As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
A thermal camera is an imaging radiometer. Its lens focuses infrared radiation onto a detector array, usually a microbolometer, and each detector element produces a signal proportional to the radiant power falling on it. The camera converts those signals into a false-colour image where colour represents apparent temperature. The key specifications are the detector resolution (how many independent temperature readings make the picture), the thermal sensitivity or NETD (the smallest temperature difference the detector can resolve), and the measurement accuracy. A camera such as the MiLESEEY TR20 illustrates the pro-level end with a 320 by 240 detector and a stated NETD below 40 millikelvin. A plug-in model such as the AMPBANK M192 shows the consumer end, where the phone supplies the screen and the camera supplies the sensor. Neither replaces a contact thermometer for absolute measurement, and neither sees through walls, glass, or water.
The sensor is a grid of tiny thermometers, and its size sets what you can see
Every thermal camera contains a focal plane array: a rectangular grid of detector elements, each of which responds to the infrared power falling on it. A 320 by 240 array has 76,800 elements. A 192 by 192 array has 36,864. Each element is one temperature reading, and each reading becomes one pixel in the displayed image. When a listing says “320 x 240 IR resolution”, that is the detector grid, not the screen resolution. The screen may be larger or smaller; it cannot add thermal detail that the detector did not capture.
This matters because a thermal camera’s ability to resolve a small object at a distance is governed by the detector grid and the lens field of view. A 50 degree wide-angle lens, as specified on the MiLESEEY TR20, covers a lot of wall in one frame, but each detector element then covers a larger patch of that wall. A narrow lens puts more elements on a smaller area. Neither is better in the abstract; the right choice depends on whether you are scanning a room or inspecting a circuit board.
Some listings quote a higher figure than the detector grid. The DCTOPT listing, for example, states a 320 by 240 detector and also a 240 by 180 “Super IR” resolution. The AMPBANK M192 states 192 by 192 native and 384 by 384 super-resolution. These figures describe processing that interpolates between detector elements; they improve the appearance of the image but do not create new independent measurements. Treat the native detector grid as the honest specification and the enhanced figure as a display feature.
NETD is the smallest temperature difference the camera can show, and it is not the same as accuracy
Thermal sensitivity, usually written as NETD (noise equivalent temperature difference), describes the smallest temperature difference the detector can distinguish from its own noise. A lower number is better. The MiLESEEY TR20 states NETD below 40 millikelvin, which is 0.04 degrees Celsius. The DCTOPT listing states below 50 millikelvin. The AMPBANK M192 states 40 millikelvin. These are plausible figures for uncooled microbolometer detectors and they tell you the camera can show subtle gradients, such as the slight warmth of a stud behind plasterboard or a failing connection under light load.
NETD is not measurement accuracy. Accuracy is how close the camera’s absolute temperature reading is to the true temperature, and it is usually quoted as a percentage of reading plus a fixed offset. The MiLESEEY TR20 states plus or minus 2 percent. The AMPBANK M192 states a maximum of plus or minus 2 degrees Celsius or 2 percent. A camera can have excellent NETD and mediocre accuracy, or the reverse. For finding anomalies, NETD matters more. For reporting a temperature, accuracy matters more, and even then the reading depends on emissivity.
One common confusion: a listing that quotes a temperature range and an accuracy figure is not quoting a resolution. The range tells you what temperatures the detector can handle without saturating. The accuracy tells you how far the displayed number may be from truth. Neither tells you the smallest object the camera can resolve, which depends on the detector grid and the lens.
Emissivity is why two cameras can disagree about the same surface
Every material radiates infrared in proportion to its emissivity, a number between zero and one. A perfect black body has an emissivity of one. Shiny metal can have an emissivity below 0.1, which means it radiates very little of its own heat and mostly reflects the heat of whatever is around it. A thermal camera that assumes an emissivity of 0.95, the usual default for organic materials and painted surfaces, will read a shiny metal surface far too low or far too high depending on what it is reflecting.
This is not a defect in the camera. It is physics. The AMPBANK M192 listing states that emissivity is adjustable within the app, and that is the correct feature to look for if you intend to measure anything other than matte surfaces. The MiLESEEY TR20 listing does not state whether emissivity is adjustable; if you need to measure bare metal, that is a question to ask before buying. The practical workaround is to cover the surface with a known-emissivity material, such as matte tape or a smear of water-based paint, and let that reach thermal equilibrium before measuring.
Reflections are the other half of the problem. Glass is opaque to most thermal infrared, so a thermal camera sees the surface of a window, not through it. A shiny metal panel may show the reflected image of the person holding the camera. Before trusting a reading, change your viewing angle; if the apparent temperature changes, you are looking at a reflection.
What the picture can and cannot tell you about a building, a circuit, or a machine
In a building, a thermal camera shows surface temperature differences. A cold patch on an interior wall in winter may be missing insulation, a thermal bridge, or a leak. The camera cannot tell you which. It tells you where to look with a moisture meter or a borescope. The same logic applies to electrical inspection: a warm connection under load is a symptom, not a diagnosis. The camera finds the anomaly; a clamp meter and a torque screwdriver find the cause.
The DCTOPT listing mentions detecting hidden pet urine stains, which is plausible because evaporating moisture cools the surface, but the camera is detecting the temperature effect, not the stain itself. The AMPBANK M192 listing mentions HVAC ductwork analysis and leak detection, which are classic thermal applications because ducts and pipes run at temperatures different from their surroundings. In every case, the camera is a screening tool. It narrows the search area; it does not confirm the fault.
Where the whole category stops being the right answer is when you need to measure temperature inside a space you cannot see, or through a material that blocks infrared. Thermal cameras do not see through walls, do not see through glass, and do not measure the temperature of a surface that is behind a reflective barrier. For those jobs you need a contact thermometer, a borescope, or a moisture meter. A thermal camera is also not a gas detector; it cannot see carbon monoxide or methane, though it can sometimes show the cold plume of a leaking refrigerant.
How to compare two thermal cameras without being misled by the headline numbers
Start with the native detector resolution. A higher grid means more independent measurements and, for a given lens, more detail. Then look at NETD; a lower figure means the camera can show smaller temperature differences. Then look at accuracy, and check whether emissivity is adjustable. Then look at the practical features: frame rate, storage, battery, and whether the camera works standalone or needs a phone. The table below compares the stated figures for the products referenced in this article. Where a listing is silent, the cell says so; do not fill the gap with an assumption.
| Model | Native detector | Enhanced resolution | NETD | Temperature range | Accuracy | Frame rate |
|---|---|---|---|---|---|---|
| MiLESEEY TR20 | 320 x 240 | Not stated | Below 40 mK | -4 to 1022 F | Plus or minus 2 percent | 25 Hz |
| Bosch PRO GTC600C | Not stated in the listing bullets | Not stated | Not stated | Not stated | Not stated | Not stated |
| AMPBANK M192 | 192 x 192 | 384 x 384 | 40 mK | -4 to 842 F | Plus or minus 2 C or 2 percent | 25 Hz |
| DCTOPT | 320 x 240 | 240 x 180 | Below 50 mK | Not stated in the listing bullets | Not stated in the listing bullets | 25 Hz |
The Bosch listing is a useful lesson in reading listings carefully. It describes storage for 500 images, voice notes, and an intuitive interface, but it does not publish detector resolution, NETD, temperature range, or accuracy. Those are the specifications that determine what the camera can measure. Without them, you cannot compare it on performance; you can only compare it on features. That is not a criticism of the product, it is a statement about the listing.
The DCTOPT listing has an internal inconsistency: it states a 320 by 240 detector and a 240 by 180 super IR resolution. The enhanced figure is lower than the native figure, which is unusual. Treat the native 320 by 240 as the meaningful number and regard the enhanced figure as unclear.
The maintenance and handling that keeps a thermal camera trustworthy
A thermal camera is an optical instrument, and its accuracy depends on the lens being clean and the detector being at a stable temperature. Most handheld cameras need a few minutes after power-on to reach thermal equilibrium, especially if they have been carried in a cold vehicle or a hot cab. The listing for the MiLESEEY TR20 does not state a warm-up time, and neither do the others; the general practice is to let the camera settle before taking a reading you intend to rely on.
Clean the lens with a soft brush and a lint-free cloth. Do not use solvent. A fingerprint on the lens scatters infrared and reduces the apparent temperature of everything you look at. The same applies to the protective window if the camera has one. Store the camera with the lens cap on and the battery removed if it is not going to be used for a long period.
Batteries are the other weak point. The Bosch GTC600C is a 12 volt max connected tool, which implies it uses the Bosch 12 volt battery platform; the listing does not state runtime. The AMPBANK M192 has no battery at all, drawing power from the phone. That is convenient but means the phone’s battery is the limiting factor. For a long inspection, a small power bank is a practical accessory.
Finally, keep a record of your settings. If you change emissivity, distance, or ambient temperature, note it. A thermal image without its settings is a picture, not a measurement. Some cameras embed the settings in the file; the listings here do not state whether they do. Assume they do not and record them separately.
The honest limit of a thermal camera
A thermal camera is a screening instrument. It shows you where surface temperatures differ, and that is often enough to find a problem quickly. It does not measure the temperature of something you cannot see, it does not see through glass or walls, and it cannot diagnose a fault on its own. If you need an absolute temperature, use a contact probe. If you need to find where to put the probe, a thermal camera is the right tool. The specifications that decide whether a particular camera is fit for that job are the native detector resolution, the NETD, the accuracy, and whether emissivity is adjustable. Everything else is convenience.
Frequently Asked Questions
Can a thermal camera see through walls?
No. Thermal cameras detect infrared radiation emitted by surfaces. Walls are opaque to that radiation, so the camera shows the surface temperature of the wall, not what is behind it. A cold patch may indicate missing insulation or a thermal bridge, but the camera cannot tell you which; you need a moisture meter or a borescope to investigate further.
Is a higher resolution always better?
A higher native detector resolution gives you more independent temperature readings and therefore more detail, which is generally better for finding small anomalies at a distance. But it also costs more and produces larger files. For scanning a room for insulation gaps, a 192 by 192 detector may be enough. For inspecting a circuit board, you need a narrow lens and as many detector elements as you can afford. The enhanced resolutions quoted by some listings are interpolation, not new measurements.
What does NETD mean in practice?
NETD is the smallest temperature difference the camera can distinguish from noise. A NETD of 40 millikelvin means the camera can show a difference of 0.04 degrees Celsius. That is fine for building inspection, where surface temperature differences are usually several degrees. It is also fine for electrical inspection under load. It is not fine for measuring the temperature of a surface that is only slightly different from ambient, which is why some medical and scientific applications use cooled detectors with much lower NETD.
Why does my thermal camera give a different reading from my contact thermometer?
The most common reason is emissivity. A thermal camera infers temperature from emitted infrared, assuming a certain emissivity. A contact thermometer measures temperature by direct conduction. On a shiny metal surface, the camera may read much lower or higher than the true temperature because the surface is reflecting its surroundings. Other reasons include the camera's accuracy specification, the distance to the target, and the size of the target relative to the detector element. Check emissivity settings first.
Do I need a thermal camera with a built-in screen?
It depends on how you work. A standalone camera with a built-in screen, such as the MiLESEEY TR20 with its 2.8 inch LCD, is convenient because you do not need a phone and the camera is always ready. A plug-in model such as the AMPBANK M192 uses the phone's screen, which may be larger and easier to share, but it ties up the phone and drains its battery. Neither is inherently better; choose based on whether you want a dedicated instrument or a phone accessory.
Can a thermal camera detect a gas leak?
A thermal camera cannot detect combustible gases or carbon monoxide directly. It can sometimes show the cold plume of a leaking refrigerant or the temperature change caused by a gas expanding, but that is not a gas detection function and it is not reliable for safety. For gas detection you need a dedicated gas detector with the appropriate sensor. A thermal camera is not a substitute for a gas sniffer or a carbon monoxide alarm.