Photographing the speed of light: Ramesh Raskar at TEDGlobal 2012

Posted by: Helen Walters

Ramesh Raskar

MIT professor Ramesh Raskar starts his talk by showing the classic Doc Edgerton photograph of an apple being shot by a bullet. It demonstrated an exposure of a millionth of a second. Wonderful, right? he asks us, to wide agreement. Only now, 50 years later, technology advances mean we can photograph a million times faster than this. “Now we can see the world at a trillion frames per second,” he says. “I present to you femto-photography, a new imaging technique so fast that it can create slow-motion videos of light in motion.”

What does this mean? Well, it means that now Raskar and co can create cameras that look around corners. “We can challenge what we mean by camera,” he says.

He shows video of a “bullet” of light passing through a Coke bottle. The film is silent and slow, and someone whoops excitedly as soon as it’s over. That guy’s grasped what Raskar now explains to the rest of us: “This would take place in less than a nanosecond. It’s slowed down by factor of 10 billion,” he explains.

Ramesh Raskar

Next, Raskar shows video of a pulse of light hitting a tomato. “It’s like throwing a stone into a pond of water,” he says of the ripples of light, describing this as how nature paints a photo. And why does the tomato continue to glow even after the light has gone? Because it is ripe, he tells us. The light bounces around inside the tomato. That means that in the future, when we have femtocameras in our cell phones, we’ll be able to check if fruit is ripe in the supermarket.

But that’s not all the femto-photography can do. It’s a superhero in its own right. By manipulating light, it’s possible to look around corners. This is really possible, he tells us. It’s not magic. They’ve done it. “We can look at the world at the speed of light.”

There’s some way to go before this will leave the lab, he acknowledges. But the implications are exciting and vast. Imagine a car that could avoid an upcoming collision around the next bend. Or a rescue operation that could detect buried people. Or new types of cardioscopes or endoscopes for healthcare. Scientists, meanwhile, can think about femto-photography as a way to solve the next generation of health imaging.

Just as Edgerton’s scientific experiments became art, perhaps too could femto-photography become a new artform. Already Raskar has noticed some strange effects within the still shots of the ripples of light. “Einstein would have loved to have seen this picture,” he says of one of the frames.

Whatever its eventual application, Raskar finishes with a wonderful announcement. All of this is now open source. “Our hope is the DIY-ers, the creatives and the research community will show us we should stop obsessing about megapixels in cameras and start focusing on the next generation in imaging. It’s about time,” he concludes. The audience laughs, loves and rises to its collective feet.

(And in a Q&A, Raskar shares another, unrelated, but equally fantastic invention: NETRA, a snap-on lens for an iPhone that can quickly diagnose nearsightedness and even cataracts.)

Photos: James Duncan Davidson

Comments (8)

  • Pingback: Femto Photography: One Trillion Frames Per Second |

  • Pingback: Femto-photography : The high speed imaging technology! | PZYCODERZ

  • Pingback: 1,000,000,000,000 Frames/Second Photography – Ramesh Raskar … | Bear Tales

  • Dale Stocker commented on Aug 29 2012

    Here I am adding another wile theory about the speed of light not being a constant as such. Again not a need to read if you are not interested, but if the proof to these theories is someday validated then at least I will know I had the right idea.
    There may be a premature assumption of the part of science that the speed of light is a constant. The fact is that we have not been able to measure this speed with any accuracy until this recent history. Here I submit to you a theory that may in time be proven one way or another, but to me there is a symmetry within this idea and the way the cosmos seems to operate.
    The relationship of the speed of light and time would be this:
    (Light speed)*(Light speed) = (Time passed) or C2=T

    The reason I am not considering a linear relationship is that we do not see a great change of the speed of light at this point. Though this may not be the exact equation, until we have found with some evidence for any relationship and the data to work with this is the closes I can come to its approximation.

    The idea in its simplest for would be this:

    At the beginning there is not time passage and no speed of light, so both are zero…
    At the first moment of time (We will for this example call a second) there is to first formation of space time and now a distance that light (We will for this example just consider all phenomena that travels at the speed of light), (We will for this example consider call this distance a meter).

    So at | Time | Speed |
    | 0 | 0 |
    | 1 | 1 |
    | 4 | 2 |
    | 9 | 3 |
    | 25 | 5 |
    | 100 | 10 |

    So trying to calculate the variation on the assumption that the beginning was 14 billion years ago.
    | 4.421088e+17 | 299,792,458 |

    T*X=C*C (X being the variation from our seconds to meters conversion ratio) and T=(C*C)/X
    (4.421088e+17)*X = (8.987551787368176e+16)

    X = (8.987551787368176e+16)/(4.421088e+17) = 0.2032882355512529

    So for 1 meter of change it would take: 2949432441.941939 seconds
    1 Millimeter then takes 2949432.441941939 or in days 34.13694955951318
    These calculations are only based on the approximation of the time factor, but if they are even close enough there should be a millimeter change in the speed of light going faster at about 35 days.

    Setting up the experiment:

    This could be done with a large 10 meter by 10 meter enclosure that can be kept under environmental control to try and eliminate any variations do to temperature. It would be sealed and vacuumed. Sensors place inside and out to insure that conditions at the time of testing are the same.
    We then place a laser at one corner to beam strait across. This then hits a mirror at a slight angle to bounce the light back but off to the side by a centimeter. Here it is again reflected but by a mirror that covers the rest of this side, back to the other side. The angle of inclination would remain and another mirror covering the other side would again bounce the light back. Each bounce would move the light beam over about 1 centimeter till it reached the end of the box. At this point the beam can either be detected or another mirror at an angle can send it back down to reiteration of bounces back two the corner of the laser were it can be detected, giving it 2 cycles though the apparitions. Within all of this a physical measurement of the enclosure and each bounce can be known and the total distance the light traveled calculated. Variation due to light fading from the multiple reflections or other factors may come into consideration. In all the measurement does not have to be exact to what we consider the speed of light, as the test will be made in all likelihood on the earth were other factors may enter in from gravity to having stray atoms in the chamber because of a imperfect vacuum. But his is not a factor within this test as were are looking for a variation of the speed over time and not the speed itself. As long as all conditions are kept standard this test may show whatever results are correct. One of the most important pieces of this experiment is the timer. Here we can use a tunable gigahertz digital oscillator. This also being hooked up to a oscilloscope so to be calibrated to an exact frequency forever test. The faster this oscillator the better to get any indication of variation, but also within the confines of the equipment from the starting pules and laser ignition to the detection. Again if all parameters remain constant then any change in the speed of light can be recorded and see if this variation is a function of time. We can run the test 10 times a day to take an average. Then every day for some time. It should take no more than a year to get the data on whether or not there is any variation that fits this theory.

    Why it happens: The expansion of space and/or the Higgs field.
    I will try to explain this as best I can with an imagination exercise. It has often been used as a method to show how the universe is expanding by placing the galaxies on the surface of a balloon and then inflating it for show how everything is moving apart. This simple example will also work for me in this explanation. So we have a balloon of such and such size. On this balloon we mark to points, A and B of course. Now, there is a certain distance between these points. Let’s imagine that there is some interaction of the nature of the speed of light, so at point A we send a photon to point B. It takes a certain amount of time. No, if while that photon is traveling the balloon is inflated a bit, then it would either take a bit more time or have to compensate for this by traveling a bit faster. As you can see in my theory it will just travel a bit faster with the expansion. This theory would give some credence to the idea that the medium (Space or the Higgs field, or even an old idea of the ether) was expanding. Though at this point that may be another assumption, the only way to know is with more valid test.

    That is all I will say at this point about this theory as it should be enough to be considered or not. If you do consider it I have other ideas you may be interested in.

  • Pingback: Fora de Ordem |

  • commented on Jul 10 2012

    Reblogged this on tum-tum stuff.

  • commented on Jun 29 2012

    If we are able to photograph light and then slow it down, does that mean we are looking at something at a speed faster than that of light?

    • Diego Vargas Urbina commented on Aug 11 2012

      No, photography needs the light to reach the digital sensor (or film) to record anything, so if we get a camera faster than the speed of light, it wouldn’t capture anything (in void) or it would be capturing light from a little bit ago. That’s why you don’t see a bubble of light floating midair in the video, you can only see the reflected light, that is, light that is interacting with something else (the table, the bottle, the cap, etc).