Night Vision Technology Explained, From Full Moon to No Moon

Night vision technology is any way of seeing a scene when there isn't enough light for your eyes. There are really only three ways to pull that off: amplify the little light that's left (image intensifier tubes), bring your own invisible light and film it (digital night vision with an infrared illuminator), or skip light entirely and read heat (thermal). A fourth, short-wave infrared, exists too, but it's still specialist gear.

What is Night Vision Tech?

Most explanations stop at that list. The more useful question is how dark each one can go before the picture falls apart, and what you actually see at that point. So this guide puts every type on the same light scale, from a full moon down to an overcast night with no moon at all, then checks the digital part against a real night out with our CIGMAN CNVPRO binoculars in the French Alps.

Why Our Eyes Give Up in the Dark

Once the light drops, your eyes switch over to rod cells. Rods are sensitive enough to respond to a single photon, but they only see in grey, at low resolution, and they're slow, so anything moving turns into a smear [1]. There's no built-in amplifier to turn a dim scene up.

I went through dark adaptation and red light in more detail in the post on using red light with your digital night vision, so I'll keep this part short.

Animals got around the same limit in two ways, and night vision gear copies both ideas. The great horned owl's tube-shaped eyes throw a bigger image onto the retina, so each receptor collects more light [2]. Cats and deer have a tapetum lucidum, a reflective layer behind the retina that bounces unused light back through for a second pass [3]. Tubes and camera sensors chase the same goal: catch more of the light that's there. Thermal does something no eye does and reads heat instead.

How Dark Can Each Type Go?

Light at night drops off fast. A full moon puts roughly 0.1 to 0.3 lux on the ground, a clear starry night with no moon about 0.001 lux, and an overcast moonless night 0.0001 lux or less [5]. That's a thousand-fold drop between a bright moon and a cloudy night, and each type of night vision gives up at a different point along the way.

Table 1: How dark each type of night vision can go

Two things stand out. Thermal and IR-lit digital don't care about the moon at all, because neither depends on light that's already in the scene. And the types that do, tubes and digital with the IR off, both run into trouble below a quarter moon. Chrzanowski's review notes that even military tubes still perform rather poorly under overcast starlight, and only the most sensitive ones keep working there [6].

What We Saw Testing Digital Night Vision on a Moonless Night

Our tester Tom Fortune took the CNVPRO to the bottom of a ski slope in the French Alps on an evening that started overcast and turned to rain. There was no moon, and apart from a faint glow from an apartment block nearby it was pitch black. He used the ski lift as a set of distance markers: a snow cannon at about 70 m, the first lift tower at about 150 m, and a mountain hut past 200 m. The full test with photos and video is in a separate post.

In low-light color mode the picture was grainy and hard to focus. He could still tell the snow cannon was orange, which is the kind of detail that helps you tell a brown animal from a grey rock. Then he pressed the IR button, and the grey, grainy screen turned into a sharp black-and-white image.

Table 2: What the CNVPRO showed with IR on, no moon, light rain

The rain cut the range at the far end and put bright streaks across the screen, but everything within 150 m stayed readable. On another night out, light fog drifted in and washed out the color mode, while the IR view still reached down a dark road. That same CNVPRO review found color mode didn't work at all in a field away from the streetlights, and that the digital zoom gets softer the further you push it.

That lines up with the light table. Below starlight, a digital unit is really an IR-lit camera, and how far you can see is how far its IR beam reaches.

Shop the CNVPRO

CIGMAN CNVPRO Night Vision Binoculars

• Starlight-grade Sony CMOS sensor, native 4K video and 52MP photos.
• 4-level IR illuminator, 25mm F/0.8 lens, 8x digital zoom.
• Up to 18 hours on a 5100mAh battery, IP54, Type-C charging.

How Digital Night Vision Gets a Picture Out of the Dark

For decades, night vision meant image intensifier tubes, the green, grainy footage from war movies. They're sensitive, but bright light can damage a standard tube, and the tube itself can't record anything [6]. Digital night vision swaps the vacuum tube for a camera sensor and a small screen, which brings recording, zoom and daytime use into the same device.

Tube Generations in One Paragraph

Since tubes are the benchmark everything else gets compared to, here's how their generations break down. Gen 2 tubes arrived in the 1970s and added a microchannel plate, a glass disc full of tiny channels that multiplies electrons before they hit the phosphor screen. Gen 3 went into production in the 1980s with a gallium arsenide photocathode, about three times more sensitive in the near infrared than Gen 2's [6]. There's officially no Gen 4: the filmless and thin-film tubes sold as Gen 3+ are still Gen 3 under the US definition, and a good modern Gen 2+ tube can match an ordinary Gen 3 on signal-to-noise [6]. For the full timeline and device types, see our guide to night vision devices.

Why Newer Sensors See More in the Dark

Inside a digital unit, the sensor sets the floor. In an older front-side illuminated (FSI) chip, the metal wiring sits above the light-sensitive layer and shades part of it.

Back-side illuminated (BSI) sensors flip the silicon so light reaches the photodiodes first, which raises quantum efficiency, the share of incoming photons that actually get counted [4]. That's why spec sheets now push starlight sensors; the CNVPRO uses a starlight-grade Sony CMOS. A better sensor still only lowers the floor, though. Chrzanowski's review of night vision technology puts the point where camera sensors start losing image quality at around 10 millilux, roughly a quarter moon [6]. Below that, the IR illuminator does the heavy lifting.

What the Processor Does to the Picture

Once the sensor has a signal, the image processor has to deal with noise. With only a few photons arriving, each pixel's reading jumps around at random, and that shows up as the crawling grain you see in low-light color mode. The processor turns the signal up, compares neighboring pixels to strip out grain without smearing edges, and evens out contrast so a streetlight doesn't blow out the rest of the frame.

You can see where that runs out. In the CNVPRO review, pointing at distant mountains in color mode gave a grainy picture with a purple tint, and switching to IR cleaned it up. The processor can't invent detail that never reached the sensor. More light can, and in the dark that light comes from the IR.

How Thermal Sees Heat Instead of Light

When there's no light left to work with, deep in a cave or under thick cloud, every light-based type runs out of signal. Thermal doesn't need any. Everything warmer than absolute zero gives off infrared energy, warmer things give off more of it, and at everyday temperatures a lot of it falls in the long-wave band around 8 to 14 µm, which is the band most thermal sensors are built for [7]. That's why a rabbit in tall grass lights up on the darkest night.

How Does the Camera Sense Heat?

Thermal cameras use a sensor called a microbolometer. Instead of counting photons like a CMOS chip, it works more like a grid of tiny thermometers. Each pixel is coated with a material such as vanadium oxide that warms up slightly when infrared hits it [7].

That temperature change shifts the material's electrical resistance. The camera reads those changes many times a second and builds a moving image out of heat differences rather than light and shadow.

The Spec That Matters More Than Resolution

NETD (noise equivalent temperature difference) is the smallest temperature difference the sensor can pick out, and lower is better. Low-cost microbolometers typically land somewhere between 50 and 500 mK [7]. That spread is why two thermal cameras with the same resolution can look very different on a cool, damp night, when the grass, the trees and the animal you're after are all close to the same temperature. For side-by-side footage, see thermal vs night vision imaging.

Why Does the Image Freeze and Click?

Thermal users are often thrown by a mechanical click followed by a split-second freeze. That's non-uniformity correction. Each pixel's response drifts over time, partly because the camera's own housing warms up and cools down, and the image slowly picks up a fixed grainy pattern [9].

To fix it, the camera closes a small shutter in front of the sensor so every pixel sees the same blank surface, measures how each one responds, and uses that to cancel the drift [9]. After that, the differences you see come from the scene, not from the camera itself.

Which Type Does What?

Here's how the three main types compare once you take them outside. Tubes glow green because of the phosphor screen, and we covered why in night vision colors explained.

Table 3: Image intensifier, digital and thermal night vision compared

The window row trips a lot of people up, especially anyone hoping to watch the yard from indoors. We tested it in can night vision see through glass.

What Is Actually New in Night Vision Technology

Most of the recent change is in software and in combining sensors, not in a new kind of tube. It helps to split what you can buy today from what is still mostly in research.

Already in Consumer Devices

• Onboard image processing: noise reduction and contrast balancing, the part of the picture a tube can't touch.
• Recording, photos and digital zoom, which Chrzanowski lists among the built-in advantages of digital devices [6].
• Adjustable IR. The CNVPRO has four IR levels, so you can turn it down up close and up for distance.

Fusion

Fusion overlays a thermal image on a light-based one, so a warm animal stands out against trees and fence lines you can actually see. The catch is that the two channels see the target from slightly different positions, and that offset changes with distance, so the images need processing to line up [6]. We went through what that means for civilian buyers in fused vs digital night vision.

Still Mostly in Research

Automatic target recognition, where the device draws a box around a person or animal and labels it, is one of the things digital imaging makes possible [6]. It isn't a feature of the consumer binoculars we've tested, so read AI on a spec sheet as image cleanup unless the maker spells out more.

Short-Wave Infrared (SWIR)

SWIR cameras usually use indium gallium arsenide (InGaAs) sensors, which pick up light from about 900 nm to 1.7 µm, beyond where silicon stops [10]. The interesting part for night use is nightglow: the upper atmosphere gives off a faint glow in that band, and measurements by Adomeit and Krieg found about ten times more light in the SWIR than in the visible band on moonless nights [8]. For now these are specialist cameras rather than something you'll find in consumer binoculars.

Conclusion

So the real difference between these technologies is where the light comes from. Tubes amplify whatever light is there, and only the most sensitive ones cope with an overcast, moonless sky. Digital units film the scene and bring their own IR, which gets you color when there's some light around, recording, and clear detail out to the reach of the beam, about 150 m in our rainy test. Thermal ignores light and reads heat, so warm things stand out fast, but with less detail than a light-based image.

If you want to identify what you're looking at and keep a recording, start with digital. If you mainly need to spot animals or people in thick cover, look at thermal. If you need to see passively in near-total darkness without any IR beam, you're looking at a high-end tube.

BUY FROM CIGMAN

Your First Choice For Night Vision Optics