How thermal imaging works and how to read the specs

The principle: heat, not light

All objects – living or not, even ice – emit thermal infrared radiation, and the amount depends on their temperature. A thermal camera captures this radiation and turns it into an image in which warmer areas look different from cooler ones. Ambient light plays no part, so thermal works in total darkness.

Most civilian thermal devices work in the long-wave infrared band (LWIR, roughly 8–14 µm) and use an uncooled microbolometer sensor. Two practical consequences:

• Ordinary glass blocks IR radiation in this band. You can’t look through a car or house window with thermal – you only see the temperature of the glass itself. That’s why thermal lenses are made from other materials, most commonly germanium.
• Thermal is not night vision. Tube-based night vision amplifies residual visible and near-IR light; thermal “sees” temperature differences. Each has different strengths – see Clip-on vs. handheld vs. fusion.

Sensor resolution: 384, 640, 1280

Resolution is the sensor’s pixel count, for example 384×288, 640×480 or 640×512. There are also 1024×768 sensors, and 1280×1024 has been commercially available since 2022. More pixels means that, with the same lens and pixel pitch, you get a wider view at the same level of detail – or more detail across the same view.

Higher resolution doesn’t automatically mean a “better device”. A 384 sensor tends to be more compact and power-efficient and can be entirely adequate at short to medium range. Always read resolution together with pixel pitch and the lens.

Pixel pitch (µm)

Pixel pitch is the distance from the centre of one pixel to the centre of the next, given in micrometres (µm). Common values are 12 µm and 17 µm.

• A smaller pitch (e.g. 12 µm) means denser pixels and, at the same resolution, a physically smaller sensor. The same magnification and range can be reached with a shorter lens, so the device can be smaller and lighter.
• A larger pitch (e.g. 17 µm) gives each pixel more area to collect radiation. Pulsar states that in modern sensors, the larger the pixel, the higher the sensitivity (the lower the NETD).

NETD (mK): sensor sensitivity

NETD (Noise Equivalent Temperature Difference) is the smallest temperature difference the sensor can distinguish from its own noise. It’s given in millikelvin (mK); 50 mK = 0.05 K. A lower number means a more sensitive sensor and a more readable image, especially in low-contrast conditions – fog, rain, humid air, game in wet grass.

Be careful when comparing: NETD is measured at a specific object temperature (typically 30 °C) and changes with it. Some manufacturers also state other test conditions, such as the lens f-number. Figures from different datasheets are therefore not always directly comparable.

Lens (mm) and field of view

The lens focal length (e.g. F19, F35, F50 – in millimetres), together with the sensor, sets the field of view (FOV) and base magnification. The rule of thumb: a longer lens gives a narrower field of view and more detail at distance; a shorter lens gives a wider view that’s better for scanning and moving through terrain.

You can work out the horizontal FOV yourself: sensor width = pixel count × pitch; FOV = 2 × arctan(sensor width / (2 × focal length)). Examples (our calculation, 12 µm pitch):

The table also shows something important: with the same pitch and lens, each pixel covers the same angle. A 640 sensor with an F35 lens therefore puts the same number of pixels on a target at a given distance as a 384 sensor with F35 – it simply shows a wider view.

Range: detection, recognition, identification (DRI)

Manufacturers often quote a “detection range” in metres. It is usually based on the Johnson criteria: detection (something is there) needs at least 1.5 pixels across the target, recognition (what it is) at least 6 pixels, and identification at least 12 pixels. Detection range is therefore always much longer than the distance at which you can reliably tell what you’re looking at – and humidity, fog or rain shorten it further.

Try how lens and sensor translate into range in our thermal range calculator.

Refresh rate (Hz)

Refresh rate is how many times per second the image updates. When you follow movement – running game, fast panning – smoothness matters; Pulsar describes 50–60 Hz as optimal for smooth motion in hunting use. Keep in mind that processing latency also affects how responsive the image feels, not just the Hz figure.

LRF and other features

LRF (laser rangefinder) is a built-in laser rangefinder that measures the distance to a target. It helps because distances are easy to misjudge in a thermal image. Other things worth comparing: colour palettes, recording, battery life and external power, water resistance and operating temperature range.

You’ll also come across calibration: an internal shutter briefly covers the microbolometer so the device can even out the sensor background. The image freezes for a fraction of a second while this happens – that’s normal.

Observation vs. aiming

The same sensor can sit in a handheld observation monocular, in a clip-on in front of a day scope or in a riflescope. Technically these are similar devices, but legally they can be treated differently: rules for observation devices and for devices used to aim a firearm vary by country, and sometimes by purpose (hunting, property protection). Thermal cameras can also fall under EU dual-use rules when exported outside the EU. See our overview at Night vision and thermal laws in Europe.

The information on this page is general, is not legal advice, and laws change. We cite the source and the date we checked it. Before buying or using a device, verify the rules with the relevant authority or hunting association. Last checked: 9 Nov 2026.

Checklist: how to read a thermal datasheet

1. Read resolution and pixel pitch together (e.g. 640×512, 12 µm).
2. Only compare NETD figures measured under the same conditions; lower = more sensitive.
3. Check the field of view from lens and sensor – does it suit how you’ll use it (scanning vs. distance)?
4. Treat detection range as an upper estimate; recognition and identification are much closer.
5. Check the refresh rate if you’ll be tracking movement.
6. Check battery life and operating temperatures for your conditions.
7. Find out whether the device complies with the rules in your country for your intended use.

If you’re looking for a specific device, see our comparison of thermal device types or thermal imaging in the NGS shop – the thermal devices we offer are tested and supported by our lab.

Sources

• Axis Communications: Thermal cameras (chapter 5) – IR emission as a function of temperature, LWIR band, microbolometers, NETD, Johnson criteria, focal length and field of view, germanium optics – axis.com (PDF) (checked 9 Nov 2026)
• Pulsar: Can I use a thermal imager to see through glass? – glass blocks 8–14 µm IR – pulsarvision.com (checked 9 Nov 2026)
• Pulsar: Thermal sensor guide – pixel pitch, NETD, 384×288, 640×480, 1024×768 sensors – pulsarvision.com (checked 9 Nov 2026)
• Pulsar: Pixel pitch explained: 12 µm vs 17 µm – pulsarvision.com (checked 9 Nov 2026)
• Pulsar: Thermal imaging real-time latency and lag explained – refresh rate – pulsarvision.com (checked 9 Nov 2026)
• Infrared Training Center: What does sensitivity (NETD) mean? – infraredtraining.com (checked 9 Nov 2026)
• Soldier Systems Daily: first commercially available 1280×1024 thermal weapon sight (2022) – soldiersystems.net (checked 9 Nov 2026)
• Pulsar Krypton 2 FXG50 – Official manufacturer manual (external link, third party – NGS is not responsible for its content) (checked 9 Nov 2026)
• Regulation (EU) 2021/821 setting up a Union regime for the control of exports of dual-use items – EUR-Lex (checked 9 Nov 2026)