Experimental camera records light in slow motion

MIT Media Lab researchers have developed a camera that can capture images at one trillion frames per second. The system enables scientists to create super slow-motion video of light itself as it travels back and forth inside a two liter bottle. Check out the video above, particularly the magic at 1:49. The technology has applications in medical imaging, materials science, and also chemical analysis. Eventually, principal investigator Ramesh Raskar thinks it could trickle down to consumer photography as well. From MIT:

The system relies on a recent technology called a streak camera, deployed in a totally unexpected way. The aperture of the streak camera is a narrow slit. Particles of light — photons — enter the camera through the slit and pass through an electric field that deflects them in a direction perpendicular to the slit. Because the electric field is changing very rapidly, it deflects late-arriving photons more than it does early-arriving ones.

The image produced by the camera is thus two-dimensional, but only one of the dimensions — the one corresponding to the direction of the slit — is spatial. The other dimension, corresponding to the degree of deflection, is time. The image thus represents the time of arrival of photons passing through a one-dimensional slice of space…

As a longtime camera researcher, Raskar also sees a potential application in the development of better camera flashes. “An ultimate dream is, how do you create studio-like lighting from a compact flash? How can I take a portable camera that has a tiny flash and create the illusion that I have all these umbrellas, and sport lights, and so on?” asks Raskar, the NEC Career Development Associate Professor of Media Arts and Sciences. “With our ultrafast imaging, we can actually analyze how the photons are traveling through the world. And then we can recreate a new photo by creating the illusion that the photons started somewhere else.”

"Trillion-frame-per-second video"


  1. Sorry, but can we please correct the language a tad? The camera can not, in fact, capture images at 1 trillion frames per second. It can create a sequence of images which SIMULATE the capture of an event at 1 trillion frames per second, but the contraption does not do what you say it does in the opening sentence. Perhaps this is pedantic but that is the sort of claim that will get stuck in people’s psyche, like the 10% of brain myth…

    1. This camera captures a single spatial line of the image over consecutive timeslices at a picosecond or sub-picosecond resolution.  By using a mirror to adjust which spatial line of the image is captured and synchronizing the illumination and capture pulses, a collection of 1-d THz videos can be assembled into a 2-d THz video.

      Good enough?

    1. No, “stop motion” is when you take a series of still images of non-moving objects and play them in sequence to create the illusion that they are moving on their own, like Gumby or Terry Gilliam’s paper cut-outs.

      1. But as a composite imager the camera is not tracking one photon in slow motion over a 30 second time frame. It’s tracking “billions and billions” of them over time, more analagous to stop motion, as I would cast it.

        On another note, this very cool project feels more like ‘reflections on the cave wall’ than the CERN research does. I’m sure it will prove useful, no matter if we ever agree about just how it works. Maybe there are magnets involved?

        1. Then the phrase you are looking for is “time lapse photography.” Except that the aperture isn’t left open. Hmm.

          1. I believe stop motion is right. They are taking a whole lot of stills and giving the illusion of motion. One difference is that each frame of video is made up of a whole lot of 1 dimensional lines each of which is shot separately, but at the same time relative to the laser pulse. Then a new frame is made up of slice pictures at a new time relative to the laser pulse.

            It’s a great idea, but “practical uses”? Nice bid for investor cash, but can’t see it myself.

          2. I believe stop motion is right. They are taking a whole lot of stills and giving the illusion of motion.

            But ALL video/motion picture techniques do that. I think the term “stop motion” should be reserved for any animation technique that makes objects appear to move by repositioning them between frames. The light beams in this case really DO move. The resulting video is still a photographic illusion of sorts, but it’s a different kind of illusion.

  2. This isn’t one-shot photography, either. It’s a method of taking many line scans and stitching them together to make the illusion of one picture. So it only works on static subjects.

    Electrical engineers have had sampling oscilloscopes for many decades that do the same thing with voltage, but somewhat slower. They have impressive specifications until you learn that they need to sample the signal thousands of times to make one trace.

  3. As a professional photographer I find the implications in the last sentence to be incredible and also scary.  On the one hand, I’ve invested tens of thousands of dollars in lighting equipment to get perfectly-lit shots (and years as an assistant and intern to use it properly), and this would completely tear down that barrier and make it all obsolete. On the flip side, I can’t wait to try it in exotic places where I can’t practically bring my lighting gear. (which is everywhere except a studio or a large truck)

    I can tell you one thing, I think that what he’s talking about is the inevitable future (I think what Lytro is doing with their light-field camera is just scratching the surface at photon-based image capture), and the kind of people who talk about “purity of photography” (and all these people are asshats who can’t adapt to new equipment or business models) are going to shit a brick. They ain’t seen nothing yet.

    1. I found the last sentence kind of funny because what kind of “sophisticated light sources” are we talking about here, that a titanium sapphire laser is considered a light weight alternative?

      1. This part:
        “An ultimate dream is, how do you create studio-like lighting from a compact flash? How can I take a portable camera that has a tiny flash and create the illusion that I have all these umbrellas, and sport lights, and so on?”

        1.  Oh, I see what you mean. I had just watched the video. But the sentence from the video was on roughly the same topic, of using this tech to replace elaborate lighting.

          What they demonstrated with this system requires incredibly fast bursts of light, measured in nanoseconds, which is itself pretty fancy lighting.

  4. The article is written inaccurately, there is no way to deflect photons with a field of any kind (except gravity). The streak camera uses electrons in a cathode ray tube which are knocked loose from an imaging  plate by photons, and the electrons are deflected to splay them out in space, so their arrival positions correspond to elapsed time.

    1. the Wiki page for a “streak camera” says:

      “Optoelectronic streak cameras work by directing the light onto a photocathode, which when hit by photons produces electrons via the photoelectric effect. The electrons are accelerated in a cathode ray tube and pass through an electric field produced by a pair of plates, which deflects the electrons sideways. By modulating the electric potential between the plates, the electric field is quickly changed to give a time-varying deflection of the electrons, sweeping the electrons across a phosphor screen at the end of the tube. A linear detector, such as a charge-coupled device (CCD) array is used to measure the streak pattern on the screen, and thus the temporal profile of the light pulse.”

      that makes a lot more sense than trying to deflect electrically-neutral particles with an electrical field.

      someone had a bit of writeup fail.

  5. While this looks really, really cool, I am suspicious that the visual demonstrations here are just computer reproductions that aren’t disclosed as such.  I could be wrong – but it just feels like it.

  6. Ok, so if the shots were of a photon, first, isn’t that infinitely small? That seems awfully big for a single photon.

    Second, what are the reflections in the sequence around 1:49 then?  Sub-photons?  I’m thinking there’s some serious terminology weirdness going on here.

    What the line in the bottle actually looks like is their pan of the mirror, not of an actual cross section of light.

    1. I was also confused.  Isn’t the only way to “see” a photon to have that photon hit your sensor.  This must be quite a few photons, some of which are scattered as they hit the media inside the coke bottle (whether that media is just air or some liquid).

  7. This is more like the effect of seeing the spokes of a wheel go backward based on lighting and/or frame rates than it is actually seeing light moving in slow motion.
    Perhaps they take us for fools, but I’m not willing to be taken – the probes are deucedly uncomfortable!

    1. Thank you. Yes! This is exactly the right analogy that others above had been struggling with (stop-motion, time-lapse, etc.).

      This is a huge number of pulses, and a huge number of photographs, slightly out-of-phase with each other so that when you put them together you get the illusion of slow movement, just light the illusion of a slow-moving propeller blade.

      I wish the write-up didn’t say that it was capturing “a trillion frames per second,” which the video doesn’t even suggest, because this is cool enough without ruining it with hyperbole and the eventual discussions correcting it that have to result.

      So, to clarify, what this is actually doing: stitching together thousands of individual high-speed photos of pulses of light, potentially taken over a period of several seconds or more, which result in the illusion of one beam of light traveling slowly.

      1. I think it’s capturing real motion, just along a narrow strip. They keep repeating the motion to give it vertical dimension.

        The whole reason the camera can only capture a narrow strip is because it’s using the vertical part of the plate to capture the motion.

        1. It’s not capturing “real motion,” because each frame of the video is actually of a different pulse of light. The illusion of motion that is created is, indeed, exactly the same as that of one pulse of light, which is what makes this very cool, but it’s not capturing the “real motion” of one pulse of light.

          It may be semantics, sure, but 1) I don’t think you’d say that the slow-moving propellor blades you see on film is the propellor’s real motion (especially when they go backwards…), and 2) it’s still flat-out completely incorrect to say that this camera is “capturing images at one trillion frames per second.”

          1. To be more precise with the semantic point, it captures real motion along a horizontal axis but relies on the repeated pulses for motion along a vertical axis. Does that seem accurate to everyone?

  8. This seems analogous to the astronomical “light echo” phenomenon whereby a star emits a large burst of light which “ripples out” over time to illuminate surrounding gas particles. See: http://www.spacetelescope.org/news/heic0617/

    The same phenomenon occurs if you were to set off a flashbulb in a dark room.  The light from the flashbulb hits our eyes before the light that bounces off of the room’s walls, but the difference is so minute that our eyes cannot detect it. What I assume the video is showing is this “ripple” in extreme slow motion. In other words, the photons from a single burst of light propagating outward into a dark space. 

  9. If you have a camera that fast and sophisticated as they imagine could re-light a scene, you don’t need extra light from a flash. But you could use it to capture distance information for 3-D applications from a single lens, perhaps even do reverse ray-tracing.

  10. Speaking as a user of scientific high speed cameras, this is very cool but I would hesitate to call it a ‘high speed camera’. Its uses are much more limited. I would lean towards the ‘extreme stop motion’ discussion above.
    Actually capturing  fast motion with this device isn’t really possible, but then, anything that isn’t light wouldn’t appear to be moving at this scale (cm). Besides the actual laser pulses used to determine the motion are 13 ns apart, so any single event will be limited to this time resolution.My criticism would be that streak cameras have been around for donkeys years doing just this sort of thing, and in fact fully gated optical high speed cameras can capture frames 2 nanoseconds apart. Whilst that’s a mere 500 million FPS, they don’t seem to mention in their paper that these things exist and are significantly smaller than their setup.

Comments are closed.