I'm curious if Single-photon avalanche diode (SPAD) arrays could replace more traditional image intensifiers.
Would be intrigued if they could be more sensitive to light, as you wouldn't need a phosphor screen to translate electrons to photons.
I've not managed to get this paper yet - https://ieeexplore.ieee.org/document/9895178/ but it sounds like it might be possible to make SPAD arrays to capture longer IR wavelengths for thermal imagery.
I can think of some downsides of a SPAD though, as you'd need a display too, which would decrease the frame rate.
Most likely not. SPADs are good for detecting single photons (I have designed electronics for this), but to quench the pixel after every detection you need significant dead-time for the pixel. Single pixel needs to be an array here, size of which defines the dynamic range of the pixel. Each element within the pixel can not be much smaller than the wavelength to be detected. That means you need significant array of arrays to do any meaningful imaging — sensor will be very large.
Thank you for that! Indeed the world has moved on since last time I worked with SPADs (2011). Canon progress you linked seems to be outlined well in several articles published by Kazuhiro Morimoto. Most importantly 3D-stacking the pixel readout with back side illumiation and getting rid of discrete "multiplication wells" which undergo avalanche independently.
So, unsurprisingly, in a completely dark room, with no (visible-range) photons available, even the best "light amplification" night vision goggles can only show you white noise produced by misfiring electrons. Apparently it's hard to come up with light amplification goggles that also work in the infrared range (which CCD sensors have no trouble with)? There are "combined" devices, but if I understood correctly these use two different technologies and then merge the image...
It's confusingly presented in the video, but they already work in the near infrared, up to ~900nm. They're used with IR illuminators and lasers all the time.
There are tubes that work on a wider spectrum, like the Photonis 4G tubes from UV to IR 350-1100nm. But I found them dimmer than the L3Harris in the video.
Infrared is a wide range. Ambient temperature black body radiation (LWIR) is typically detected between 7..14um. Refractive optics that works from there to visible range (380..700nm) is to my knowledge still unsolved problem.
Image intensifier tubes don't care what color the incoming photons carry. Anything within reasonable wavelength/voltage/frequency that can pass and focus on input side of IIT triggers secondary electron release events, a la armor piercing shell causing spalling on an armor plate. Then the released electron hits exit side charged at high voltage and excites good old green or white phosphor paint. Frequency of that release events converts to light intensity on the screen and the user observes it through eyepiece optics.
Standard photomultipliers are single pixel vacuum tube, nightvision IIT devices are hex array of pixels. They are built on microchannel plate technology; a bundle of optical fiber, twisted, impregnated in glass into a sushi roll, silced angled, etched, electroplated, stacked, to create a circular glass plate filled with couple layers of microscopic conical voids, each working to same effect as multi-stage PMTs when charged.
I suppose, in theory, all lights from long-wave IR to X-rays can be all handled by a single IIT tube, but in reality they can't, and thermal cameras that resolves heat images works in a completely different principle; instead of using vacuum tube technology that excite on incoming photons, they receive microscopic changes in temperatures caused by incoming light focused by IR-transparent lenses, projected onto an array of individual heat sensor devices(microbolometers) that are MEMS fabricated on top of typically Sony CMOS sensors.
Combining an IIT with a CMOS camera should be possible and I guess will improve contrast/minimum light level necessary, but the tube already sees near IR that it reasonably can, and also, because the light at the exit/user side of IIT is just luminescence from phosphor, adding heat-sensitive detectors there won't yield much.
IMO, if you just want to maybe buy and use an NVG, videos from (rich/financially irresponsible)gun people such as Hop(Hoplopfheil) or InRangeTV are more brief and potentially useful. Anything beyond, I found on Google(pre-covid).
They totally avoided the obvious "light amplification but with an IR-pass filter in front of the photocathode and/or a photocathode that is particularly IR-sensitive".
I say obvious because I'm convinced these have been in production for decades. The things the security apparatus considers secure is laughable and a sign that they could be easily fall behind.
To each their own. "Black silicon" and other high-sensitivity CMOS cameras are way better in starry nights, from what I understand, but conventional night visions are more useful in moonless/starless nights, as well as poorly lit indoors. Assuming you have any reasons to "operate" under starless skies.
Both of the above really require active illumination to be effective. As long as you're not worried about other people with night vision, that's not a problem.
Above that... it gets expensive and complicated very quickly.
I enjoy the veritasium youtube channel but it's important to remember that videos like this are enabled by the military as recruiting and propaganda aids. The same is true of channels like Smarter Every Day when they partner with the coast guard or military.
Calling this propaganda is a bit much in my opinion. Of course the military has an interest in creating a video like this but I never felt that this video was an ad.
Derek has always been very ethical about his sponsorships, never going for NordVPN or shitty mobile games but services that promote science (like Nebula or Brilliant).
Absolutely. I always wonder how these videos actually get made. Did the channel reach out to the military, or did some military PR arm reach out to the channel and ask if they were interested in showcasing night vision tech? Hard to imagine anything other than the latter, which is fairly insidious behavior.
For a short time, I had access to photomultiplier night vision hardware when I was in the army in ~2016. I remember spending at least a few hours just looking up at the night sky, seeing it in a way I never had before. Super cool experience, would recommend. Too bad the hardware is so expensive!
I remember having to pee during an exercise while we were under strict light discipline. We were equipped with NVGs, and like Derek experienced, your field of view and depth of vision is very limited. This isn't such a big problem unless you want to drive a car, or like me, use the toilet. I started peeing a good meter in front of the toilet, making a huge mess. The lieutenant was not very impressed.
I haven't watched the whole video yet, but I own a very high-quality PVS-14. I've mostly solved the depth of field issue for myself with an iris in front of the objective lens.
There is a tradeoff to make between available light and DoF, but outdoors I can turn it to about 1/3 open and see clearly from ~2' to infinity.
I found a well-made iris mechanism on eBay, designed for microscopes. It fits the threads on the objective lens exactly, and has another set of threads on the outside to which a sacrificial lens can be added for protection if desired. I'm really happy with it.
I have a simple but very high-quality night vision setup that I regularly use. Working remotely, my sleep schedule is often pretty weird, and as a result I sometimes find it difficult to get outside and exercise during the day. I bought my stuff to let me go for walks at night.
I have a single PVS-14, but it's an Elbit third-gen white phosphorus tube with a FOM of ~2400. I also picked up a Crye "Night Cap" soft head mount for it. It's a little less practical in that it takes a bit longer to don, but it's a solid mount and doesn't look quite as intimidating as a helmet IMO.
In total I have about $4,200 in my head-mounted setup.
One thing about the video that stood out to me - the video shows a lot more optical noise than I actually see, especially if I'm outdoors or using active illumination. That level of noise is only visible for me on overcast nights with a new moon.
22 comments
[ 0.27 ms ] story [ 58.1 ms ] threadWould be intrigued if they could be more sensitive to light, as you wouldn't need a phosphor screen to translate electrons to photons.
I've not managed to get this paper yet - https://ieeexplore.ieee.org/document/9895178/ but it sounds like it might be possible to make SPAD arrays to capture longer IR wavelengths for thermal imagery.
I can think of some downsides of a SPAD though, as you'd need a display too, which would decrease the frame rate.
I just came across this - https://petapixel.com/2023/04/03/canons-new-sensor-enables-l... which sounds interesting.
There are tubes that work on a wider spectrum, like the Photonis 4G tubes from UV to IR 350-1100nm. But I found them dimmer than the L3Harris in the video.
Standard photomultipliers are single pixel vacuum tube, nightvision IIT devices are hex array of pixels. They are built on microchannel plate technology; a bundle of optical fiber, twisted, impregnated in glass into a sushi roll, silced angled, etched, electroplated, stacked, to create a circular glass plate filled with couple layers of microscopic conical voids, each working to same effect as multi-stage PMTs when charged.
I suppose, in theory, all lights from long-wave IR to X-rays can be all handled by a single IIT tube, but in reality they can't, and thermal cameras that resolves heat images works in a completely different principle; instead of using vacuum tube technology that excite on incoming photons, they receive microscopic changes in temperatures caused by incoming light focused by IR-transparent lenses, projected onto an array of individual heat sensor devices(microbolometers) that are MEMS fabricated on top of typically Sony CMOS sensors.
Combining an IIT with a CMOS camera should be possible and I guess will improve contrast/minimum light level necessary, but the tube already sees near IR that it reasonably can, and also, because the light at the exit/user side of IIT is just luminescence from phosphor, adding heat-sensitive detectors there won't yield much.
IMO, if you just want to maybe buy and use an NVG, videos from (rich/financially irresponsible)gun people such as Hop(Hoplopfheil) or InRangeTV are more brief and potentially useful. Anything beyond, I found on Google(pre-covid).
1: https://www.youtube.com/watch?v=sHflvwNJ3cs
2: https://www.youtube.com/watch?v=F3KftiMHYVU
I say obvious because I'm convinced these have been in production for decades. The things the security apparatus considers secure is laughable and a sign that they could be easily fall behind.
https://kottke.org/17/04/incredible-low-light-camera-turns-n...
and the quality of nightvision is much better than the one shown in this video. Or am I missing something?
The goggles you see folks wearing on their heads is analog amplification, so has zero lag.
There are use-cases for both.
The cheapest would be an IR camera, a small display, and a lens. Here's a 3D printed kit, with links to parts and communications channels for makers: https://www.tindie.com/products/justcallmekoko/bpnvg-3d-prin...
A step up from that would be the lower end commercial offerings like the Sionyx Aurora: https://www.sionyx.com/collections/aurora-night-vision-camer...
Both of the above really require active illumination to be effective. As long as you're not worried about other people with night vision, that's not a problem.
Above that... it gets expensive and complicated very quickly.
Derek has always been very ethical about his sponsorships, never going for NordVPN or shitty mobile games but services that promote science (like Nebula or Brilliant).
There is a tradeoff to make between available light and DoF, but outdoors I can turn it to about 1/3 open and see clearly from ~2' to infinity.
I found a well-made iris mechanism on eBay, designed for microscopes. It fits the threads on the objective lens exactly, and has another set of threads on the outside to which a sacrificial lens can be added for protection if desired. I'm really happy with it.
I have a single PVS-14, but it's an Elbit third-gen white phosphorus tube with a FOM of ~2400. I also picked up a Crye "Night Cap" soft head mount for it. It's a little less practical in that it takes a bit longer to don, but it's a solid mount and doesn't look quite as intimidating as a helmet IMO.
In total I have about $4,200 in my head-mounted setup.
One thing about the video that stood out to me - the video shows a lot more optical noise than I actually see, especially if I'm outdoors or using active illumination. That level of noise is only visible for me on overcast nights with a new moon.