics.
630
00:21:32,875 --> 00:21:34,575
And, you know, that skill,
631
00:21:35,674 --> 00:21:37,674
it was I just think that's why physicists
632
00:21:37,674 --> 00:21:39,434
are so successful when they go to other
633
00:21:39,434 --> 00:21:41,434
fields, is they just have this magical ability
634
00:21:41,434 --> 00:21:43,615
to solve problems in other fields. Yeah.
635
00:21:44,179 --> 00:21:45,700
That kind of fun. Thank you very much.
636
00:21:45,700 --> 00:21:46,679
Yeah. Thank you.
637
00:21:53,859 --> 00:21:57,220
That was Margaret Harris in conversation with Pat
638
00:21:57,220 --> 00:21:57,720
Hanrahan,
639
00:21:58,515 --> 00:22:00,375
who has won three Oscars
640
00:22:00,755 --> 00:22:02,775
in addition to the ACM's
641
00:22:03,154 --> 00:22:04,855
AM Turing Award.
642
00:22:05,154 --> 00:22:07,414
Thanks to both of them for a fascinating
643
00:22:07,634 --> 00:22:08,134
conversation.
644
00:22:08,994 --> 00:22:11,875
And thank you for listening to the Physics
645
00:22:11,875 --> 00:22:13,335
World weekly podcast
646
00:22:13,909 --> 00:22:15,529
in 2025.
647
00:22:16,150 --> 00:22:17,450
I'm Hamish Johnston,
648
00:22:17,750 --> 00:22:20,409
and our producer is Fred Iles.
649
00:22:20,950 --> 00:22:22,089
We'll be back again
650
00:22:22,390 --> 00:22:24,089
in 2026.
00:00:08,960 --> 00:00:11,919
Hello, and welcome to the Physics World weekly
2
00:00:11,919 --> 00:00:12,419
podcast,
3
00:00:12,884 --> 00:00:15,945
the last episode of 2025.
4
00:00:16,484 --> 00:00:18,664
And it features Pat Hanrahan,
5
00:00:19,125 --> 00:00:19,945
who studied
6
00:00:20,245 --> 00:00:22,585
nuclear engineering and biophysics
7
00:00:23,285 --> 00:00:25,704
before becoming a founding employee
8
00:00:26,164 --> 00:00:26,984
of Pixar
9
00:00:27,449 --> 00:00:28,669
Animation Studios
10
00:00:29,210 --> 00:00:30,589
in the nineteen eighties.
11
00:00:31,289 --> 00:00:34,429
That conversation is coming up after this message
12
00:00:34,489 --> 00:00:35,469
about a webinar
13
00:00:35,850 --> 00:00:36,909
that we're presenting
14
00:00:37,289 --> 00:00:39,229
with Leo Cancer Care.
15
00:00:39,905 --> 00:00:41,364
It's called unlocking
16
00:00:41,905 --> 00:00:45,284
novel radiation beams for cancer treatment
17
00:00:45,664 --> 00:00:47,844
with upright patient positioning,
18
00:00:48,384 --> 00:00:51,045
and it will feature a panel of leading
19
00:00:51,104 --> 00:00:51,604
clinicians
20
00:00:51,984 --> 00:00:52,804
and experts.
21
00:00:53,789 --> 00:00:56,770
They will discuss how the shift in patient
22
00:00:56,909 --> 00:00:57,409
positioning
23
00:00:57,869 --> 00:00:59,570
is enabling the exploration
24
00:01:00,109 --> 00:01:02,049
of new treatment geometries
25
00:01:02,829 --> 00:01:03,570
and supporting
26
00:01:03,949 --> 00:01:04,689
the development
27
00:01:05,150 --> 00:01:05,969
of advanced
28
00:01:06,564 --> 00:01:08,344
future cancer therapies.
29
00:01:09,284 --> 00:01:12,424
The live event is aimed at medical physicists,
30
00:01:13,284 --> 00:01:14,584
radiation oncologists,
31
00:01:15,685 --> 00:01:16,185
radiotherapy
32
00:01:16,805 --> 00:01:17,305
planners,
33
00:01:17,685 --> 00:01:18,584
and researchers
34
00:01:18,965 --> 00:01:19,864
in the field.
35
00:01:20,700 --> 00:01:24,780
Participants will learn about this paradigm shift in
36
00:01:24,780 --> 00:01:25,280
radiation
37
00:01:25,579 --> 00:01:26,079
therapy,
38
00:01:26,939 --> 00:01:28,239
gain an understanding
39
00:01:28,619 --> 00:01:30,560
of emerging and novel
40
00:01:31,019 --> 00:01:31,519
radiation
41
00:01:31,819 --> 00:01:33,040
beam modalities,
42
00:01:33,944 --> 00:01:34,765
and discover
43
00:01:35,064 --> 00:01:36,045
how foundational
44
00:01:36,825 --> 00:01:37,325
radiation
45
00:01:37,704 --> 00:01:38,204
physics
46
00:01:38,825 --> 00:01:39,325
translates
47
00:01:39,784 --> 00:01:41,805
into real world impact
48
00:01:42,105 --> 00:01:43,484
in cancer care.
49
00:01:44,265 --> 00:01:47,179
The webinar is on the February 17
50
00:01:47,640 --> 00:01:48,780
at 4PM
51
00:01:49,239 --> 00:01:49,739
GMT.
52
00:01:50,520 --> 00:01:52,380
That's 11AM
53
00:01:52,680 --> 00:01:54,299
Eastern Standard Time.
54
00:01:54,920 --> 00:01:57,420
You can register for this free event
55
00:01:57,719 --> 00:01:59,579
on the Physics World website.
56
00:02:02,165 --> 00:02:05,765
Earlier this year, Physics World's Margaret Harris attended
57
00:02:05,765 --> 00:02:06,584
the Heidelberg
58
00:02:07,045 --> 00:02:07,545
Laureate
59
00:02:07,844 --> 00:02:09,305
Forum in Germany.
60
00:02:10,004 --> 00:02:14,104
This international event brings together early career researchers
61
00:02:14,689 --> 00:02:17,349
with some winners of the most prestigious
62
00:02:17,650 --> 00:02:19,189
awards in mathematics
63
00:02:19,569 --> 00:02:21,030
and computer science.
64
00:02:21,650 --> 00:02:24,610
There, she spoke to a laureate who's also
65
00:02:24,610 --> 00:02:28,389
won three academy awards. Here's that conversation.
66
00:02:36,794 --> 00:02:39,114
So I'm talking to Pat Hanrahan, who won
67
00:02:39,114 --> 00:02:41,995
the Association for Computing Machineries AM Turing Award
68
00:02:41,995 --> 00:02:42,974
in 2019
69
00:02:43,310 --> 00:02:45,870
for contributions to three d computer graphics or
70
00:02:45,870 --> 00:02:46,370
CGI.
71
00:02:47,069 --> 00:02:48,509
So you're best known for that type of
72
00:02:48,509 --> 00:02:50,269
work, but you actually started out, I think,
73
00:02:50,269 --> 00:02:52,989
in nuclear engineering Yes. And then you got
74
00:02:52,989 --> 00:02:55,229
a PhD in biophysics. Yes. So how did
75
00:02:55,229 --> 00:02:56,935
you get into those subjects? Tell me about
76
00:02:56,935 --> 00:02:58,935
nuclear engineering first. Well, when I was in
77
00:02:58,935 --> 00:03:00,615
school, I I really felt I wanted to
78
00:03:00,615 --> 00:03:03,094
become a physicist. Anyway, I like math too,
79
00:03:03,094 --> 00:03:04,935
and I like science in general, but I
80
00:03:04,935 --> 00:03:07,415
think physics was the field that was most
81
00:03:07,415 --> 00:03:09,270
interesting to me. And so, yeah, when I
82
00:03:09,270 --> 00:03:11,030
went to the University of Wisconsin actually, it
83
00:03:11,030 --> 00:03:12,790
was it was it was doubly nice because
84
00:03:12,790 --> 00:03:14,469
they had this really nice, what they call,
85
00:03:14,469 --> 00:03:15,370
honors program,
86
00:03:15,750 --> 00:03:17,270
where you have a lot of freedom in
87
00:03:17,270 --> 00:03:19,270
what classes you took and all that. So,
88
00:03:19,270 --> 00:03:21,349
you know, not just take physics, but do
89
00:03:21,349 --> 00:03:24,064
other things as well. Take chemistry, biology, other
90
00:03:24,064 --> 00:03:24,564
sciences.
91
00:03:24,864 --> 00:03:26,625
So yeah. So I was always interested in
92
00:03:26,625 --> 00:03:28,704
physic. I love my intro physics course. I
93
00:03:28,704 --> 00:03:29,364
can still
94
00:03:29,664 --> 00:03:32,144
remember the lectures and, you know, going through
95
00:03:32,144 --> 00:03:33,044
it all. And,
96
00:03:33,344 --> 00:03:35,104
yeah. So I just loved physics, you know,
97
00:03:35,104 --> 00:03:36,884
all the way through the undergraduate
98
00:03:37,185 --> 00:03:38,084
physics, mechanics,
99
00:03:38,730 --> 00:03:39,230
electrodynamics,
100
00:03:40,169 --> 00:03:42,569
quantum mechanics. I just loved all those courses,
101
00:03:42,569 --> 00:03:43,550
my favorite courses.
102
00:03:43,849 --> 00:03:46,169
But why nuclear engineering specifically? Why not? Well,
103
00:03:46,169 --> 00:03:47,610
I think it was mostly because of this
104
00:03:47,610 --> 00:03:50,090
honors program. I was actually I actually and
105
00:03:50,090 --> 00:03:51,770
I when I went there, the chairman of
106
00:03:51,770 --> 00:03:52,349
the department,
107
00:03:52,650 --> 00:03:53,790
offered me a job,
108
00:03:54,665 --> 00:03:56,344
and he was the chair of the nuclear
109
00:03:56,344 --> 00:03:58,745
regulatory commission at the time. So I ended
110
00:03:58,745 --> 00:04:00,665
up, you know, in in this department. But,
111
00:04:00,665 --> 00:04:02,025
you know, I was he was working for
112
00:04:02,025 --> 00:04:03,784
the nuclear regulatory commission. This was, you know,
113
00:04:03,784 --> 00:04:05,165
like, in the late seventies.
114
00:04:05,705 --> 00:04:08,025
You know, nuclear power was having you know,
115
00:04:08,025 --> 00:04:10,120
people were pretty down on that. Right? And
116
00:04:10,120 --> 00:04:11,400
when I was reading about all this stuff,
117
00:04:11,400 --> 00:04:12,919
I got pretty down on it too because
118
00:04:12,919 --> 00:04:14,680
they weren't really addressing a lot of the
119
00:04:14,680 --> 00:04:16,439
currents. I think that's maybe changing a little
120
00:04:16,439 --> 00:04:17,740
bit, but I was definitely
121
00:04:18,279 --> 00:04:20,620
not that interested especially at towards the end.
122
00:04:21,160 --> 00:04:23,560
So I wanted to work on other areas
123
00:04:23,560 --> 00:04:24,220
of physics.
124
00:04:24,654 --> 00:04:26,415
I did a lot of theoretical physics. You
125
00:04:26,415 --> 00:04:28,095
know? Like, when I was a graduate student,
126
00:04:28,095 --> 00:04:29,935
I I had to take, like, I don't
127
00:04:29,935 --> 00:04:32,415
know, five courses in quantum mechanics, you know,
128
00:04:32,415 --> 00:04:34,574
quantum field theory. You know? I was really
129
00:04:34,574 --> 00:04:37,079
headed towards the theoretical physics thing, but I
130
00:04:37,079 --> 00:04:38,519
don't know what you I I the the
131
00:04:38,519 --> 00:04:41,000
low level of physics just wasn't really exciting
132
00:04:41,000 --> 00:04:42,439
me. So I I wanted to be a
133
00:04:42,439 --> 00:04:43,899
physicist, but I
134
00:04:44,360 --> 00:04:46,279
wanted to work on sort of higher level
135
00:04:46,279 --> 00:04:46,779
problems.
136
00:04:47,560 --> 00:04:50,439
So I I admired, like, Francis Crick, and
137
00:04:50,439 --> 00:04:51,660
he was, like, a theoretical
138
00:04:52,524 --> 00:04:53,904
biologist in some sense.
139
00:04:54,285 --> 00:04:55,725
And I thought, well, maybe I could become,
140
00:04:55,725 --> 00:04:57,264
like, a theoretical biologist.
141
00:04:57,964 --> 00:04:59,584
And then I met this person,
142
00:05:00,045 --> 00:05:01,964
and I was, he had was one of
143
00:05:01,964 --> 00:05:04,865
the founders of the computational neuroscience program.
144
00:05:05,319 --> 00:05:07,800
Or well, I'm sorry. Not computational neuroscience. That
145
00:05:07,800 --> 00:05:11,160
didn't exist. Neuroscience program. And, he was working
146
00:05:11,160 --> 00:05:13,319
on neurobiology of nematodes. He had worked for
147
00:05:13,319 --> 00:05:14,139
Sydney Brenner.
148
00:05:14,600 --> 00:05:16,279
And so I thought, well, I could I
149
00:05:16,279 --> 00:05:17,560
could do you know, he and they weren't
150
00:05:17,560 --> 00:05:19,319
being very quantitative, and I thought I could,
151
00:05:19,319 --> 00:05:21,555
like, apply what I know about physics to
152
00:05:21,555 --> 00:05:23,314
help them, you know, build up models of
153
00:05:23,314 --> 00:05:25,154
the nervous system. And I I started doing
154
00:05:25,154 --> 00:05:27,715
some simulations and things like that. Yeah. And
155
00:05:27,715 --> 00:05:28,995
is that about the time that you got
156
00:05:28,995 --> 00:05:31,314
first involved in Pixar, the project that became
157
00:05:31,314 --> 00:05:33,395
the first Toy Story film? No. That took
158
00:05:33,395 --> 00:05:34,675
a little while. I mean, you know, you
159
00:05:34,675 --> 00:05:37,170
have to realize computer graphics didn't exist then.
160
00:05:37,170 --> 00:05:39,410
It really didn't exist. I mean, you know,
161
00:05:39,410 --> 00:05:41,649
it just did not exist. My advice so
162
00:05:41,649 --> 00:05:43,170
it turned out it's sort of funny story.
163
00:05:43,170 --> 00:05:46,069
Sydney Brenner, you know, Nobel laureate in biology.
164
00:05:46,610 --> 00:05:48,129
There was he had two a bunch of
165
00:05:48,129 --> 00:05:51,165
colleagues and students, Francis Crick. There was another
166
00:05:51,165 --> 00:05:53,245
guy named John White and another guy named
167
00:05:53,245 --> 00:05:54,064
John Sulston.
168
00:05:54,444 --> 00:05:56,145
And John White is famous for
169
00:05:56,524 --> 00:05:57,024
determining,
170
00:05:57,645 --> 00:05:59,725
the structure of the nervous system of c
171
00:05:59,725 --> 00:06:01,245
l. So they all worked on C. Elegans,
172
00:06:01,245 --> 00:06:03,185
which is a a free living nematode.
173
00:06:03,680 --> 00:06:06,240
So, John White had, like, figured out the
174
00:06:06,240 --> 00:06:08,879
entire wiring grant by sectioning the words and
175
00:06:08,879 --> 00:06:11,360
worm and tracing all the neurons. And,
176
00:06:11,839 --> 00:06:14,019
John Solston had, worked on
177
00:06:14,560 --> 00:06:17,680
developmental biology, and he recorded, you know, how
178
00:06:17,680 --> 00:06:18,339
the cells
179
00:06:18,814 --> 00:06:20,574
divided, you know, from an egg all the
180
00:06:20,574 --> 00:06:22,435
way to a full grown one.
181
00:06:22,975 --> 00:06:24,975
And, anyways, the three of them all got
182
00:06:24,975 --> 00:06:25,954
really the same computers.
183
00:06:27,294 --> 00:06:27,794
And,
184
00:06:28,334 --> 00:06:30,495
and so, you know, my my adviser was
185
00:06:30,495 --> 00:06:32,014
in that group, and he was trying to
186
00:06:32,014 --> 00:06:33,555
convince me to work on computers.
187
00:06:34,129 --> 00:06:36,370
And I didn't have any background in computers.
188
00:06:36,370 --> 00:06:37,829
So I started simulating,
189
00:06:38,370 --> 00:06:39,589
this nervous system,
190
00:06:40,129 --> 00:06:42,209
and I had you, write these programs that
191
00:06:42,209 --> 00:06:43,829
simulated. I was, like, solving,
192
00:06:44,209 --> 00:06:47,169
like, basically cable equations for neurons because these
193
00:06:47,169 --> 00:06:49,189
neurons are, like, you know, electrical cables.
194
00:06:49,995 --> 00:06:52,074
But I needed to do visualization. I had
195
00:06:52,074 --> 00:06:53,435
to be able to see my result. Like,
196
00:06:53,435 --> 00:06:55,915
you just so my roommate introduced me to
197
00:06:55,915 --> 00:06:57,275
computer graphics. He said, oh, by the way,
198
00:06:57,275 --> 00:06:58,975
do you know about this graphic stuff?
199
00:06:59,355 --> 00:07:00,975
You know, I just sort of got enchanted
200
00:07:01,115 --> 00:07:02,235
by what I could do with it because
201
00:07:02,235 --> 00:07:03,835
I could take all my theories, like, whether
202
00:07:03,835 --> 00:07:06,580
it was simulation of neurons or whatever I
203
00:07:06,580 --> 00:07:08,819
was interested in and see it, you know,
204
00:07:08,819 --> 00:07:11,080
make it really concrete. And it just became
205
00:07:11,779 --> 00:07:13,460
even more fun. You know? Like, the the
206
00:07:13,460 --> 00:07:14,439
physics was suddenly
207
00:07:14,740 --> 00:07:16,500
infinitely more fun because I could see it.
208
00:07:16,500 --> 00:07:18,564
You know? Like, I could do animations,
209
00:07:18,944 --> 00:07:20,305
not much animation, but I could at least
210
00:07:20,305 --> 00:07:21,904
make pictures and so on. So I sort
211
00:07:21,904 --> 00:07:24,145
of got seduced by that idea, and it
212
00:07:24,145 --> 00:07:26,225
it was a wide open field. You know?
213
00:07:26,225 --> 00:07:28,305
Physics and, you know, biology was, you know,
214
00:07:28,305 --> 00:07:30,680
sort of slow paced. This was, like, completely
215
00:07:30,680 --> 00:07:32,759
wide open. So, like like like, the first
216
00:07:32,759 --> 00:07:34,280
week I worked on it, I, like, wrote
217
00:07:34,280 --> 00:07:36,280
a paper. I got accepted in the major
218
00:07:36,280 --> 00:07:37,800
conference. You know? Whereas in physics, you know,
219
00:07:37,800 --> 00:07:39,319
I'd been a grad student for a couple
220
00:07:39,319 --> 00:07:41,560
years, never written a paper. And so then
221
00:07:41,560 --> 00:07:43,080
I decided I had you go learn more
222
00:07:43,080 --> 00:07:44,220
about computer graphics,
223
00:07:44,794 --> 00:07:46,714
but, you know, because there was nobody there.
224
00:07:46,714 --> 00:07:48,954
There was only one or two places in
225
00:07:48,954 --> 00:07:50,794
the world that was doing computer graphics, and
226
00:07:50,794 --> 00:07:53,435
eventually, I ended up at Pixar a couple
227
00:07:53,435 --> 00:07:54,495
years later. Yeah.
228
00:07:55,435 --> 00:07:58,314
And a lot of your research on computer
229
00:07:58,314 --> 00:08:01,720
generated imagery focuses on getting the simulated world
230
00:08:01,720 --> 00:08:02,459
to display
231
00:08:03,079 --> 00:08:05,000
things in the way it's grounded in physics.
232
00:08:05,000 --> 00:08:06,519
Yeah. Yeah. You know, you literally wrote the
233
00:08:06,519 --> 00:08:09,479
book about physically based rendering. Yeah. What are
234
00:08:09,479 --> 00:08:11,639
the benefits of that approach, and what are
235
00:08:11,639 --> 00:08:14,055
some of the main challenges in it? That's
236
00:08:14,055 --> 00:08:15,035
a great question.
237
00:08:15,335 --> 00:08:17,095
You know, I love that question. I mean,
238
00:08:17,095 --> 00:08:18,694
you know, you know, we're we're creating a
239
00:08:18,694 --> 00:08:21,335
virtual world, and the world, you know, has
240
00:08:21,335 --> 00:08:23,495
physics in it. I mean, you know, and
241
00:08:23,495 --> 00:08:25,175
just imagine. I mean, what we did, we're
242
00:08:25,175 --> 00:08:26,694
just trying to create a world that, like,
243
00:08:26,694 --> 00:08:29,120
if somebody looked at it had all the
244
00:08:29,120 --> 00:08:31,439
phenomena that we think of in physics and
245
00:08:31,439 --> 00:08:33,759
even biology as well, but all these phenomena.
246
00:08:33,759 --> 00:08:35,160
So, you know, like, what are some of
247
00:08:35,160 --> 00:08:37,299
the found well, a cloud might move. Right?
248
00:08:37,759 --> 00:08:40,019
Wind will move the leaves of the tree.
249
00:08:40,079 --> 00:08:41,519
You know, you could have an object, a
250
00:08:41,519 --> 00:08:42,339
rock falling.
251
00:08:42,725 --> 00:08:44,725
You know, get out something a building kind
252
00:08:44,725 --> 00:08:47,045
of blowing apart. You could have, you know,
253
00:08:47,045 --> 00:08:50,485
cloth, you know, conforming to somebody's skin. You
254
00:08:50,485 --> 00:08:52,965
have the muscles moving the body. I mean,
255
00:08:52,965 --> 00:08:54,965
there's just physics everywhere in the world. Right?
256
00:08:54,965 --> 00:08:56,570
And so, you know, if we're trying to
257
00:08:56,570 --> 00:08:57,970
simulate all I mean, we're trying to make
258
00:08:57,970 --> 00:08:59,690
a picture of all this stuff. I mean,
259
00:08:59,690 --> 00:09:01,050
how else would you do it? I mean,
260
00:09:01,050 --> 00:09:02,889
you you have to in some sense, you
261
00:09:02,889 --> 00:09:04,889
have to somehow capture the physics of the
262
00:09:04,889 --> 00:09:06,490
everyday world. It's I I call it the
263
00:09:06,490 --> 00:09:08,970
everyday world because it's at the human scale.
264
00:09:08,970 --> 00:09:11,115
You know? It's what we see around us.
265
00:09:11,355 --> 00:09:13,534
And I mostly worked on,
266
00:09:13,914 --> 00:09:16,075
simulating the appearance of it, right, which is
267
00:09:16,075 --> 00:09:17,595
like, you know, you know, why does it
268
00:09:17,595 --> 00:09:19,274
look the way it does? Like, why how
269
00:09:19,274 --> 00:09:22,315
do, like, materials scatter light, and how does
270
00:09:22,315 --> 00:09:25,259
light get transported through the environment, and, you
271
00:09:25,259 --> 00:09:27,899
know, even how cameras simulate cameras and simulate
272
00:09:27,899 --> 00:09:29,820
light sources. So, yeah, so if you look
273
00:09:29,820 --> 00:09:31,420
at the what we actually ended up doing,
274
00:09:31,420 --> 00:09:33,179
if you look at that book, you know,
275
00:09:33,179 --> 00:09:34,080
we write down
276
00:09:34,460 --> 00:09:36,700
a bunch of equations that describe the scattering
277
00:09:36,700 --> 00:09:39,279
of light off surfaces and off infinite media,
278
00:09:39,465 --> 00:09:41,245
and then we just solve this
279
00:09:42,024 --> 00:09:44,264
solve for the lightning distribution in the world.
280
00:09:44,264 --> 00:09:46,345
You know? And it's a just a giant
281
00:09:46,345 --> 00:09:47,725
computational science problem.
282
00:09:48,105 --> 00:09:49,945
But the challenge is you have all these
283
00:09:49,945 --> 00:09:50,925
real world materials,
284
00:09:52,345 --> 00:09:53,884
which are both complex.
285
00:09:54,529 --> 00:09:57,009
You know, they're not just like ideal mirrors
286
00:09:57,009 --> 00:09:58,929
and, you know, one type of cars where
287
00:09:58,929 --> 00:10:00,690
I just, like, opened up an optics book
288
00:10:00,690 --> 00:10:02,290
and said, how much is in this book
289
00:10:02,290 --> 00:10:04,389
that would help us do this? Almost nothing.
290
00:10:04,690 --> 00:10:06,210
You know what I mean? Because, you know,
291
00:10:06,210 --> 00:10:07,889
how does light scatter off skin? How does
292
00:10:07,889 --> 00:10:10,085
light scatter off a rock? How does light
293
00:10:10,085 --> 00:10:12,565
scatter off a cloud? Just not detailed stuff
294
00:10:12,565 --> 00:10:14,644
in there. So we'd often have to come
295
00:10:14,644 --> 00:10:15,625
up with, you know,
296
00:10:16,004 --> 00:10:16,504
approximations
297
00:10:16,884 --> 00:10:17,784
and models
298
00:10:18,164 --> 00:10:20,325
for all these things. And I worked a
299
00:10:20,325 --> 00:10:22,164
lot. Like, I I worked on my papers
300
00:10:22,164 --> 00:10:23,464
as I'm modeling skin.
301
00:10:24,419 --> 00:10:26,339
Hair. How does light scatter from hair? How
302
00:10:26,339 --> 00:10:28,899
does light scatter from cloth? Stuff like that.
303
00:10:28,899 --> 00:10:30,339
But, anyways, when it all is said and
304
00:10:30,339 --> 00:10:31,940
done, you get, you know, like, a billion
305
00:10:31,940 --> 00:10:32,440
objects
306
00:10:32,820 --> 00:10:34,679
each with really complex materials,
307
00:10:35,059 --> 00:10:37,375
and then you have to solve for the
308
00:10:37,375 --> 00:10:38,115
lighting distribution,
309
00:10:38,574 --> 00:10:41,454
which is really complicated because the objects have
310
00:10:41,454 --> 00:10:44,274
complicated shape. So we use Monte Carlo techniques.
311
00:10:44,495 --> 00:10:46,174
You know? In fact, just like what they
312
00:10:46,174 --> 00:10:48,815
did up during the war to simulate the
313
00:10:48,815 --> 00:10:49,794
neutron transport,
314
00:10:50,110 --> 00:10:51,870
And that's actually was the connection back to
315
00:10:51,870 --> 00:10:54,509
my work in nuclear engineering. I had learned
316
00:10:54,509 --> 00:10:57,230
about transport theory and Monte Carlo techniques, you
317
00:10:57,230 --> 00:10:59,309
know, in school. And so, you know, I
318
00:10:59,309 --> 00:11:01,009
I sort of also just sort of gravitated
319
00:11:01,070 --> 00:11:02,590
towards those problems because those are the ones
320
00:11:02,590 --> 00:11:03,570
I knew to sell.
321
00:11:03,904 --> 00:11:06,144
So you actually won, one of your three
322
00:11:06,144 --> 00:11:08,544
Academy Awards, Oscars, in other words Yeah. Is
323
00:11:08,544 --> 00:11:11,585
specifically for developing that software that simulates the
324
00:11:11,585 --> 00:11:13,665
subsurface scattering of light Yes. Yeah. In in
325
00:11:13,665 --> 00:11:16,304
materials that are translucent. Yeah. Yeah. Why is
326
00:11:16,304 --> 00:11:18,144
that so difficult, and why is it so
327
00:11:18,144 --> 00:11:18,644
important
328
00:11:19,340 --> 00:11:21,980
for film and computer graphics? Yeah. Well, okay.
329
00:11:21,980 --> 00:11:23,980
Why is it important? Well, I'm looking at
330
00:11:23,980 --> 00:11:26,080
you, and your skin looks really healthy.
331
00:11:27,019 --> 00:11:28,300
I mean, you know, I mean, you know,
332
00:11:28,300 --> 00:11:30,940
in real people, skin is really important. Right?
333
00:11:30,940 --> 00:11:33,340
It has to look right. And until then,
334
00:11:33,340 --> 00:11:34,399
it looked like plastic.
335
00:11:34,715 --> 00:11:36,154
You know, it looked like, yeah, just were
336
00:11:36,154 --> 00:11:38,235
completely covered in makeup. You know? You know,
337
00:11:38,235 --> 00:11:40,554
makeup and things like that, you know, almost
338
00:11:40,554 --> 00:11:41,915
like pink. You know what I mean? They're
339
00:11:41,915 --> 00:11:43,915
just very matte and, you know, very I
340
00:11:43,915 --> 00:11:45,995
mean, you know, that's why they're matte. They're
341
00:11:45,995 --> 00:11:47,740
not they did not you know, don't have
342
00:11:47,740 --> 00:11:49,180
all the complexity of real skin, but real
343
00:11:49,180 --> 00:11:51,820
skin has oil. It has texture to it.
344
00:11:51,820 --> 00:11:52,480
It has
345
00:11:52,940 --> 00:11:54,700
a bunch of properties that are very different
346
00:11:54,700 --> 00:11:57,519
than these idealized plastic like, paint like material.
347
00:11:57,820 --> 00:11:59,899
So so, obviously, for the movies, they want
348
00:11:59,899 --> 00:12:02,559
people to look, you know, natural and appealing.
349
00:12:03,144 --> 00:12:04,664
So that was one that was sort of
350
00:12:04,664 --> 00:12:06,985
a motivation for it. What was interesting, the
351
00:12:06,985 --> 00:12:08,365
discovery was that
352
00:12:08,825 --> 00:12:09,565
most materials
353
00:12:10,105 --> 00:12:13,225
are actually somewhat translucent, and subsurface scatter is
354
00:12:13,225 --> 00:12:15,929
actually more common than just normal surface scatter.
355
00:12:15,929 --> 00:12:17,690
So what happened in surface scatter, which you
356
00:12:17,690 --> 00:12:19,450
learn about in physics, you know, light ray
357
00:12:19,450 --> 00:12:21,769
hits the surface and just gets reflected off
358
00:12:21,769 --> 00:12:24,170
the interface between, say, the air and the
359
00:12:24,170 --> 00:12:26,809
the material itself. Yeah. Yeah. But in most
360
00:12:26,809 --> 00:12:28,809
materials, they're actually dielectric. And by the way,
361
00:12:28,809 --> 00:12:30,410
that's what happens in the metal. You know?
362
00:12:30,410 --> 00:12:32,334
Like, light can't go into a metal because
363
00:12:32,334 --> 00:12:32,995
it's conductive,
364
00:12:33,534 --> 00:12:34,914
so it has to get reflected.
365
00:12:35,294 --> 00:12:38,095
Okay? That's why metals are shiny. But most
366
00:12:38,095 --> 00:12:40,334
materials are dielectrics. You know? They have an
367
00:12:40,334 --> 00:12:42,815
index of refraction. And so most light goes
368
00:12:42,815 --> 00:12:44,815
into the materials. It doesn't get reflected. You
369
00:12:44,815 --> 00:12:46,589
know? You just look at it physically. And
370
00:12:46,589 --> 00:12:48,829
why are the dialectics? Well, they're rocks. You
371
00:12:48,829 --> 00:12:51,069
know? They're based on water. Like, we have
372
00:12:51,069 --> 00:12:53,009
a lot of water in our human body.
373
00:12:53,069 --> 00:12:54,929
You know? They're made of things like fibers,
374
00:12:55,149 --> 00:12:56,449
anything like biological.
375
00:12:57,149 --> 00:12:59,534
And most minerals are not metals. They're,
376
00:12:59,934 --> 00:13:02,095
you know, they're dielectric. So almost everything around
377
00:13:02,095 --> 00:13:03,634
us is a dielectric,
378
00:13:04,495 --> 00:13:06,815
and what happens is the light goes into
379
00:13:06,815 --> 00:13:09,134
the material and then it scatters around a
380
00:13:09,134 --> 00:13:11,054
little bit, and then it comes up. It
381
00:13:11,054 --> 00:13:13,294
looks like it's reflecting off it. Let's actually
382
00:13:13,294 --> 00:13:15,240
go on and scatter around, come out, And
383
00:13:15,240 --> 00:13:17,019
that's called subsurface reflection.
384
00:13:17,639 --> 00:13:19,320
You know, it's been studied, but it's never
385
00:13:19,320 --> 00:13:21,159
really been studied very much in natural no
386
00:13:21,159 --> 00:13:23,799
normal everyday materials like skin or whatever. So
387
00:13:23,799 --> 00:13:25,899
we we develop you know, like a physicist,
388
00:13:26,200 --> 00:13:28,600
you know, just, you know, we read models
389
00:13:28,600 --> 00:13:30,415
about how you might do that. I used
390
00:13:30,415 --> 00:13:33,075
to read books like people like in atmospheric
391
00:13:33,375 --> 00:13:36,254
sciences study that, meteorology study that, like, if
392
00:13:36,254 --> 00:13:38,035
you observe a planetary atmosphere
393
00:13:38,575 --> 00:13:40,335
from a distance, like, it's like you have
394
00:13:40,335 --> 00:13:43,554
this layer of gas around the planet,
395
00:13:43,879 --> 00:13:45,559
and so it's gonna be sort of like
396
00:13:45,559 --> 00:13:47,159
that. The light goes into the atmosphere and
397
00:13:47,159 --> 00:13:48,839
gets scattered. So there, you know, there's things
398
00:13:48,839 --> 00:13:50,039
known about it, but they were all like
399
00:13:50,039 --> 00:13:52,139
things like that, like, you know, planetary atmosphere.
400
00:13:52,519 --> 00:13:54,360
So we yeah. We we just took a
401
00:13:54,360 --> 00:13:56,200
lot of those ideas that they had developed
402
00:13:56,200 --> 00:13:58,085
and and, you know, sort of, you know,
403
00:13:58,085 --> 00:14:00,264
mod used them to model things like skin.
404
00:14:00,644 --> 00:14:02,165
And in the same thing with hair, what
405
00:14:02,165 --> 00:14:03,625
happens is hair is mostly,
406
00:14:04,245 --> 00:14:06,325
translucent as well. It's like a little optical
407
00:14:06,325 --> 00:14:08,165
fiber. And so light goes into it, bounces
408
00:14:08,165 --> 00:14:10,085
around, and comes out. So these just lead
409
00:14:10,085 --> 00:14:12,264
to very different appearances than
410
00:14:12,629 --> 00:14:15,110
metals and things like that. So yeah. So
411
00:14:15,110 --> 00:14:16,230
I I did a lot of work on
412
00:14:16,230 --> 00:14:18,809
on, you know, just modeling different materials. And
413
00:14:18,870 --> 00:14:21,210
the subsurface scattering works, I mean, since everything
414
00:14:21,429 --> 00:14:23,750
looked like plastic, as soon as we wrote
415
00:14:23,750 --> 00:14:26,149
the paper, I mean, every studio in the
416
00:14:26,149 --> 00:14:27,210
world immediately,
417
00:14:28,414 --> 00:14:30,975
adopted them. And we we published the source
418
00:14:30,975 --> 00:14:34,095
code, and it was relatively simple model, so
419
00:14:34,095 --> 00:14:35,855
they could just, like, take it and put
420
00:14:35,855 --> 00:14:36,995
it into their system.
421
00:14:37,454 --> 00:14:39,294
And so yeah. So they like, within a
422
00:14:39,294 --> 00:14:41,389
year or two, everybody was using it sort
423
00:14:41,389 --> 00:14:43,230
of very quickly. And but, you know, it's
424
00:14:43,309 --> 00:14:44,669
it was I think it was the first
425
00:14:44,669 --> 00:14:47,870
time an academic paper had ever gotten an
426
00:14:47,870 --> 00:14:48,769
Academy Award.
427
00:14:50,829 --> 00:14:52,769
What would you say is the most difficult
428
00:14:53,115 --> 00:14:55,995
CGI problem either you've solved or you've seen
429
00:14:55,995 --> 00:14:58,875
someone else solve by some building on your
430
00:14:58,875 --> 00:14:59,375
work?
431
00:14:59,914 --> 00:15:01,455
Well, I I I think the
432
00:15:01,914 --> 00:15:04,095
I think the hardest thing to solve
433
00:15:04,794 --> 00:15:06,554
was not just, like, how to make, you
434
00:15:06,554 --> 00:15:08,940
know, skin look right. I mean, that's really
435
00:15:08,940 --> 00:15:10,160
insane, hard problem.
436
00:15:10,779 --> 00:15:12,639
It's how to build a system
437
00:15:13,259 --> 00:15:14,639
that would include everything.
438
00:15:15,179 --> 00:15:16,620
You know what I mean? Like, it's like
439
00:15:16,620 --> 00:15:18,620
this it's like sort of grand unified theory.
440
00:15:18,620 --> 00:15:21,519
Right? How can I build one rendering system
441
00:15:21,924 --> 00:15:23,625
that would handle any shape
442
00:15:23,925 --> 00:15:27,144
and any material and any light source,
443
00:15:27,605 --> 00:15:28,985
you know, and any kind of,
444
00:15:29,365 --> 00:15:32,165
you know, participating media and and just build
445
00:15:32,165 --> 00:15:34,325
a system that could do it all? And
446
00:15:34,325 --> 00:15:35,605
the problem is you if you if you
447
00:15:35,605 --> 00:15:37,259
can't do a bunch of stuff, like, say
448
00:15:37,259 --> 00:15:38,720
you couldn't do the atmosphere.
449
00:15:39,179 --> 00:15:40,700
I mean, it's just not gonna be useful.
450
00:15:40,700 --> 00:15:42,620
Right? And if you can't do people, it's
451
00:15:42,620 --> 00:15:44,559
not gonna be useful. You can't do plants.
452
00:15:44,860 --> 00:15:46,460
You know, like plants were a huge problem
453
00:15:46,460 --> 00:15:48,300
because there's so many leaves and they have
454
00:15:48,300 --> 00:15:50,460
such complicated scattering, but you have billions of
455
00:15:50,460 --> 00:15:52,559
them. You know? Billions of leaves and
456
00:15:53,035 --> 00:15:55,274
dandelions and stuff. So, you know, just having
457
00:15:55,274 --> 00:15:57,514
a system that could handle all this stuff
458
00:15:57,514 --> 00:15:59,615
and scale to the complexities,
459
00:16:00,475 --> 00:16:02,955
I mean, it's it's mostly an engineering problem,
460
00:16:02,955 --> 00:16:04,495
but it was it was a conceptualization
461
00:16:05,195 --> 00:16:07,610
problem. Like, how do you break down,
462
00:16:08,070 --> 00:16:09,929
you know, the world in some sense
463
00:16:10,389 --> 00:16:12,870
into all these what are some you know,
464
00:16:12,870 --> 00:16:14,070
you know, we sort of use, like, object
465
00:16:14,070 --> 00:16:16,070
oriented, but what are all the objects you
466
00:16:16,070 --> 00:16:18,389
need, and how are they all interrelated? How
467
00:16:18,389 --> 00:16:20,309
do they all work together? And then so
468
00:16:20,309 --> 00:16:22,605
that you can, like, plug in something new.
469
00:16:22,605 --> 00:16:24,205
Like, when we had that skin model, like,
470
00:16:24,205 --> 00:16:25,565
you just plug it in. You know, the
471
00:16:25,565 --> 00:16:27,804
system had the design. So that was actually
472
00:16:27,804 --> 00:16:29,485
my first academy where we had this system
473
00:16:29,485 --> 00:16:30,705
we called Reyes,
474
00:16:31,165 --> 00:16:33,585
R E Y E S, for the California.
475
00:16:33,725 --> 00:16:35,825
There's this park called Point Reyes.
476
00:16:36,740 --> 00:16:38,339
It's a very beautiful place. It was near
477
00:16:38,339 --> 00:16:41,779
Lucasfilm, so everybody knew about it. But it
478
00:16:41,779 --> 00:16:44,259
it was the acronym for renders everything you
479
00:16:44,259 --> 00:16:46,419
ever saw. So, you know, the first thing
480
00:16:46,500 --> 00:16:48,259
what I mostly worked on at Pixar was
481
00:16:48,259 --> 00:16:50,024
just how to build this system. You know?
482
00:16:50,105 --> 00:16:52,105
How how to build a system that renders
483
00:16:52,105 --> 00:16:54,264
everything you ever saw. And, you know, it's
484
00:16:54,425 --> 00:16:55,404
it was very,
485
00:16:56,024 --> 00:16:58,264
very challenging and non trivial to think of
486
00:16:58,264 --> 00:17:00,264
how to design this thing so it would
487
00:17:00,264 --> 00:17:01,705
work. And it's still being used. I mean,
488
00:17:01,705 --> 00:17:03,144
I did this many, many years ago. But,
489
00:17:03,144 --> 00:17:04,904
you know, once you get the basic framework
490
00:17:04,904 --> 00:17:05,549
in place,
491
00:17:06,110 --> 00:17:07,869
you know, you can just keep enhancing it
492
00:17:07,869 --> 00:17:09,630
because you can add stuff to it and
493
00:17:09,630 --> 00:17:11,070
they it just sort of you know, you
494
00:17:11,070 --> 00:17:13,309
have the right design when something new comes
495
00:17:13,309 --> 00:17:15,070
along and just sort of clicks into place.
496
00:17:15,070 --> 00:17:16,670
You know what I mean? So that that
497
00:17:16,750 --> 00:17:18,190
I think that was the hardest part. I
498
00:17:18,190 --> 00:17:20,194
mean, you know, maybe not quite as mathematical,
499
00:17:20,335 --> 00:17:22,515
but just how to conceive of this thing
500
00:17:22,654 --> 00:17:24,595
in a way that, you know, was both
501
00:17:24,654 --> 00:17:27,214
conceptually elegant, but also worked. You know what
502
00:17:27,214 --> 00:17:29,315
I mean? It could be efficient and and
503
00:17:29,454 --> 00:17:30,815
and so on. So that I think that
504
00:17:30,815 --> 00:17:32,759
was the hardest problem. Yeah. What would you
505
00:17:32,759 --> 00:17:35,559
say then is the the most difficult unsolved
506
00:17:35,559 --> 00:17:37,400
problem? What what can't you do now that
507
00:17:37,400 --> 00:17:38,680
you really like to be able to do
508
00:17:38,680 --> 00:17:39,740
in computer graphics?
509
00:17:40,519 --> 00:17:41,880
Well, I think that I have a couple
510
00:17:41,880 --> 00:17:43,400
of opinions. I mean, you know, on the
511
00:17:43,400 --> 00:17:45,340
on the so we would call this photorealistic
512
00:17:45,720 --> 00:17:47,545
rendering and so on. I mean, I don't
513
00:17:47,625 --> 00:17:49,144
I mean, pretty much, you know, we've been
514
00:17:49,144 --> 00:17:50,825
working on it so long. I mean, I
515
00:17:50,825 --> 00:17:52,424
don't know if you go to movies. Some
516
00:17:52,424 --> 00:17:53,705
of the people do, but, you know, the
517
00:17:53,705 --> 00:17:55,865
movies are pretty convincing at this point. You
518
00:17:55,865 --> 00:17:57,464
know? And even like Toy Story was the
519
00:17:57,464 --> 00:17:59,224
first one that we did. It was a
520
00:17:59,224 --> 00:18:00,365
cartoon movie.
521
00:18:00,664 --> 00:18:02,205
Now, you know, you see movies
522
00:18:02,920 --> 00:18:04,380
routinely that just
523
00:18:04,920 --> 00:18:05,900
is completely indistinguishable
524
00:18:06,519 --> 00:18:07,019
from,
525
00:18:07,320 --> 00:18:09,640
you know, from things like Avatar and, you
526
00:18:09,640 --> 00:18:12,119
know, all these movie. So it's pretty hard.
527
00:18:12,119 --> 00:18:13,240
I mean, I don't know if I can
528
00:18:13,240 --> 00:18:14,059
think of really,
529
00:18:14,599 --> 00:18:17,559
one thing. I find natural scenes, outdoor scenes,
530
00:18:17,559 --> 00:18:19,305
sort of still somewhat unsatisfying.
531
00:18:20,085 --> 00:18:21,765
I mean, remember, they still fake a lot
532
00:18:21,765 --> 00:18:23,465
of stuff. They don't model everything
533
00:18:24,085 --> 00:18:25,924
and so on, so they can avoid, you
534
00:18:25,924 --> 00:18:27,865
know, showing you things that they can't do.
535
00:18:27,924 --> 00:18:30,005
But I find, you know, really if I
536
00:18:30,005 --> 00:18:31,865
take a walk in the forest or something,
537
00:18:32,119 --> 00:18:33,960
you know, there's something about it that I
538
00:18:33,960 --> 00:18:35,880
don't I don't see in the computer, but
539
00:18:35,880 --> 00:18:37,240
I can't quite pin it down. That's the
540
00:18:37,240 --> 00:18:39,160
other problem. What? You're you're looking at this
541
00:18:39,160 --> 00:18:40,940
stuff. Like, you look I'm looking at skin.
542
00:18:41,240 --> 00:18:43,240
Like, what are we not getting about it?
543
00:18:43,240 --> 00:18:45,480
You know? There's something underlying the phenomena of
544
00:18:45,480 --> 00:18:47,734
the scattering. So I don't quite know exactly
545
00:18:47,795 --> 00:18:48,615
for these
546
00:18:48,994 --> 00:18:49,494
natural,
547
00:18:50,195 --> 00:18:53,234
environments what we're getting wrong. But, anyway, I
548
00:18:53,234 --> 00:18:54,994
think that's one. But I actually think, you
549
00:18:54,994 --> 00:18:55,494
know,
550
00:18:55,955 --> 00:18:58,615
a more interesting thing is not just photorealistic
551
00:18:59,154 --> 00:19:01,029
effects, but, you know, how to make it
552
00:19:01,029 --> 00:19:02,710
easy. You know, like, I gave a talk
553
00:19:02,710 --> 00:19:05,589
once about starting with Leonardo, how, you know,
554
00:19:05,589 --> 00:19:08,970
we sort of reproduced what Leonardo had discovered,
555
00:19:09,029 --> 00:19:11,349
right, like perspective and shading and all that
556
00:19:11,349 --> 00:19:13,289
stuff, and we just computerized it.
557
00:19:13,865 --> 00:19:15,465
But then the event art sort of went
558
00:19:15,465 --> 00:19:17,785
in lots of different directions. It wasn't just
559
00:19:17,785 --> 00:19:19,865
about realism. You know? So, like, maybe how
560
00:19:19,865 --> 00:19:21,485
to make other kinds of pictures,
561
00:19:22,105 --> 00:19:24,424
like, either more abstract kind of pictures like
562
00:19:24,424 --> 00:19:26,930
an artist might do, or more functional kind
563
00:19:26,930 --> 00:19:28,390
of pictures, like, in visualizations
564
00:19:28,690 --> 00:19:30,529
of things. You know, just how you know,
565
00:19:30,529 --> 00:19:32,609
because there you're not just simulating the physics.
566
00:19:32,609 --> 00:19:34,130
You have to ask but you still you
567
00:19:34,130 --> 00:19:36,049
know, it might be some formal technique. Let's
568
00:19:36,049 --> 00:19:37,589
say we had some theory of aesthetics
569
00:19:38,265 --> 00:19:40,345
or something like that. Maybe we could formalize
570
00:19:40,345 --> 00:19:42,025
that and have a computer do that kind
571
00:19:42,025 --> 00:19:43,384
of stuff. Yeah. You see some of that
572
00:19:43,384 --> 00:19:45,785
generated by AI right now. But, yeah, just,
573
00:19:45,785 --> 00:19:47,465
you know, other there are different kinds of
574
00:19:47,465 --> 00:19:47,965
picture.
575
00:19:48,585 --> 00:19:50,505
And I think we could make a lot
576
00:19:50,505 --> 00:19:52,904
of other interesting pictures besides just sort of
577
00:19:52,904 --> 00:19:53,404
photographic,
578
00:19:53,750 --> 00:19:55,529
You know, sort of a classic thing, like,
579
00:19:55,830 --> 00:19:57,509
first we once we had a camera, like,
580
00:19:57,509 --> 00:19:58,789
you know, what do we why do we
581
00:19:58,789 --> 00:20:00,710
need to you know, we don't want it
582
00:20:00,710 --> 00:20:02,950
to we have cameras to take realistic pictures.
583
00:20:02,950 --> 00:20:05,109
We don't need, you know, artists to draw
584
00:20:05,109 --> 00:20:07,634
perfectly realistic pictures. Yeah. So you're still looking
585
00:20:07,634 --> 00:20:10,115
for the Pixar version of Picasso. Yeah. Yeah.
586
00:20:10,115 --> 00:20:12,455
Exactly. Yeah. Yeah. Yeah. Exactly. Exactly.
587
00:20:13,075 --> 00:20:14,994
So final question. How if you could just
588
00:20:14,994 --> 00:20:16,275
sum it up, how do you think your
589
00:20:16,275 --> 00:20:18,835
training in physics has helped you in your
590
00:20:18,835 --> 00:20:19,335
career?
591
00:20:19,759 --> 00:20:21,359
Well, I certainly learned a lot of phys
592
00:20:21,519 --> 00:20:23,619
you know, physical phenomena, you know,
593
00:20:24,240 --> 00:20:25,059
like transport,
594
00:20:25,519 --> 00:20:26,339
light optics,
595
00:20:26,960 --> 00:20:28,980
and mechanics, classical mechanics.
596
00:20:29,440 --> 00:20:32,085
You know, lots of, particular things I learned.
597
00:20:32,325 --> 00:20:33,924
But the best thing I learned was how
598
00:20:33,924 --> 00:20:34,825
to solve problems.
599
00:20:36,085 --> 00:20:37,684
I mean, you know, I mean, you just
600
00:20:37,845 --> 00:20:39,365
you know, the problems they gave you to
601
00:20:39,365 --> 00:20:40,804
solving, you just had to solve. And there
602
00:20:40,804 --> 00:20:42,404
was a certain way of solving them. You
603
00:20:42,404 --> 00:20:44,404
know? How do you like, you notice I
604
00:20:44,404 --> 00:20:46,484
use the word modeling and simulation a lot.
605
00:20:46,484 --> 00:20:48,849
You know? Like, you have physical models and,
606
00:20:49,230 --> 00:20:51,230
you know, models are useful, but they're not
607
00:20:51,230 --> 00:20:52,509
the whole thing. You have to, you know,
608
00:20:52,509 --> 00:20:54,049
you have to pick the right model
609
00:20:54,509 --> 00:20:56,269
for, you know, you know, like, start out
610
00:20:56,269 --> 00:20:57,649
with really phenomenological
611
00:20:57,950 --> 00:20:59,470
like models, but they can still be super
612
00:20:59,470 --> 00:21:02,049
useful when you also want more fundamental models.
613
00:21:02,174 --> 00:21:03,855
But there's many models, and you have to
614
00:21:03,855 --> 00:21:05,695
pick the right one. And then and so
615
00:21:05,695 --> 00:21:07,394
you just have to pick the right one.
616
00:21:07,695 --> 00:21:08,975
And, you know, a lot of people, they
617
00:21:08,975 --> 00:21:10,735
they just they don't know how to pick
618
00:21:10,735 --> 00:21:12,414
the right model. You know what I mean?
619
00:21:12,414 --> 00:21:13,775
And then you and then you have to
620
00:21:13,775 --> 00:21:14,755
crank through whatever
621
00:21:15,089 --> 00:21:17,569
it takes to sort of make it solvable.
622
00:21:17,569 --> 00:21:19,169
You know what I mean? Like, you know,
623
00:21:19,169 --> 00:21:21,169
it might involve spherical harmonics or it might
624
00:21:21,169 --> 00:21:22,690
involve a bunch of stuff. And then some
625
00:21:22,690 --> 00:21:24,690
of that stuff sort of just difficult. Right?
626
00:21:24,690 --> 00:21:26,769
You just gotta have the the you have
627
00:21:26,769 --> 00:21:29,410
to be, like, powerful problem solver to actually
628
00:21:29,410 --> 00:21:31,434
turn that into, like, something that works. So
629
00:21:31,434 --> 00:21:32,875
I think that's what I learned in physics.
630
00:21:32,875 --> 00:21:34,575
And, you know, that skill,
631
00:21:35,674 --> 00:21:37,674
it was I just think that's why physicists
632
00:21:37,674 --> 00:21:39,434
are so successful when they go to other
633
00:21:39,434 --> 00:21:41,434
fields, is they just have this magical ability
634
00:21:41,434 --> 00:21:43,615
to solve problems in other fields. Yeah.
635
00:21:44,179 --> 00:21:45,700
That kind of fun. Thank you very much.
636
00:21:45,700 --> 00:21:46,679
Yeah. Thank you.
637
00:21:53,859 --> 00:21:57,220
That was Margaret Harris in conversation with Pat
638
00:21:57,220 --> 00:21:57,720
Hanrahan,
639
00:21:58,515 --> 00:22:00,375
who has won three Oscars
640
00:22:00,755 --> 00:22:02,775
in addition to the ACM's
641
00:22:03,154 --> 00:22:04,855
AM Turing Award.
642
00:22:05,154 --> 00:22:07,414
Thanks to both of them for a fascinating
643
00:22:07,634 --> 00:22:08,134
conversation.
644
00:22:08,994 --> 00:22:11,875
And thank you for listening to the Physics
645
00:22:11,875 --> 00:22:13,335
World weekly podcast
646
00:22:13,909 --> 00:22:15,529
in 2025.
647
00:22:16,150 --> 00:22:17,450
I'm Hamish Johnston,
648
00:22:17,750 --> 00:22:20,409
and our producer is Fred Iles.
649
00:22:20,950 --> 00:22:22,089
We'll be back again
650
00:22:22,390 --> 00:22:24,089
in 2026.
