How Do Infrared Cameras Work: A Plain-Language Guide
Infrared cameras work by detecting heat-based radiation that every object above absolute zero emits, then converting that signal into a visible image. The physics underneath is straightforward once it is unpacked: warm objects emit electromagnetic radiation at wavelengths longer than visible light, a specialized detector array measures that radiation, and processing electronics convert the measurements into a false-color image that the user can interpret. This guide is the physics-focused companion to the device-focused overview — it walks through what infrared radiation is, how the microbolometer detector works, and how the camera converts incoming heat into the pseudo-color images that home inspectors and homeowners see on a screen.
What infrared radiation is
Visible light is a narrow slice of the electromagnetic spectrum, spanning roughly 380 to 750 nanometers in wavelength. Beyond the red end of the visible band lies the infrared region, which extends from about 750 nanometers to one millimeter. Infrared is invisible to the human eye but carries energy that we feel as heat.
Every object above absolute zero (0 Kelvin, or -273.15 Celsius) emits infrared radiation. The amount of radiation and its spectral distribution depend on the object’s temperature. Cooler objects emit less radiation at longer wavelengths; hotter objects emit more radiation at shorter wavelengths. This relationship — described by Planck’s law and the Stefan-Boltzmann law — is the physical foundation of thermal imaging.
At everyday temperatures (objects between about -40 and 100 degrees Celsius), most of the infrared radiation emitted is in the long-wave infrared band, roughly 7 to 14 micrometers. This is the band thermal cameras for building diagnostics are designed to detect.
Why ordinary cameras cannot do this
Standard digital cameras use silicon detectors that respond to visible light and a portion of the near-infrared band (up to about 1.1 micrometers). They cannot detect the long-wave infrared radiation emitted by everyday objects at room temperature because silicon is essentially blind to wavelengths longer than the near-infrared.
Glass optics also stop working at these wavelengths. Ordinary silica glass absorbs long-wave infrared almost completely. A thermal camera lens must be made of a material that transmits LWIR — typically germanium, which looks dark gray to the eye but is transparent to the infrared band the detector needs.
These two physical constraints — that visible-light sensors and glass optics do not work at LWIR wavelengths — are why thermal imaging requires a specialized device rather than a software trick applied to ordinary photography.
For a device-focused overview rather than a physics walk-through, see the device overview guide and the home inspection tools hub.
The microbolometer detector
The detector at the heart of almost every consumer and prosumer thermal camera is the microbolometer array. A microbolometer is a tiny thermometer that responds to absorbed infrared radiation by changing temperature, which in turn changes its electrical resistance.
How a single microbolometer pixel works
A microbolometer pixel is a thin film of vanadium oxide or amorphous silicon suspended on a microscopic bridge structure with thermal isolation. Incoming infrared radiation strikes the film and warms it. The film’s electrical resistance changes with temperature. Readout electronics measure that resistance change and convert it to a digital value.
The thermal isolation is essential. The film must be sensitive enough to detect the very small amount of radiation incoming from a single pixel of the scene, which means it cannot be in significant thermal contact with the chip substrate or it would simply equilibrate to room temperature. The bridge structure achieves this by suspending the film on tiny supports across a small vacuum gap.
Pixel arrays
A modern microbolometer array contains tens of thousands to over a million individual pixels, arranged in a rectangular grid. Common sizes are 160×120 (about 19,000 pixels), 320×240 (about 77,000 pixels), 640×512 (about 328,000 pixels), and 1280×1024 (about 1.3 million pixels). Each pixel produces an independent temperature measurement on the order of 30 to 60 milli-Kelvin sensitivity (NETD).
Why uncooled detectors changed the market
Early thermal imaging used cooled photon detectors that required cryogenic cooling to operate. These detectors were expensive, fragile, and limited to industrial and military applications. The development of room-temperature microbolometers in the late 1990s and early 2000s brought thermal imaging into the consumer market by eliminating the cooling requirement. Modern handheld thermal cameras can fit in a pocket and run on a small battery — neither possible with cooled detectors.
How the signal becomes an image
The raw output of a microbolometer array is a set of resistance measurements, one per pixel. Several processing steps turn that data into the image a user sees on a screen.
Calibration
The camera applies a calibration table that maps each pixel’s resistance measurement to a temperature value. This calibration is performed at the factory and periodically updated through the camera’s flat-field calibration cycle (the brief pause and click users hear when the camera “shutters” — a mechanical shutter briefly blocks the lens to provide a uniform reference for the detector array).
Emissivity correction
Different surfaces emit infrared radiation differently even at the same temperature. The ratio of actual emission to ideal-blackbody emission is called emissivity, with values from 0 to 1. Most building materials have emissivity values around 0.9; polished metal has very low emissivity (around 0.05); glass and water have high emissivity (around 0.95). Accurate temperature readings require the camera to apply an emissivity correction for each surface.
Most building diagnostic work uses a default emissivity value (typically 0.95) that works adequately across common materials. Professional work involving precise temperature measurement requires adjusting emissivity per surface.
Pseudo-color mapping
The processed temperature data is mapped to a color palette for display. Common palettes include:
- Iron / rainbow. Black to dark purple to red to orange to yellow to white as temperature increases. The most familiar palette for general use.
- Gray scale. Black to white as temperature increases. Often preferred for forensic work where subtle gradients matter more than dramatic contrast.
- Hot iron / cold iron. Variations on the iron palette emphasizing one end of the temperature range.
- Custom palettes. Some cameras allow user-defined palettes for specific applications.
The pseudo-color is a visualization choice, not an inherent property of the data. The same image looks very different in different palettes; trained users sometimes switch palettes to surface different patterns in the same data.
Optional image fusion
Cameras with a built-in visible-light sensor can fuse the visible image with the thermal image, overlaying visible edges onto the thermal data. The fusion improves spatial localization of thermal anomalies — making it easier to know which wall, electrical box, or pipe the anomaly is associated with.
What affects image quality
Several factors determine how useful a thermal image is.
Detector resolution. More pixels means more detail and longer effective working distance.
Sensitivity (NETD). Smaller temperature differences are detectable with better NETD.
Temperature delta. The bigger the temperature difference between the area of interest and surrounding areas, the easier the imaging. Inspections during cold weather (large indoor-outdoor delta) produce clearer envelope imagery than inspections in mild weather.
Solar loading. Direct sunlight heats surfaces unevenly and masks the subsurface patterns of interest. Building thermography works best at night or under heavy overcast.
Surface reflectivity. Highly reflective surfaces (polished metal, glass at certain angles) reflect ambient temperatures and produce misleading readings.
Atmospheric conditions. Humidity, rain, fog, and particulates absorb infrared and degrade image quality, especially at longer range.
How interpretation works in practice
The physics produces an image; interpretation produces a diagnosis. A trained thermographer interprets a thermal image by combining several pieces of information:
- The thermal pattern itself — shape, intensity, gradient.
- The visual context — what is actually in front of the camera.
- The building construction — wall assembly, insulation type, framing layout.
- Environmental conditions — temperature differential, wind direction, solar loading history.
- Confirmatory measurements — moisture meter readings, electrical load checks, ventilation tests.
An image of a cool patch on a ceiling is data. The same image interpreted in context might suggest an active roof leak, an air leak around a recessed light, a cold patch where insulation is missing, or a duct loss into the attic. Distinguishing between these possibilities requires the full set of inputs, not the image alone.
Why understanding the physics matters for users
Homeowners and inspectors who understand the underlying physics make better use of the equipment they have. Three practical consequences of the physics matter regularly:
The camera shows surface temperatures only. Knowing this prevents the misinterpretation that the camera sees “through” things. Hidden conditions are inferred from surface effects.
Time of day and weather matter. Knowing that solar loading masks building patterns guides inspectors to schedule envelope work for nighttime or overcast conditions, when the imaging is reliable.
Reflective surfaces need different handling. Knowing that polished metal does not emit at the same rate as drywall prevents the diagnostic error of reading reflected temperatures as actual surface temperatures.
Common artifacts and how to recognize them
Thermal images contain artifacts that can mislead users who interpret them naively. Recognizing the common artifacts prevents diagnostic errors.
Reflections. Smooth surfaces — glass, polished metal, glossy paint — reflect the thermal radiation of other objects in the scene rather than emitting their own temperature. A polished refrigerator door imaged from across a kitchen reflects the operator’s body heat as a vague warm shape. Recognizing reflections means knowing what the actual surface looks like and noting when the thermal image disagrees with material expectations.
Solar loading patterns. Exterior surfaces that received direct sun earlier in the day retain residual heat for hours. Walls facing west show solar loading patterns into the evening. Inspectors waiting to image west-facing walls until after the surface has cooled (often a few hours after sunset) get cleaner imagery.
Convection shapes. Warm air rising from heat sources produces vertical thermal streaks in front of walls and against cold windows. The streaks are real air movements but can be confused with thermal anomalies in the wall behind them.
Camera self-heating. The camera body warms during use. The lens and detector can drift slightly in apparent temperature over a session, especially in extreme ambient conditions. Periodic flat-field calibration (the brief shutter cycles) corrects for most of this drift, but extended sessions in challenging conditions may require manual recalibration breaks.
Ghost images. Some detector technologies produce faint residual images of bright thermal sources for short periods after the camera has moved. The effect is minimal in modern uncooled microbolometers but can appear after imaging very hot sources at close range.
How thermal cameras have changed since early adoption
The history of consumer-accessible thermal imaging is worth noting because it shapes current product expectations. In the mid-1990s, a thermal imaging system suitable for building diagnostics cost tens of thousands of dollars and required cooled detectors with periodic detector replacement. Uncooled microbolometer development through the late 1990s and early 2000s brought costs down by an order of magnitude. The introduction of smartphone-attached thermal modules around 2014 brought thermal imaging into the consumer price range under $500.
The trajectory of resolution improvements has been similarly dramatic. Early consumer thermal cameras at 80×60 pixels are now superseded by 256×192 and 320×240 at the same price point. The 640×512 sensors that defined professional inspection work a decade ago now appear in mid-tier prosumer units.
For users new to thermal imaging in 2026, the capability available at every price point is dramatically better than it was even five years ago. The price-to-capability curve has steepened in the user’s favor across the entire range, from $300 smartphone attachments through $4,000 professional handhelds.
The role of training in actually using the technology well
Despite the improving hardware, training remains the limiting factor on most thermal imaging outcomes. The physics described in this guide is foundational, but actually interpreting building thermal images well requires building science knowledge, experience pattern-matching anomalies across many buildings, and discipline about confirming findings with other measurements. ASNT Level I thermography certification is the entry-level credential for professional work; Level II provides more independent analytical authority. Manufacturer-specific training adds equipment-specific competence. None of this training is required for a homeowner using a basic unit casually, but the gap between casual use and diagnostic use is wider than most new owners expect.
References
- Thermographic Inspections — Department of Energy
- Infrared Thermography Standards — InterNACHI
- Building Science Resources — ASHRAE