Return to Video

See invisible motion, hear silent sounds. Cool? Creepy? We can't decide

  • 0:01 - 0:06
    So over the past few centuries,
  • 0:06 - 0:09
    microscopes have revolutionized our world.
  • 0:09 - 0:12
    They revealed to us a tiny world
  • 0:12 - 0:14
    of objects, life, and structures
  • 0:14 - 0:17
    that are too small for us
    to see with our naked eyes.
  • 0:17 - 0:20
    They are a tremendous contribution
    to science and technology.
  • 0:20 - 0:23
    Today I'd like to introduce you
    to a new type of microscope,
  • 0:23 - 0:26
    a microscope for changes.
  • 0:26 - 0:29
    It doesn't use optics
    like a regular microscope
  • 0:29 - 0:31
    to make small objects bigger,
  • 0:31 - 0:34
    but instead it uses a video camera
  • 0:34 - 0:36
    and image processing
  • 0:36 - 0:38
    to reveal to us the tiniest motions
  • 0:38 - 0:41
    and color changes in objects and people,
  • 0:41 - 0:45
    changes that are impossible
    for us to see with our naked eyes,
  • 0:45 - 0:46
    and it lets us look at our world
  • 0:46 - 0:49
    in a completely new way.
  • 0:49 - 0:50
    So what do I mean by color changes?
  • 0:50 - 0:53
    Our skin, for example,
    changes its color very slightly
  • 0:53 - 0:55
    when the blood flows under it.
  • 0:55 - 0:58
    That change is incredibly subtle,
  • 0:58 - 1:00
    which is why, when you
    look at other people,
  • 1:00 - 1:02
    when you look at the person
    sitting next to you,
  • 1:02 - 1:06
    you don't see their skin
    or their face changing color.
  • 1:06 - 1:08
    When we look at
    this video of Steve here,
  • 1:08 - 1:10
    it appears to us like a static picture,
  • 1:10 - 1:14
    but once we look at this video
    through our new, special microscope,
  • 1:14 - 1:16
    suddenly we see
    a completely different image.
  • 1:16 - 1:19
    What you see here are small changes
  • 1:19 - 1:20
    in the color of Steve's skin,
  • 1:20 - 1:23
    magnified 100 times
  • 1:23 - 1:25
    so that they become visible.
  • 1:25 - 1:28
    We can actually see a human pulse.
  • 1:28 - 1:31
    We can see how fast
    Steve's heart is beating,
  • 1:31 - 1:33
    but we can also see the actual way
  • 1:33 - 1:37
    that the blood flows in his face.
  • 1:37 - 1:40
    And we can do that not just
    to visualize the pulse
  • 1:40 - 1:44
    but also to actually
    measure our heart rates,
  • 1:44 - 1:46
    and we can do it with regular cameras
  • 1:46 - 1:49
    and without touching the patients.
  • 1:49 - 1:52
    So here you see the pulse and heart rate
  • 1:52 - 1:55
    we extracted from a neonatal baby
  • 1:55 - 1:58
    from a video we took
    with a regular DSLR camera,
  • 1:58 - 1:59
    and the heart rate measurement we get
  • 1:59 - 2:01
    is as accurate as the one you'd get
  • 2:01 - 2:04
    with a standard monitor in a hospital.
  • 2:04 - 2:07
    And it doesn't even have to be
    a video we recorded.
  • 2:07 - 2:10
    We can do it essentially
    with other videos as well.
  • 2:10 - 2:14
    So I just took a short clip
    from "Batman Begins" here
  • 2:14 - 2:15
    just to show Christian Bale's pulse.
  • 2:15 - 2:18
    (Laughter)
  • 2:18 - 2:20
    And you know, presumably
    he's wearing makeup,
  • 2:20 - 2:21
    the lighting here is kind of challenging,
  • 2:21 - 2:24
    but still, just from the video,
    we're able to extract his pulse
  • 2:24 - 2:26
    and show it quite well.
  • 2:26 - 2:28
    So how do we do all that?
  • 2:28 - 2:31
    We basically analyze the changes
  • 2:31 - 2:33
    in the light that's recorded
  • 2:33 - 2:35
    at every pixel in the video over time,
  • 2:35 - 2:37
    and then we crank up those changes.
  • 2:37 - 2:39
    We make them bigger
    so that we can see them.
  • 2:39 - 2:41
    The tricky part is that those signals,
  • 2:41 - 2:42
    those changes that we're after,
  • 2:42 - 2:44
    are extremely subtle,
  • 2:44 - 2:47
    so we have to be very careful
    when you try to separate them
  • 2:47 - 2:51
    from noise that always exists in videos.
  • 2:51 - 2:54
    So we use some clever
    image processing techniques
  • 2:54 - 2:58
    to get a very accurate measurement
    of the color at each pixel in the video,
  • 2:58 - 3:00
    and then the way the color
    changes over time,
  • 3:00 - 3:03
    and then we amplify those changes.
  • 3:03 - 3:07
    We make them bigger to create those types
    of enhanced videos, or magnified videos,
  • 3:07 - 3:09
    that actually show us those changes.
  • 3:09 - 3:11
    But it turns out we can do that
  • 3:11 - 3:13
    not just to show tiny changes in color
  • 3:13 - 3:16
    but also tiny motions,
  • 3:16 - 3:19
    and that's because the light
    that gets recorded in our cameras
  • 3:19 - 3:22
    will change not only if the color
    of the object changes,
  • 3:22 - 3:24
    but also if the object moves.
  • 3:24 - 3:26
    So this is my daughter
  • 3:26 - 3:28
    when she was about two months old.
  • 3:28 - 3:31
    It's a video I recorded
    about three years ago.
  • 3:31 - 3:35
    And as new parents, we all want
    to make sure our babies are healthy,
  • 3:35 - 3:37
    that they're breathing,
    that they're alive of course.
  • 3:37 - 3:39
    So I too got one of those baby monitors
  • 3:39 - 3:41
    so that I could see my daughter
    when she was asleep.
  • 3:41 - 3:43
    And this is pretty much what you'll see
  • 3:43 - 3:45
    with a standard baby monitor.
  • 3:45 - 3:47
    You know, you can see the baby's sleeping,
  • 3:47 - 3:48
    but there's not too much
    information there.
  • 3:48 - 3:50
    There's not too much we can see.
  • 3:50 - 3:51
    Wouldn't it be better,
  • 3:51 - 3:53
    or more informative, or more useful,
  • 3:53 - 3:56
    if instead we could look
    at the view like this.
  • 3:56 - 3:59
    So here I took the motions
  • 3:59 - 4:02
    and I magnified them 30 times,
  • 4:02 - 4:06
    and then I could clearly see that my
    daughter was indeed alive and breathing.
  • 4:06 - 4:09
    (Laughter)
  • 4:09 - 4:11
    Here is a side by side comparison.
  • 4:11 - 4:13
    So again, in the source video,
    in the original video,
  • 4:13 - 4:14
    there's not too much we can see,
  • 4:14 - 4:16
    but once we magnify the motions,
  • 4:16 - 4:18
    the breathing becomes much more visible.
  • 4:18 - 4:20
    And it turns out, there's
    a lot of phenomena
  • 4:20 - 4:22
    we can reveal and magnify
  • 4:22 - 4:24
    with our new motion microscope.
  • 4:24 - 4:27
    We can see how our veins and arteries
  • 4:27 - 4:28
    are pulsing in our bodies.
  • 4:28 - 4:30
    We can see that our eyes
  • 4:30 - 4:33
    are constantly moving
    in this wobbly motion.
  • 4:33 - 4:35
    And that's actually my eye,
  • 4:35 - 4:37
    and again this video was taken
    right after my daughter was born,
  • 4:37 - 4:42
    so you can see I was not
    getting too much sleep.
  • 4:42 - 4:44
    Even when a person is sitting still,
  • 4:44 - 4:46
    there's a lot of information
    we can extract
  • 4:46 - 4:50
    about their breathing patterns,
    small facial expressions.
  • 4:50 - 4:52
    Maybe we could use those motions
  • 4:52 - 4:55
    to tell us something about
    our thoughts or our emotions.
  • 4:55 - 4:58
    We can also magnify small
    mechanical movements,
  • 4:58 - 5:00
    like vibrations in engines,
  • 5:00 - 5:03
    that can help engineers detect
    and diagnose machinery problems,
  • 5:03 - 5:06
    or see how our buildings and structures
  • 5:06 - 5:08
    sway in the wind and react to forces.
  • 5:08 - 5:10
    Those are all things that our society
  • 5:10 - 5:13
    knows how to measure in various ways,
  • 5:13 - 5:15
    but measuring those motions is one thing,
  • 5:15 - 5:17
    and actually seeing those
    motions as they happen
  • 5:17 - 5:20
    is a whole different thing.
  • 5:20 - 5:23
    And ever since we discovered
    this new technology,
  • 5:23 - 5:24
    we made our code available online
  • 5:24 - 5:27
    so that others could use
    and experiment with it.
  • 5:27 - 5:29
    It's very simple to use.
  • 5:29 - 5:31
    It can work on your own videos.
  • 5:31 - 5:34
    Our collaborators at Quantum Research
    even created this nice website
  • 5:34 - 5:37
    where you can upload your videos
    and process them online,
  • 5:37 - 5:39
    so even if you don't have any experience
  • 5:39 - 5:40
    in computer science or programming,
  • 5:40 - 5:42
    you can still very easily experiment
    with this new microscope.
  • 5:42 - 5:45
    And I'd like to show you
    just a couple of examples
  • 5:45 - 5:48
    of what others have done with it.
  • 5:48 - 5:54
    So this video was made by
    a YouTube user called Tamez85.
  • 5:54 - 5:55
    I don't know who that user is,
  • 5:55 - 5:58
    but he, or she, used our code
  • 5:58 - 6:01
    to magnify small belly
    movements during pregnancy.
  • 6:01 - 6:03
    It's kind of creepy.
  • 6:03 - 6:05
    (Laughter)
  • 6:05 - 6:07
    People have used it to magnify
  • 6:07 - 6:10
    pulsing veins in their hands,
  • 6:10 - 6:13
    and you know, it's not real science
    unless you use guinea pigs,
  • 6:13 - 6:17
    and apparently this guinea pig
    is called Tiffany,
  • 6:17 - 6:20
    and this YouTube user claims
    it is the first rodent on Earth
  • 6:20 - 6:23
    that was motion-magnified.
  • 6:23 - 6:24
    You can also do some art with it.
  • 6:24 - 6:28
    So this video was sent to me
    by a design student at Yale.
  • 6:28 - 6:30
    she wanted to see
    if there's any difference
  • 6:30 - 6:31
    in the way her classmates move.
  • 6:31 - 6:33
    She made them all stand still,
  • 6:33 - 6:35
    and then magnified their motions
  • 6:35 - 6:37
    It's like seeing still pictures
  • 6:37 - 6:39
    come to life.
  • 6:39 - 6:41
    And the nice thing with all those examples
  • 6:41 - 6:43
    is that we had nothing to do with them.
  • 6:43 - 6:47
    We just provided this new tool,
    a new way to look at the world,
  • 6:47 - 6:52
    and then people find other interesting,
    new, and creative ways of using it.
  • 6:52 - 6:54
    But we didn't stop there.
  • 6:54 - 6:57
    This tool not only allows us
    to look at the world in a new way,
  • 6:57 - 7:00
    it also redefines what we can do
  • 7:00 - 7:03
    and pushes the limits of what
    we can do with our cameras.
  • 7:03 - 7:05
    So as scientists, we started wondering,
  • 7:05 - 7:09
    what other types of physical phenomena
    produce tiny motions
  • 7:09 - 7:12
    that we could now use
    our cameras to measure?
  • 7:12 - 7:15
    And one such phenomena
    that we focused on recently
  • 7:15 - 7:16
    is sound.
  • 7:16 - 7:18
    Sound, as we all know,
    is basically changes
  • 7:18 - 7:20
    in air pressure that
    travel through the air.
  • 7:20 - 7:22
    Those pressure waves hit objects
  • 7:22 - 7:24
    and they create small vibrations in them,
  • 7:24 - 7:26
    which is how we hear
    and how we record sound.
  • 7:26 - 7:30
    But it turns out that sound
    also produces visual motions.
  • 7:30 - 7:33
    Those are motions
    that are not visible to us
  • 7:33 - 7:36
    but are visible to a camera
    with the right processing.
  • 7:36 - 7:37
    So here are two examples.
  • 7:37 - 7:41
    This is me demonstrating
    my great singing skills.
  • 7:44 - 7:45
    (Laughter)
  • 7:45 - 7:48
    And I took a high speed video
    of my throat while I was humming.
  • 7:48 - 7:49
    Again, if you stare at that video,
  • 7:49 - 7:50
    there's not too much
    you'll be able to see,
  • 7:50 - 7:52
    but once we magnify the motions
  • 7:52 - 7:53
    100 times,
  • 7:53 - 7:55
    we can see all the motions and ripples
  • 7:55 - 7:59
    in the neck that are involved
    in producing the sound.
  • 7:59 - 8:01
    That signal is there in that video.
  • 8:01 - 8:04
    We also know that singers
    can break a wine glass
  • 8:04 - 8:05
    if they hit the correct note.
  • 8:05 - 8:07
    So here, we're going to play a note
  • 8:07 - 8:10
    that's in the resonance
    frequency of that glass
  • 8:10 - 8:12
    through a loudspeaker that's next to it.
  • 8:12 - 8:13
    Once we play that note
  • 8:13 - 8:16
    and magnify the motions 250 times,
  • 8:16 - 8:18
    we can very clearly see how the glass
  • 8:18 - 8:22
    vibrates and resonates
    in response to the sound.
  • 8:22 - 8:25
    It's not something you're used
    to seeing every day,
  • 8:25 - 8:28
    but this made us think,
    it gave us this crazy idea:
  • 8:28 - 8:31
    can we actually invert this process
  • 8:31 - 8:34
    and recover sound from video
  • 8:34 - 8:36
    by analyzing the tiny vibrations
  • 8:36 - 8:38
    that sound waves create in object
  • 8:38 - 8:40
    and essentially convert those
  • 8:40 - 8:42
    back into the sounds that produce them.
  • 8:42 - 8:47
    In this way, we can turn
    everyday objects into microphones.
  • 8:47 - 8:49
    So that's exactly what we did.
  • 8:49 - 8:52
    So here's an empty bag of chips
    that was lying on a table,
  • 8:52 - 8:54
    and we're going to turn that bag of chips
  • 8:54 - 8:56
    into a microphone by filming it
    with a video camera
  • 8:56 - 8:58
    and analyzing the tiny motions
  • 8:58 - 9:00
    that sound waves create in it.
  • 9:00 - 9:02
    So here's the sound
    that we played in the room.
  • 9:02 - 9:07
    (Music: "Mary Had A Little Lamb")
  • 9:11 - 9:14
    And this is a high-speed video
    we recorded of that bag of chips.
  • 9:14 - 9:15
    Again, it's playing.
  • 9:15 - 9:17
    There's no chance you'll be able
    to see anything going on in that video
  • 9:17 - 9:18
    just by looking at it,
  • 9:18 - 9:22
    but here's the sound we were able
    to recover just by analyzing
  • 9:22 - 9:24
    the tiny motions in that video.
  • 9:24 - 9:27
    (Music: "Mary Had A Little Lamb")
  • 9:41 - 9:42
    I call it -- Thank you.
  • 9:42 - 9:48
    (Applause)
  • 9:50 - 9:52
    I call it the visual microphone.
  • 9:52 - 9:56
    We actually extract audio signals
    from video signals.
  • 9:56 - 9:59
    And just to give you a sense
    of the scale of the motions here,
  • 9:59 - 10:00
    a pretty loud sound
  • 10:00 - 10:04
    will cause that bag of chips
    to move less than a micrometer.
  • 10:04 - 10:07
    That's one thousandth of a millimeter.
  • 10:07 - 10:09
    That's how tiny the motions are
  • 10:09 - 10:10
    that we are now able to pull out
  • 10:10 - 10:14
    just by observing how light
    bounces off objects
  • 10:14 - 10:16
    and gets recorded by our cameras.
  • 10:16 - 10:19
    We can recover sounds
    from other objects like plants.
  • 10:19 - 10:25
    (Music: "Mary Had A Little Lamb")
  • 10:27 - 10:29
    And we can recover speech as well.
  • 10:29 - 10:32
    So here's a person speaking in a room.
  • 10:32 - 10:36
    Voice: Mary had a little lamb
    whose fleece was white as snow,
  • 10:36 - 10:41
    and everywhere that Mary went,
    that lamb was sure to go.
  • 10:41 - 10:43
    Michael Rubinstein: And here's
    that speech again recovered
  • 10:43 - 10:46
    just from this video
    of that same bag of chips.
  • 10:46 - 10:53
    Voice: Mary had a little lamb
    whose fleece was white as snow,
  • 10:53 - 10:56
    and everywhere that Mary went,
    that lamb was sure to go.
  • 10:56 - 10:59
    MR: We used "Mary Had A Little Lamb"
  • 10:59 - 11:02
    because those are said to be
    the first words that Thomas Edison
  • 11:02 - 11:05
    spoke into his photograph in 1877.
  • 11:05 - 11:08
    It was one of the first sound
    recording devices in history.
  • 11:08 - 11:10
    It basically directed the sounds
  • 11:10 - 11:11
    onto a diaphragm
  • 11:11 - 11:15
    that vibrated a needle that essentially
    engraved the sound on a tin foil
  • 11:15 - 11:17
    that was wrapped around the cylinder.
  • 11:17 - 11:19
    Here's a demonstration
  • 11:19 - 11:21
    of recording and replaying sound
  • 11:21 - 11:23
    with Edison's phonograph.
  • 11:23 - 11:26
    Voice (Shouting): Testing,
    testing, one two three.
  • 11:26 - 11:30
    Mary had a little lamb
    whose fleece was white as snow,
  • 11:30 - 11:34
    and everywhere that Mary went,
    the lamb was sure to go.
  • 11:34 - 11:36
    Testing, testing, one two three.
  • 11:36 - 11:40
    Mary had a little lamb
    whose fleece was white as snow,
  • 11:40 - 11:46
    and everywhere that Mary went,
    the lamb was sure to go.
  • 11:46 - 11:50
    MR: And now, 137 years later,
  • 11:50 - 11:54
    we're able to get sound
    in pretty much similar quality
  • 11:54 - 11:56
    but by just watching objects
  • 11:56 - 11:58
    vibrate to sound with cameras,
  • 11:58 - 12:00
    and we can even do that when the camera
  • 12:00 - 12:02
    is 15 feet away from the object
  • 12:02 - 12:04
    behind soundproof glass.
  • 12:04 - 12:07
    So this is the sound we were able
    to recover in that case.
  • 12:07 - 12:13
    Voice: Mary had a little lamb
    whose fleece was white as snow,
  • 12:13 - 12:17
    and everywhere that Mary went,
    the lamb was sure to go.
  • 12:18 - 12:21
    MR: And of course, surveillance is
    the first application that comes to mind.
  • 12:21 - 12:24
    (Laughter)
  • 12:24 - 12:28
    But it might actually be useful
    for other things as well.
  • 12:28 - 12:31
    Maybe in the future, we'll be able
    to use it, for example,
  • 12:31 - 12:33
    to recover sound across space,
  • 12:33 - 12:37
    because sound can't travel
    in space, but light can.
  • 12:37 - 12:39
    We've only just begun exploring
  • 12:39 - 12:42
    other possible uses
    for this new technology.
  • 12:42 - 12:45
    It lets us see physical processes
    that we know are there
  • 12:45 - 12:49
    but that we've never been able
    to see with our own eyes until now.
  • 12:49 - 12:51
    This is our team.
  • 12:51 - 12:53
    Everything I showed you today
    is a result of a collaboration
  • 12:53 - 12:55
    with this great group
    of people you see here,
  • 12:55 - 12:57
    and I encourage you and welcome you
  • 12:57 - 12:58
    to check out our website,
  • 12:58 - 12:59
    try it out yourself,
  • 12:59 - 13:02
    and join us in exploring
    this world of tiny motions.
  • 13:02 - 13:04
    Thank you.
  • 13:04 - 13:05
    (Applause)
Title:
See invisible motion, hear silent sounds. Cool? Creepy? We can't decide
Speaker:
Michael Rubinstein
Description:

more » « less
Video Language:
English
Team:
closed TED
Project:
TEDTalks
Duration:
13:18
  • 108
    00:05:30,708 --> 00:05:33,901
    Our collaborators at Quantum Research
    even created this nice website
    ->
    Our collaborators at Quanta Research
    even created this nice website
    # See https://videoscope.qrilab.com/

    114
    00:05:48,470 --> 00:05:53,787
    So this video was made by
    a YouTube user called Tamez85.
    ->
    So this video was made by
    a YouTube user called Tomez85.
    # See https://www.youtube.com/user/Tomez85

  • 237
    00:12:39,157 --> 00:12:42,176
    other possible uses
    for this new technology.
    ->
    all the possible uses
    for this new technology.

English subtitles

Revisions Compare revisions