Infrared Thermography Guide: SWIR, MWIR and LWIR Imaging

Infrared thermography measures radiation, but not every infrared band measures heat. SWIR cameras (roughly 900–1700 nm) mostly record reflected light, so they behave like visible cameras that can see through silicon, smoke and haze. MWIR cameras (3–5 μm) and LWIR cameras (8–14 μm) record emitted radiation, which is what makes true temperature measurement possible. Choose SWIR for material inspection, moisture detection and starlight environments; choose MWIR for hot targets at long range; choose LWIR for ambient-temperature thermography such as electrical, building and industrial safety inspections.

Key Takeaways

• Thermography is not one band. SWIR images reflected light while MWIR and LWIR image emitted heat, so a SWIR camera cannot measure temperature the way a true thermal camera does.
• MWIR cameras detect radiation in the 3–5 μm band, and under moderate-to-high humidity conditions the MWIR band generally offers better atmospheric transmission than the LWIR band, making MWIR well suited for detecting engines and flares and for long-range surveillance.
• LWIR covers 8–14 μm, the window where a 300 K scene radiates most strongly, making it the default choice for non-contact thermometry, building surveys and industrial safety inspections.
• SWIR wins where heat fails. Silicon turns transparent above about 1100 nm and water absorbs strongly near 1450 nm, so SWIR suits material inspection and moisture detection.
• AttosTek covers 1–14,000 nm with 100% factory-tested, CE/FCC-certified cameras and small-batch customization, so one supplier can support a program that mixes all three bands.

What Infrared Thermography Actually Measures

Every infrared image starts with photons, but the source of those photons changes the meaning of the picture. Two mechanisms matter. Reflected light bounces off a surface after arriving from the sun, the sky, a lamp or a laser; solar reflectance is the daylight version of that signal, and it is what makes non-thermal imaging possible at all. Emitted radiation leaves the object itself because everything above absolute zero radiates according to its temperature. Thermography in the strict sense means measuring that second component and converting it into a temperature reading.

The split happens around 2–3 μm. Below that, in the short-wave infrared, reflected light usually dominates and the image carries material information rather than thermal information. Above roughly 3 μm, emitted radiation takes over, and the atmosphere offers two useful windows: 3–5 μm for MWIR cameras and 8–14 μm for LWIR cameras. Wien’s displacement law explains the division of labour. A 300 K scene peaks near 10 μm, while a 600 K target peaks near 4.8 μm, so ambient-temperature thermography lands in LWIR and hot-target work lands in MWIR.

Accurate temperature measurement adds one more requirement: emissivity. A shiny metal surface reflects more than it emits, so a thermal camera reading it directly will under-report temperature. Coatings, reference emitters and calibration curves solve this, and they are the reason a thermography program is judged by its measurement setup, not by camera resolution alone.

SWIR Imaging: Reflected Light, Not Heat

Short-wave infrared spans roughly 900–1700 nm, and it is best understood as non-thermal imaging. A SWIR camera collects reflected light exactly as a visible camera does, which produces familiar-looking, high-resolution images that operators read without special training. Under daylight, solar reflectance dominates and delivers crisp contrast. At night, the same sensor can still work in starlight environments, because the night sky emits enough short-wave infrared for long exposures to build a usable image. Neither case measures temperature, which is the defining trait of non-thermal imaging.

Three material behaviours make SWIR valuable in industry. Silicon becomes transparent above about 1100 nm, so wafer and semiconductor inspection can look through the surface instead of at it. Water absorbs strongly near 1450 nm, which turns moisture detection into a high-contrast measurement rather than a subtle one. Many plastics, pigments and coatings separate clearly across the short-wave infrared, which is why material inspection and sorting lines often pair a SWIR camera with a hyperspectral camera; the hyperspectral camera resolves narrow spectral bands that a single SWIR channel would average out.

AttosTek builds SWIR cameras from 400–1700 nm in the A-Series, 900–1700 nm in the B-Series and up to 900–2200 nm in the C-Series, 300-1700 nm in the Q-Series, using InGaAs, CQD or MCT detectors with cooling from uncooled to -80°C and interfaces covering USB3, GigE, CoaXPress, CameraLink and MIPI. You can compare the full line on the SWIR camera range.

MWIR Cameras: The Hot-Target Band

MWIR cameras cover 3–5 μm, where emitted radiation from hot objects carries far more signal per degree than it does at longer wavelengths. That physical advantage shows up as range: under comparable conditions, MWIR usually has better atmospheric transmission performance than LWIR, which is the number that decides most long-range projects. Shorter wavelengths also allow smaller optics for the same angular resolution, and the band is less sensitive to humidity and atmospheric attenuation than the long-wave window.

A representative cooled core, the MWIR1503TCL, uses a T2SL detector at 640 × 512 with 15 μm pixels, sensitive from 3.7–4.8 μm, paired with a linear Stirling cooler. It reaches operating temperature in about 7 minutes, holds NETD at ≤22 mK, draws under 12 W at steady state, weighs ≤420 g, keeps the bad-pixel rate below 0.3% and is rated for 10,000 h MTTF across -40 to +60°C. Output is CameraLink at 14-bit or 8-bit with RS422 control, and on-board processing covers AGC and bad-pixel replacement. AttosTek also lists MCT and InSb options plus resolutions up to 2048 × 2048 for programs that need more pixels across the same field.

Typical jobs for MWIR cameras are engine test stands, exhaust and flare monitoring, furnace and kiln imaging, long-range surveillance, unmanned payloads and traffic imaging through smoke. Unlike SWIR, MWIR cameras need no illumination and no solar reflectance to produce a usable image at night. For a side-by-side view of cooled cores and complete units, see the MWIR camera family.

LWIR Imaging: Ambient-Temperature Thermography

LWIR covers 8–14 μm, the atmospheric window where objects near room temperature radiate most strongly. A scene at 300 K therefore produces a strong signal with no illumination at all, which is why LWIR is the band people mean when they talk about thermography for temperature measurement. It is also the cheapest route into thermal imaging, because uncooled VOx microbolometers work well at these wavelengths and start in seconds on a few watts. It is the band MWIR cameras trade away when the target is not hot enough to stand out at 3–5 μm.

The application list follows directly from that physics: electrical panel and switchgear surveys, building envelope and insulation diagnostics, mechanical overheating checks, fire search and rescue, perimeter monitoring in darkness, and medical screening. Industrial safety inspections rely on it heavily, because a thermal survey finds a loose connection or a failing bearing before it becomes a failure. AttosTek’s LWIR line spans 640 × 512, 1024 × 768 and 1280 × 1024 resolutions with GigE, USB3.0, CameraLink and MIPI interfaces, plus cooled MCT, T2SL and HOT variants for finer thermal resolution or longer integration times.

The limits are physical, not commercial. Long wavelengths need larger apertures for the same detail, lenses depend on special infrared materials with high-efficiency anti-reflection coatings, and heavy rain, fog or window glass will block the signal. Browse the uncooled and cooled LWIR cameras to compare formats.

Choosing the Right Band: A Practical Comparison

Four questions settle the decision: what is the target temperature, how far away is it, do you need a calibrated temperature number, and what can the platform carry in power and weight. Answer those, and the band usually picks itself. The table below compresses the comparison for common industrial and scientific jobs.

Two combinations appear again and again in real projects. Industrial inspection lines pair a SWIR camera for material and moisture contrast with an LWIR camera for thermal verification, because the two bands answer different questions about the same part. Long-range security programs often run LWIR for wide detection and MWIR cameras for identification at distance, since the shorter band keeps detail where the long-wave image turns into a blob. Starlight environments are the exception where a security team may prefer SWIR, because reflected light from the night sky needs no thermal contrast at all. Hyperspectral and multispectral add-ons sit naturally on the SWIR side, where spectral signatures are strongest.

Building a Multi-Band Program with AttosTek

Most imaging problems do not respect a single band, which is why supplier breadth matters as much as any one specification. AttosTek covers the wavelength range from 1 to 14,000 nm across UV-VIS, SWIR, MWIR and LWIR cameras, lenses and optical systems, so a team can standardise on one source for an entire multi-band program instead of stitching together three vendors. That breadth matters most on lines that combine material inspection, moisture detection and industrial safety inspections in one workflow.

Three commitments support that range. Every product is 100% factory tested and fully CE/FCC certified. Custom imaging solutions are tailored to the application even for small orders, covering sensor type, spectral response, cooling, interfaces and housing. And 7×24 expert support stays available through integration, calibration and troubleshooting. Each camera ships with AttosView control software and an SDK supporting C/C++, C#/VB.NET, Python and Java, with LabVIEW, MATLAB, Micro-Manager, DirectShow and TWAIN compatibility for existing setups.

For worked examples and application footage across the three bands, the AttosTek YouTube channel at @attostek.official is a practical starting point. When a standard model does not fit your target temperature, pixel count or mechanical envelope, describe the scene to the AttosTek customization team and they will come back with a configuration and a CIF quotation.

Frequently Asked Questions