Slashdot Mirror


World's Fastest Camera Captures 4.4 Trillion Frames Per Second

Diggester writes Japanese researchers have recently designed a motion picture camera which is capable of capturing 4.4 trillion frames per second, making it the fastest camera in the world. The technique that allows for such speed is called STAMP (sequentially timed all-optical mapping photography). The research paper, published in the journal Nature Photonics has the full details.

59 of 94 comments (clear)

  1. Playback at 24 FPS. by grub · · Score: 4, Funny


    If it's possible to play this back at 24 FPS, we can shoot that 3 minute homemade porn we've always wanted!

    --
    Trolling is a art,
    1. Re: Playback at 24 FPS. by Anonymous Coward · · Score: 3, Interesting

      Tell me I got my math wrong: one second of video would take 4650 years to play back at 30fps?

      What is supposed to trigger the start of a recording that can only last such a short time?

    2. Re: Playback at 24 FPS. by Anonymous Coward · · Score: 1

      I did the same calculation as you about the mind boggling amount of samples in a second vs. a 30 fps camera, and got 4650.77 years too. Watching one second unfold over 4650.77 years might just be unfathomable to me. An incredible amount of things has happened in 4000+ years of human history vs. one second.

    3. Re: Playback at 24 FPS. by bondsbw · · Score: 1

      Watching one second unfold over 4650.77 years might just be unfathomable to me.

      No, I'm pretty sure it is exactly unfathomable to any of us.

      --
      All my liberal friends think I'm a conservative, all my conservative friends think I'm a liberal.
    4. Re: Playback at 24 FPS. by michelcolman · · Score: 1

      Would make one hell of an art project, I guess.

    5. Re: Playback at 24 FPS. by RivenAleem · · Score: 2

      My sex-life?

  2. Yeah well... by Morpeth · · Score: 2

    but does it go to 11 ?

    --

    'The unexamined life is not worth living' - Socrates
  3. Re:Four times faster than existing. by bobbied · · Score: 1

    What are we going to see that we couldn't before?

    Audio vibrations in a potato chip bag perhaps? Just saying...

    --
    "File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
  4. Re:human eye by bobbied · · Score: 1

    You remember that "We can recover audio from a video of a potato chip bag" article right? 4.4 Trillion FPS might be overkill for that though.

    --
    "File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
  5. This is really cool technology! by Anonymous Coward · · Score: 5, Funny

    This technology sounds totally cool. I'd like to see them use it to take pictures of somebody viewing the shitty Slashdot Beta site. They could capture the user's extreme boredom as this user waits for the shitty Beta site's page and all its shitty JavaScript and CSS crap to initially load. Then they could capture the trillionth of a second when the person notices that it's the shitty Beta site rather than the Classic site, and the person's anger starts to grow. The photos would progressively show the anger turning into madness, and then finally utter and complete disappointment and despair once the shitty Beta site has finally loaded. The photos could also capture the formation and flow of the very first of many teardrops to cascade down this poor victim's cheeks as the user struggles in vain to read the stories' small text with poor contrast. These trillions upon trillions of frames of total anguish could be examined in excruciating detail, so the awful nature of the Slashdot Beta site could be truly comprehended.

    1. Re:This is really cool technology! by sg_oneill · · Score: 1

      Hahaha if I had mod points ,they'd be all your sir! Classic comment (as opposed to shitty beta comment)

      --
      Excuse the Unicode crap in my posts. That's an apostrophe, and slashdot is busted.
  6. Re:human eye by iamwhoiamtoday · · Score: 2

    That's if you want to watch it in realtime. Things, such as bullet dynamics, railguns, physics labs, so on so forth, would have great use for such a camera.

    FOR SCIENCE!!!

  7. Watching the footage by Anonymous Coward · · Score: 1

    At 4.4 trillion frames a second, playing back one second of footage at 24fps would take over 5,580 years.

    1. Re: Watching the footage by Anonymous Coward · · Score: 2, Funny

      Finally, a cable network with no reruns.

  8. Motion Picture Camera? by QuasiSteve · · Score: 1

    designed a motion picture camera which is capable of ... 450 x 450 pixel pictures

    I guess if you're targeting a 320x240 device, that counts... otherwise, not so much.

    ( 450x450 is still pretty impressive at that frame rate. )

  9. Picoseconds... by mythosaz · · Score: 2

    Picosecond resolution... ....only 30 or so more zeros until we hit one frame per plank time.

    1. Re:Picoseconds... by mythosaz · · Score: 1

      I was talking about wooden boards, you insensitive clod!

      [Also, thanks...]

  10. Some details by JerryLove · · Score: 5, Informative

    First FTA: There's a mention of a previous camera...
    "Back in 2011, researchers from MIT created a high-speed camera that captured light passing through an empty bottle in slow motion by acquiring visual data at one trillion frames a second – to the STAMP cam, more than four times faster than this, even the speed of light could be as stimulating as watching paint dry."

    That's misleading. The camera in 2011 didn't do amazingly high FPS capture. What it did have was very short capture with precise timing. That video of a laser moving through a bottle was actually thousands of successive laser shots. More like stop-motion than video.

    Now this camera I see fewer details on. I do see that one thing it seems to do is to divide a laser with a prism and use the separation to make virtual frames by using different receptors.

    Let me make an analogy. If you took a normal RGB color sensor from a camera, and exposed it, and during that exposure you fired a red flash, then a green flash than a blue flash one after the other. Take your resulting picture and break it into three by color and you have 3 "frames". They appear to be doing this with a large number of wavelengths.

  11. Things that go fast by Earthquake+Retrofit · · Score: 3, Interesting

    It seems to me a camera ilke would be useful for viewing things that happen very quickly, for instance, particle collisions in an atom smasher.

    --
    Fifty years of Yippie! 1968-2018
    1. Re:Things that go fast by BenFenner · · Score: 2

      After thinking about this for a beat, I went from "great idea" to "how in the hell would that work?" pretty quickly.

      I have admittedly not read the article, but if the camera captures photos, are there photons flying around in particle colliders just flying around bouncing off sub-atomic particles all over the place, enough to get a video of it all happening? I get the feeling photons don't interact with the particles much if at all, which is why now they can only see where they end up (trapped or puncture a gold film?) and guess their path with math...

    2. Re:Things that go fast by joe_frisch · · Score: 4, Interesting

      Most particle physics happens on much faster time scales than picoseconds. There is some slower physics but that can generally be measured by looking at the verticies where tracks diverge and calculating the time it too particles to get to those vertices.

      For measuring beams rather than the individual particle collisions we can use transverse deflection structures (a sort of streak-camera on steroids) to get to resolutions of a few femtoseconds.

      The original article is a nice technique, but whether it is the fastest depends on how you define "camera". It is probably the fastest for 2-d images, but there are much faster 1-d imagers.

    3. Re:Things that go fast by penguinoid · · Score: 1

      It seems to me a camera like that would be useful for viewing things that happen very quickly, for instance, particle collisions in an atom smasher.

      Or...
      the period during which a new politician is still honest and competent.
      the attention span of the average teenager.
      the length of time a non-staff member can use Beta before losing all faith in humanity.
      the length of time DRM prevents a non-crippled version appearing on TPB.
      a flash of lightning.

      At 4.4 trillions fps played back at 24 fps, the average lightning bolt would take 1.5 hours. I'm not sure the camera is fast enough to capture any of the others.

      --
      Don't waste your vote! Vote for whoever you want, unless you live in a swing state it won't matter anyways
  12. Re:human eye by RelaxedTension · · Score: 2

    Exactly. But until they capture and show something better than this I say Meh.

  13. 4.4 trillion frames per second? by Trogre · · Score: 1

    Still not enough for the Mythbusters.

    (On a more serious note though, how on Earth do they manage to store even a few microseconds of the footage from this beast?)

    --
    "Nine times out of ten, starting a fire is not the best way to solve the problem." - my wife
    1. Re:4.4 trillion frames per second? by Baloroth · · Score: 4, Informative

      (On a more serious note though, how on Earth do they manage to store even a few microseconds of the footage from this beast?)

      They don't. From the full paper:

      In our proof-of-principle demonstration, the total number of frames was limited to six due to our simple embodiment of the SMD (Supplementary Figs 3 and 4), but can be increased up to 100 by increasing the number of periscopes in the periscope array of the SMD or by using a more complex design (see Methods and Supplementary Section ‘Improvements in STAMP's specifications’)

      You can't just record an indefinite length movie with this thing, you basically need to alter the hardware to record longer segments (since it has different physical elements detect different frames of the signal).

      --
      "None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
  14. Doesn't that come with another problem? by Opportunist · · Score: 4, Interesting

    I mean, the speed of light is 299,792,458,000 Millimeters per second. Maybe I miscalculated something (I always get confused with the way the US names its powers of 10), but doesn't that mean that in 15 frames of this movie, light only moves for about a millimeter? Someone with more background in physics may shed some light onto this (no pun intended), but when you're dealing with stuff SO fast that it approaches the speed of light, isn't measuring and recording subject to the problem that you cannot transport information (and thus also the result of your experiment to the observing camera) faster than said speed of light?

    --
    We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
    1. Re:Doesn't that come with another problem? by fisted · · Score: 1

      I mean, the speed of light is 299,792,458,000 Meters per second.

      FTFY

    2. Re:Doesn't that come with another problem? by glwtta · · Score: 2

      Yup, I'm getting 0.06813 mm per frame.

      Pretty damn impressive, but then again, that thing apparently can only record 6 frames at a time, making the whole article misleading to the point of being complete bullshit.

      --
      sic transit gloria mundi
    3. Re: Doesn't that come with another problem? by Anonymous Coward · · Score: 1

      If your first F isn't "fixed", I'd agree. Count the zeroes, and learn how to spell "metres". xx lol thx.

    4. Re:Doesn't that come with another problem? by Rich0 · · Score: 1

      Well, they don't transport anything anywhere with their solution. But, if you had a technology that could actually offload the data to bulk storage and thus operate "indefinitely" the speed of light would not create any problems whatsoever.

      Just send the data down a wire or fiber optic cable. Sure, the first byte will only travel a few mm before you have to send the second, but there is no law saying that you can only have one packet of data in the cable at a time. Speed of light impacts latency, not bandwidth. If you want to be able to retransmit lost packets/etc then that latency will have a HUGE impact on your buffering requirements at both ends. We're talking transmitting trillions of data points before getting the first ack.

      The reality is that it will be a long time before this technology could be operated for more than a handful of frames at a time. We don't have any technologies with the necessary bandwidth to transfer this kind of data volume quickly. Even if a frame only required a kilobyte of space to store, you're talking about petabytes per second here. Still, the speed of light is not a fundamental limitation on bandwidth.

    5. Re:Doesn't that come with another problem? by thieh · · Score: 1

      The system is at steady state (photons are already at the lens/detector) by the time you start, so when you do, you are taking the photons that are closest to your detectors first, whether they are in front of the lens or next to your detector. What I am interested in is that at trillions of frames per second where do you get hundreds of TB/s to PB/s of bandwidth between camera and storage? I do not recall stuffs you buy at a store go that fast.

    6. Re:Doesn't that come with another problem? by Vitriol+Angst · · Score: 1

      If you have a trillion cameras all operating in sequence such that they are triggered exactly so one activates a trillionth of a second after the other, than you've got exactly the same information via speed of light as just one camera. The only question is how much light each device is sampling. The shorter the time window, the more sensitive the measurement of light.

      But I don't see any problem with speed of light; you are just sampling what hits the sensor at a faster rate.

      I'm just wondering who would sit around for a few weeks to watch a humming bird beat its wings a few times. We need some hyperbolic metaphors so we can comprehend how fast this is...

      --
      >>"ad space available -- low rates!!!"
    7. Re:Doesn't that come with another problem? by jfengel · · Score: 1

      The scale they're trying to capture is far smaller than that: they want to capture molecules moving in a chemical reaction. They're moving much, much less than a millimeter.

      The speed of light isn't the problem there. The entire frame comes in to them at the speed of light, just as in an ordinary camera. The trick is being able to capture this frame and move on to the next one, which they do with a very clever beam-splitting setup. (I haven't gotten all the details yet, but I gather that it's like sending the light beams to multiple cameras at once, each of which takes a very, very short image at a very, very precisely calibrated time. And it doesn't actually have a trillion cameras; it takes frames at that rate, but only a few of them, one per camera.)

    8. Re: Doesn't that come with another problem? by fisted · · Score: 1

      the first F was supposed to mean "failed" ;).
      You're right. Not sure what i was thinking...

  15. Re:Four times faster than existing. by 50000BTU_barbecue · · Score: 1

    You can build a Forward Mass Sensor and use a magnetic cannon to launch an iridium sphere nearby...

    --
    Mostly random stuff.
  16. Re:Four times faster than existing. by wooferhound · · Score: 2

    What are we going to see that we couldn't before?

    Light traveling through a glass bottle

    --
    We are Dead Stars looking back Up at the Sky
  17. Re:Four times faster than existing. by Anonymous Coward · · Score: 1

    Just kidding, I eat smegma!

    Slashdot user sillybilly

  18. Re:Always needing more... by Anonymous Coward · · Score: 1

    Ah, yes. Plank time: the ultimate endpoint for pirate physicists.

  19. Re:Bandwidth and Storage by Noah+Haders · · Score: 1

    420*500*16 bit pixel depth*4.4 trillion frames per second = 1,800 TB/sec. raw, obvously, because i don't know if anything can compress that fast. my ssd can do 500 MB / sec. Is that close?

  20. REALLY SLOW motion... by Kazoo+the+Clown · · Score: 1

    Stretch those femtoseconds into hours, nice...

  21. 4.4 trillion frames per second, and by danknight48 · · Score: 1

    high pixel resolution (450×450 pixels).
    http://www.nature.com/nphoton/...

    Will I need to update my TV to 4K to play it? Enough said.

    1. Re:4.4 trillion frames per second, and by Dogtanian · · Score: 1

      high pixel resolution (450Ã--450 pixels) [..]
      Will I need to update my TV to 4K to play it?

      WTF? 450 x 450 is effectively sub-SD (640 x 480 or 720 x 576).

      Not that this is a criticism if one is shooting a mindboggling trillions of frames per second, but in terms of spatial resolution alone, it'd probably disappoint even on your existing HD set.

      Also, you'll probably complain that the high frame rate makes it look like a cheap soap opera and prefer it when they go back to 24 fps. ;-)

      In all seriousness, trillions of frames per second... WTF?! How does that even work? Modern CPUs operate in the low GHz (i.e. *billions* of ops per second) range, "trillions" is a thousand times faster (THz). Even assuming the use of specialised hardware, how does one even begin to deal with handling frames at that speed?

      Ignoring the amount of capacity that might be required to store such high frame rates- we could assume it might just be recording a very short burst- one still needs to get it off the sensor (or whatever) and store it somewhere in an average of one-trillionth of a second. Even a Class-10 SD card probably isn't going to cut it here(!)

      --
      "Slashdot - News and Chat Sites Deviant". (Click "homepage" link above for details).
    2. Re:4.4 trillion frames per second, and by Rich0 · · Score: 1

      The concept is like a film camera. You capture the data first, and deal with it later. No CPU is involved in the actual collection of each frame, and they capture a VERY limited number of frames as a result.

      When they did bullet time in the first matrix movie they just lined up a whole bunch of cameras and triggered them at very short intervals. Each camera only recorded a single frame. You can collect/process the data later - the critical part was capturing the data. Most high-speed data collection works the same way - extremely fast sampling times, but capture is done into a small buffer and then offloaded later.

  22. Re:Four times faster than existing. by gl4ss · · Score: 2

    audio? you'd need 44 trillion samples/s for capturing audio?

    I suppose you could sell it to audiophiles for inspecting their equipment though still..

    --
    world was created 5 seconds before this post as it is.
  23. Re:Four times faster than existing. by zAPPzAPP · · Score: 1

    But... uhm... you need free light to 'see'.
    So how could you see the light while it is still traveling inside the glass?

  24. 4.4 trillion FPS by Arancaytar · · Score: 1

    1 second / 4.4 trillion * 3E8 m/s = 68 microns.

    That's the distance light covers between frames. Wow.

  25. Atmospheric nuclear testing by Alioth · · Score: 1

    Pity this didn't exist during the days of atmospheric nuclear testing.

  26. finally! by excelsior_gr · · Score: 1

    Finally something that can show us your mom falling on her ass in slow motion!

  27. Peter Jackson by PsyMan · · Score: 1

    Will cream his pants, it can't be long before he will be using this to film the first 12 movies in the long overdue LOTR reboot.

  28. Re:Four times faster than existing. by Anonymous Coward · · Score: 1

    You can notice the effects between two sets of dense metal spheres from gravity using your naked eye (Cavendish did so over 200 years ago), and if you're concerned with precise positioning you wouldn't need a camera, but something that just measures the position really well (e.g. an interferometer would do so to far more precision than you could easily set up with a microscope). And if you are trying to disprove general relativity's statement that gravity travels at c, you don't need cameras with trillions of frames a second, just an oscilloscope with nanosecond level resolution that you can find in a student lab these days. Or if you insist on using a camera, ICCD cameras with frame controls below 1 ns are commercially available off the shelf. More likely you will find the issue isn't the speed of recording devices, which far exceed those requirements, but the difficult in building such a setup that has a low enough noise floor.

  29. Any viewer available? by fabrica64 · · Score: 1

    It will take a long time to see that second!

  30. Re:Four times faster than existing. by ArcadeMan · · Score: 2

    So how could you see the light while it is still traveling inside the glass?

    Magnets. Lots and lots of magnets.

  31. Re:Bandwidth and Storage by rwise2112 · · Score: 1

    420*500*16 bit pixel depth*4.4 trillion frames per second = 1,800 TB/sec. raw, obvously, because i don't know if anything can compress that fast. my ssd can do 500 MB / sec. Is that close?

    So, how much storage are we going to need for the next Hobbit movie?

    --

    "For every expert, there is an equal and opposite expert"
  32. Doesn't that come with another problem? by marcroelofs · · Score: 1

    Red light is 430 trillion hertz. So about 100 red light waves per frame.

  33. Re:Bandwidth and Storage by HornWumpus · · Score: 1

    None at all.

    --
    John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
  34. Re:But, can it... by HornWumpus · · Score: 1

    That's easy. Just ask for a BJ.

    --
    John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
  35. Re:Four times faster than existing. -the answer is by lucien86 · · Score: 2

    There's a very simple test of the general relativity law about the speed of gravity - and it has already failed it spectacularly.. Put simply if gravity moved at the speed of light then the gravitational fields of black holes should collapse in on themselves leaving them externally massless. If you take the simplest classical model of black hole with a central singularity then the inner event horizons require higher and higher escape velocities towards the centre and gravity must jump over the whole lot to escape from the edge. This requires A - an absolute frame for space ruling out a curved space time, and B sets a minimum speed for gravity which is close to FTL Simultaneous - ie nearly 'infinitely' fast.
    This model is not accepted in modern physics - but it is powerful and simple (orders of magnitude simpler than most others), and has no obvious point where it fails. (KISS - Keep It Simple Stupid - Occam's Razor) The real problem of course is that physicists are very stubborn and don't like their favoured solutions being contradicted - even in the face of direct evidence...

    --
    Below the speed of light Special Relativity is one of the most accurate theories in physics - above the speed of light..
  36. Re:Four times faster than existing. by silent-listener · · Score: 1

    The audio vibration in a chip bag was a nice example of high speed pictures and software using this pictures to filter out unseen movements. They work with a speed up to 6000 fps. With higher high speed we can understand / tackle hopeful vibration in constructions and machinery, as combustion engines and air planes.

  37. Re:Four times faster than existing. by MakerDusk · · Score: 1

    Yes, it would be fairly useful for inspections in science and engineering: a way to measure precision, as well as observe rapid chemical reactions. Speaking of rapid chemical reactions, I wouldn't mind seeing a mythbuster style explosion that slow.