Using Missile Tech To See Like Predator

[Artem Litvinovich] wanted to see by heat vision like in the Predator movies. He not only succeeded but went on to see in color, medium-wave IR, short-wave IR, and ultraviolet using a very unique approach since his effort began back in 2009.

He started with a box based on the basic pinhole camera concept. In the box is a physical X-Y digitizer moving a photodiode to collect the thousands of points needed to create a picture. First all he got, due to the high signal amplification, was the 60 cycle hum that permeates our lives. A Faraday cage around the box helped but metal foil around the sensor and amplifier finally eliminated the noise. Now he had pictures in the near infrared (NIR).

Some years passed, he built a 3d printer, and updated the mechanism and the box. Using filters his system now could detect colors and NIR. But that wasn’t enough for [Artem]. As the project continued he added the ability to see ultraviolet and short-wave infrared.

The real challenge was seeing into the medium wave infrared (MWIR). Since these are the sensors used by heat-seeking missiles, the Japanese supplier wanted a ton of paperwork proving he wasn’t building something nefarious… like missiles.

[Artem], who lives in Moscow, turned to a local supplier to obtain sensors often used in Russian missiles. Sensor drift now caused problems and the pinhole no longer sufficed. He added a ZnSe lens in a 3D printed holder and straightened out the drift problem. He would like to tackle long-wave infrared but that requires liquid nitrogen cooling of the sensor. Hey, the missile sensor issue didn’t stop him so we won’t be surprised to see him succeed here as well.

This is a hack of major proportions, taking years to evolve, with multiple learning curves. An impressive feat and a project writeup well worth reading. Given his inventiveness and determination we anticipate he’ll work around the cooling problem before long.

[Our thanks to Nils Hitze for the tip.]

30 thoughts on “Using Missile Tech To See Like Predator

  1. It seems a very impressive build overall. I think the significance might be a bit overstated though. MWIR cameras are not all that rare – certainly not “never seen before by anyone”. I have an uncooled MWIR microbolometer on my desk right now. Same for LWIR – that’s really cheap to obtain. FLIR Lepton cameras are $180 on Digikey.

    I’d be really interested to see if he could build a compressive imager ( with the same detector. That’s a great way to improve the speed of the imager versus scanning.

    1. I’ve headr about it a few days back, and I had the same thought first.
      I guess he meant they are rare and almost never seen, because of the way he is capturing the images.
      The infared/lepton camera they just react to “heat” (they sense the temperature increase/decrease on the sensor). His camera works like any other camera, it’s just the sensor that ca see a certain range of wavelenght.
      I guess, it’s not really the same as juste sensing heat.

      Btw, this feat is really amazing. I wonder why he doesn’t align multiple diode, in order to accelerate the process.

      1. At some point us HAD readers have to stop paying attention to any opinion HAD writers give, excet Bill Heard and various people named Eliot.

        Most importantly, if a HAD writer says something ‘is the most’, ‘is the best’, ‘is brand new’, ‘has never been seen’, maybe that guy who wrote the anti-benchof Filter could write a filter that replaces these terms with more reasonable ones? IE instead of “best ever”, the plugin could replace it with “is a good example of”.

        One note – this article does not claim this guys cool project is ‘never seen’ – you claimed that, apparently?

          1. Spoken like a master of the English language. Tell me, do four syllable words exist inside the cavernous expanse between your ears? Is it true that the best negative description you can come up with stares you in the face when you shave in the morning? Inquiring, socially functional minds want to know.

      2. Sensing the (radiated) heat is obviously a way to sense electromagnetic radiation. It may have a different spectral response and bandwidth than other sensor principles.
        You can also sense visible light thermally, if you want:
        I have a 50W white LED and when I hold my hand in front of it, I sense heat. Although the diode is not overheating. To filter out thermal radiation I used a thick glass plate (which is transparent to VIS but not to M/LIR). I could still sense the “heat” radiation in more or less unchanged intensity. So what i felt was the heat of absorbed visual light on my skin.

  2. wow, physically scanning and digitizing a single photo diode until you build a picture? Obviously only useful for pictures of stationary objects and very low resolution.

        1. There is no loss of resolution by using pinhole. All you lose is intensity of light, which in turn requires large exposure times to acquire the same amount of photons. This requires static objects, otherwise you basically reflect the same surface in different points on the screen, causing blur.

          1. Resolution goes up with hole reduction, until you hit diffraction effect, there is an optimal hole size given screen distance and light color (see wikipedia page for details).

        2. Insane detail here:

          Long story short: as you increase the pinhole-sensor distance, you get better and better resolution, but it’s darker so you need longer exposures or a better sensor. Practically, if you can get the image plane 10-30 cm back from the hole, you get pretty decent images. Many film-based pinholes shoot at the lower end of that range and look goo d.

      1. from the article :
        “The amount of sensitivity needed to pick up the faint light from the pinhole is massive. And we live in the age of AC electricity – an omnipresent hum at 50Hz.”

        about the author:
        “About Artem Litvinovich
        Artem is a hacker and garage tinkerer based in Moscow. His day job is making software for a telecom company. You can check out his projects at

        The pictures are all taken in moscow, so what are you talking about ?

      2. He tells how he tries to get a MWIR diode but the US would not allow it

        “This was my first encounter with ITAR – the “how dare you want interesting stuff?” restrictions. Before then I never realized just how USA, uh…, loves the whole world.”

        Then he says:
        “Fortunately, we in Russia also know how to make heat-seeking missiles, and after some searching I found a local supplier that was more than happy to provide me with a couple of 3.4 μm InAs photodiodes

        And the author of the camera article is not Venkatesh Rao who lives in Seatle but Artem Litvinovich,
        Rao is just the guy who runs the site it seems.

  3. I wonder if you could do something to move the sensor much more quickly? Or put a lens or pinhole in front of it, and move that. Something like a stack of piezos or a voice coil. Obviously you wouldn’t have the range of movement, you’d need to somehow compensate for that. Mirrors or lenses or something. Perhaps have a vibrating mirror shine onto the detector. Use the principle of leverage to get a larger range of motion

      1. That’s the way thermal imaging was done in commercial – and really expensive – cameras, before array sensors were available. But especially with the the pinhole arrangement you can not reduce the exposure time too much. It is like a really high f-stop – very little light.

    1. More quickly == less time to integrate signal, less amplifier noise averaging == more noisy pixels.

      The Engineer’s Dilemma. Damned if you do, damned if you don’t :-)

      1. Yeah but you don’t need infinite slowness. Depends how sensitive his setup is to noise, and how sensitive generally.

        So many things in engineering and science work like the V=IR triangle. More of one, less of the other two. Nice symmetrical universe we live in.

  4. A quick look at Farnell shows that a cheapest photodiode comes at around 0.2€. 10×10 would cost ~20€ and would speed-up the scanning process by 100 times. So why to bother with a single-one? More noise? Well, apply some correlation and filtering…

    1. Wouldn’t a diagonal aragment of the diodes be more efficient? The 10 cm long printhead of the 3D printer I work with does this and gets an amazing resolution on a two cm strip

    2. These are special medium-wave infra-red diodes, that you can only get if you have a license to manufacture heat-seeking missiles, or from the shop on the corner in Moscow. Probably a bit dearer.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.