0:00:00.902,0:00:06.856
Doc Edgerton inspired us[br]with awe and curiosity

0:00:06.880,0:00:12.118
with this photo of a bullet[br]piercing through an apple,

0:00:12.142,0:00:15.994
and exposure just a millionth of a second.

0:00:17.661,0:00:24.550
But now, 50 years later,[br]we can go a million times faster

0:00:24.574,0:00:29.336
and see the world[br]not at a million or a billion,

0:00:29.360,0:00:32.413
but one trillion frames per second.

0:00:33.865,0:00:37.529
I present to you[br]a new type of photography,

0:00:37.553,0:00:39.588
femto-photography,

0:00:39.612,0:00:44.188
a new imaging technique so fast

0:00:44.212,0:00:48.484
that it can create slow motion[br]videos of light in motion.

0:00:49.413,0:00:54.272
And with that, we can create cameras[br]that can look around corners,

0:00:54.296,0:00:56.269
beyond line of sight,

0:00:56.293,0:01:01.149
or see inside our body without an x-ray,

0:01:01.173,0:01:04.980
and really challenge[br]what we mean by a camera.

0:01:06.440,0:01:09.574
Now if I take a laser pointer[br]and turn it on and off

0:01:09.598,0:01:12.401
in one trillionth of a second --

0:01:12.425,0:01:15.212
which is several femtoseconds --

0:01:15.236,0:01:19.233
I'll create a packet of photons[br]barely a millimeter wide.

0:01:20.360,0:01:22.769
And that packet of photons, that bullet,

0:01:22.793,0:01:24.719
will travel at the speed of light,

0:01:24.743,0:01:28.701
and again, a million times faster[br]than an ordinary bullet.

0:01:29.622,0:01:34.133
Now, if you take that bullet[br]and take this packet of photons

0:01:34.157,0:01:37.489
and fire into this bottle,

0:01:37.513,0:01:41.084
how will those photons[br]shatter into this bottle?

0:01:41.864,0:01:44.452
How does light look in slow motion?

0:01:44.552,0:01:47.552
[Light in Slow Motion ...[br]10 Billion x Slow]

0:02:06.041,0:02:07.588
Now, the whole event --

0:02:07.612,0:02:14.599
(Applause)

0:02:14.623,0:02:18.124
Now remember, the whole event[br]is effectively taking place

0:02:18.148,0:02:19.811
in less than a nanosecond --

0:02:19.835,0:02:22.236
that's how much time[br]it takes for light to travel.

0:02:22.260,0:02:26.907
But I'm slowing down in this video[br]by a factor of 10 billion,

0:02:26.931,0:02:29.479
so you can see the light in motion.

0:02:29.503,0:02:30.573
(Laughter)

0:02:30.597,0:02:32.884
But Coca-Cola did not[br]sponsor this research.

0:02:32.908,0:02:35.010
(Laughter)

0:02:35.034,0:02:37.057
Now, there's a lot going on in this movie,

0:02:37.081,0:02:39.701
so let me break this down[br]and show you what's going on.

0:02:39.725,0:02:42.659
So the pulse enters[br]the bottle, our bullet,

0:02:42.683,0:02:45.221
with a packet of photons[br]that start traveling through

0:02:45.245,0:02:47.058
and that start scattering inside.

0:02:47.082,0:02:49.289
Some of the light leaks,[br]goes on the table,

0:02:49.313,0:02:51.925
and you start seeing[br]these ripples of waves.

0:02:51.949,0:02:54.957
Many of the photons[br]eventually reach the cap

0:02:54.981,0:02:57.871
and then they explode[br]in various directions.

0:02:57.895,0:02:59.783
As you can see, there's a bubble of air

0:02:59.807,0:03:01.449
and it's bouncing around inside.

0:03:01.473,0:03:03.923
Meanwhile, the ripples[br]are traveling on the table,

0:03:03.947,0:03:05.947
and because of the reflections at the top,

0:03:05.971,0:03:09.718
you see at the back of the bottle,[br]after several frames,

0:03:09.742,0:03:11.441
the reflections are focused.

0:03:13.210,0:03:18.599
Now, if you take an ordinary bullet

0:03:18.623,0:03:21.623
and let it go the same distance[br]and slow down the video --

0:03:21.647,0:03:23.868
again, by a factor of 10 billion --

0:03:23.892,0:03:28.285
do you know how long you'll have to sit[br]here to watch that movie?

0:03:28.309,0:03:29.897
(Laughter)

0:03:29.921,0:03:33.430
A day, a week? Actually, a whole year.

0:03:34.309,0:03:36.566
It'll be a very boring movie --

0:03:36.590,0:03:37.981
(Laughter)

0:03:38.005,0:03:41.377
of a slow, ordinary bullet in motion.

0:03:42.700,0:03:46.057
And what about some[br]still-life photography?

0:03:53.112,0:03:58.098
You can watch the ripples,[br]again, washing over the table,

0:03:58.122,0:04:00.780
the tomato and the wall in the back.

0:04:00.804,0:04:03.594
It's like throwing a stone[br]in a pond of water.

0:04:07.403,0:04:11.063
I thought: this is how[br]nature paints a photo,

0:04:11.087,0:04:13.650
one femto frame at a time,

0:04:13.674,0:04:17.931
but of course our eye sees[br]an integral composite.

0:04:19.923,0:04:22.057
But if you look at this tomato[br]one more time,

0:04:22.081,0:04:24.826
you will notice, as the light[br]washes over the tomato,

0:04:24.850,0:04:26.514
it continues to glow.

0:04:26.538,0:04:28.565
It doesn't become dark. Why is that?

0:04:28.589,0:04:30.887
Because the tomato is actually ripe,

0:04:30.911,0:04:33.300
and the light is bouncing[br]around inside the tomato,

0:04:33.324,0:04:36.744
and it comes out after several[br]trillionths of a second.

0:04:37.714,0:04:42.053
So in the future, when this femto-camera[br]is in your camera phone,

0:04:42.077,0:04:44.031
you might be able to go to a supermarket

0:04:44.055,0:04:47.110
and check if the fruit is ripe[br]without actually touching it.

0:04:47.134,0:04:48.820
(Laughter)

0:04:48.844,0:04:52.533
So how did my team at MIT[br]create this camera?

0:04:53.906,0:04:55.850
Now, as photographers, you know,

0:04:55.874,0:04:59.835
if you take a short exposure photo,[br]you get very little light.

0:04:59.859,0:05:03.367
But we're going to go a billion times[br]faster than your shortest exposure,

0:05:03.391,0:05:05.320
so you're going to get hardly any light.

0:05:05.344,0:05:07.257
So what we do is we send that bullet --

0:05:07.281,0:05:09.777
that packet of photons --[br]millions of times,

0:05:09.801,0:05:12.884
and record again and again[br]with very clever synchronization,

0:05:12.908,0:05:14.847
and from the gigabytes of data,

0:05:14.871,0:05:17.091
we computationally weave together

0:05:17.115,0:05:20.014
to create those femto-videos I showed you.

0:05:21.196,0:05:25.346
And we can take all that raw data[br]and treat it in very interesting ways.

0:05:26.015,0:05:27.832
So, Superman can fly.

0:05:27.856,0:05:30.114
Some other heroes can become invisible.

0:05:30.610,0:05:35.392
But what about a new power[br]for a future superhero:

0:05:35.416,0:05:37.277
To see around corners.

0:05:38.094,0:05:42.563
The idea is that we could[br]shine some light on the door,

0:05:42.587,0:05:45.238
it's going to bounce, go inside the room,

0:05:45.262,0:05:47.668
some of that is going to reflect[br]back on the door,

0:05:47.692,0:05:49.175
and then back to the camera.

0:05:49.199,0:05:52.196
And we could exploit[br]these multiple bounces of light.

0:05:52.687,0:05:55.354
And it's not science fiction.[br]We have actually built it.

0:05:55.378,0:05:57.444
On the left, you see our femto-camera.

0:05:57.468,0:05:59.823
There's a mannequin hidden behind a wall,

0:05:59.847,0:06:02.352
and we're going to bounce[br]light off the door.

0:06:02.749,0:06:06.242
So after our paper was published[br]in Nature Communications,

0:06:06.266,0:06:08.408
it was highlighted by Nature.com,

0:06:08.432,0:06:10.908
and they created this animation.

0:06:10.932,0:06:12.216
(Music)

0:06:12.240,0:06:15.104
[A laser pulse is fired]

0:06:15.128,0:06:18.000
(Music)

0:06:18.024,0:06:20.891
Ramesh Raskar: We're going to fire[br]those bullets of light,

0:06:20.915,0:06:24.291
and they're going to hit this wall,

0:06:24.315,0:06:26.946
and because of the packet of the photons,

0:06:26.970,0:06:29.080
they will scatter in all the directions,

0:06:29.104,0:06:31.445
and some of them will reach[br]our hidden mannequin,

0:06:31.469,0:06:34.851
which in turn will again[br]scatter that light,

0:06:34.875,0:06:39.446
and again in turn, the door will reflect[br]some of that scattered light.

0:06:40.683,0:06:44.179
And a tiny fraction of the photons[br]will actually come back to the camera,

0:06:44.203,0:06:45.354
but most interestingly,

0:06:45.378,0:06:48.461
they will all arrive[br]at a slightly different time slot.

0:06:48.926,0:06:53.759
(Music)

0:06:53.783,0:06:56.296
And because we have a camera[br]that can run so fast --

0:06:56.320,0:06:59.402
our femto-camera --[br]it has some unique abilities.

0:06:59.426,0:07:02.244
It has very good time resolution,

0:07:02.268,0:07:05.163
and it can look at the world[br]at the speed of light.

0:07:06.114,0:07:09.371
And this way, we know the distances,[br]of course to the door,

0:07:09.395,0:07:11.172
but also to the hidden objects,

0:07:11.196,0:07:14.353
but we don't know which point[br]corresponds to which distance.

0:07:14.377,0:07:18.422
(Music)

0:07:18.446,0:07:21.876
By shining one laser,[br]we can record one raw photo,

0:07:21.900,0:07:24.924
which, if you look on the screen,[br]doesn't really make any sense.

0:07:24.948,0:07:27.068
But then we will take[br]a lot of such pictures,

0:07:27.092,0:07:29.252
dozens of such pictures,[br]put them together,

0:07:29.276,0:07:31.734
and try to analyze[br]the multiple bounces of light,

0:07:31.758,0:07:35.209
and from that, can we see[br]the hidden object?

0:07:35.233,0:07:37.367
Can we see it in full 3D?

0:07:38.113,0:07:40.042
So this is our reconstruction.

0:07:40.066,0:07:45.196
(Music)

0:07:45.220,0:07:52.196
(Applause)

0:07:52.791,0:07:54.260
Now, we have some ways to go

0:07:54.284,0:07:57.419
before we take this[br]outside the lab on the road,

0:07:57.443,0:08:01.172
but in the future, we could create[br]cars that avoid collisions

0:08:01.196,0:08:02.677
with what's around the bend.

0:08:03.581,0:08:07.668
Or we can look for survivors[br]in hazardous conditions

0:08:07.692,0:08:10.916
by looking at light reflected[br]through open windows.

0:08:11.604,0:08:17.475
Or we can build endoscopes that can see[br]deep inside the body around occluders,

0:08:17.499,0:08:19.199
and also for cardioscopes.

0:08:19.223,0:08:21.589
But of course,[br]because of tissue and blood,

0:08:21.613,0:08:22.868
this is quite challenging,

0:08:22.892,0:08:24.827
so this is really a call for scientists

0:08:24.851,0:08:27.158
to start thinking about femto-photography

0:08:27.182,0:08:29.063
as really a new imaging modality

0:08:29.087,0:08:32.607
to solve the next generation[br]of health-imaging problems.

0:08:33.082,0:08:36.939
Now, like Doc Edgerton,[br]a scientist himself,

0:08:36.963,0:08:39.266
science became art --

0:08:39.290,0:08:41.952
an art of ultra-fast photography.

0:08:42.880,0:08:44.031
And I realized

0:08:44.055,0:08:47.496
that all the gigabytes of data[br]that we're collecting every time,

0:08:47.520,0:08:49.941
are not just for scientific imaging.

0:08:49.965,0:08:55.504
But we can also do a new form[br]of computational photography,

0:08:55.528,0:08:58.060
with time-lapse and color coding.

0:08:59.147,0:09:00.616
And we look at those ripples.

0:09:00.640,0:09:01.798
Remember:

0:09:01.822,0:09:07.295
The time between each of those ripples[br]is only a few trillionths of a second.

0:09:08.602,0:09:10.809
But there's also something[br]funny going on here.

0:09:10.833,0:09:13.629
When you look at the ripples[br]under the cap,

0:09:13.653,0:09:15.836
the ripples are moving away from us.

0:09:16.851,0:09:18.765
The ripples should be moving towards us.

0:09:18.789,0:09:20.161
What's going on here?

0:09:20.185,0:09:26.975
It turns out, because we're recording[br]nearly at the speed of light,

0:09:26.999,0:09:28.888
we have strange effects,

0:09:28.912,0:09:32.173
and Einstein would have loved[br]to see this picture.

0:09:32.197,0:09:33.236
(Laughter)

0:09:33.260,0:09:36.307
The order at which events[br]take place in the world

0:09:36.331,0:09:39.657
appears in the camera[br]sometimes in reversed order.

0:09:41.046,0:09:44.241
So by applying the corresponding[br]space and time warp,

0:09:44.265,0:09:46.632
we can correct for this distortion.

0:09:48.606,0:09:52.955
So whether it's for photography[br]around corners,

0:09:52.979,0:09:56.975
or creating the next generation[br]of health imaging,

0:09:56.999,0:09:59.891
or creating new visualizations,

0:09:59.915,0:10:01.539
since our invention,

0:10:01.563,0:10:06.688
we have open-sourced all the data[br]and details on our website,

0:10:06.712,0:10:13.665
and our hope is that the DIY,[br]the creative and the research communities

0:10:13.689,0:10:19.194
will show us that we should stop obsessing[br]about the megapixels in cameras --

0:10:19.218,0:10:20.497
(Laughter)

0:10:20.521,0:10:24.909
and start focusing[br]on the next dimension in imaging.

0:10:25.607,0:10:26.917
It's about time.

0:10:27.337,0:10:28.489
Thank you.

0:10:28.513,0:10:35.383
(Applause)