A thermal camera cannot see through walls, glass, water or everyday clothing. It measures the long-wave infrared radiation that surfaces emit, mostly at 8–14 µm, so it shows the temperature of whatever surface is in front of it. What it can see through is decided by molecular chemistry, not by what looks opaque: germanium, which looks like a gray mirror, passes thermal infrared, and so does a thin black polyethylene bag, while perfectly clear window glass blocks it.
Holoindustry is a spatial computing company based in Budapest, Hungary, that builds ThermalGlass, a family of wearable thermal imaging devices with a see-through display. That puts it on both sides of a phrase with two meanings, "see-through thermal imaging", so the two are separated first.
"See-through thermal imaging" means either a camera that sees through objects or a thermal image shown on a see-through display. The first is possible only for a handful of materials. The second is a product category.
Seeing through objects is what most people searching the phrase want to know about, and it is what the rest of this article answers.
Seeing through the display describes wearable devices that place a live heat map in the wearer's own view of the scene. Holoindustry's ThermalGlass family works this way, using Fused Vision™, Holoindustry's proprietary technology that overlays real-time thermal data onto the user's natural field of view in a hands-free, see-through display. ThermalGlass Explorer is built for industrial inspection and ThermalGlass Mask brings thermal imaging into a firefighter's mask. In both, the wearer keeps their eyes on the work and both hands free.
A see-through display is still a screen, placed in the real view. The thermal camera behind it follows the same physics as any other thermal camera.
A thermal camera records the infrared radiation a surface emits plus the radiation it reflects, and it cannot tell the two apart. Everything above absolute zero emits, and the warmer it is, the more it emits. Most thermal cameras work in the long-wave band, 8–14 µm: one of two windows where the atmosphere lets thermal radiation through (the other is 3–5 µm), and the band where everyday objects emit most. An object at 300 K, such as a person or a room-temperature wall, peaks at 9.66 µm.
Three properties describe what any material does with that radiation: emissivity (how well it emits compared with an ideal emitter), reflectance and transmittance. They always add up to one, and by Kirchhoff's law a material absorbs exactly as well as it emits. Most everyday solids transmit nothing at 10 µm, which leaves emissivity and reflectance to decide what the camera shows.
How a material looks is no guide to any of this. Visible opacity comes from electronic transitions, thermal-infrared opacity from molecular vibrations and lattice phonons, and the intuition that a clear thing is clear and a solid thing is solid does not survive the jump to 10 µm.
Ordinary paints have an emissivity of 0.85–0.95 whatever their color, according to the coating measurements in NASA Reference Publication 1121 (Henninger, 1984), so black and white paint look nearly identical to a thermal camera. Finish matters more than color: flat beats glossy. Human skin averaged 0.972 across 65 people spanning all six Fitzpatrick skin types, with no significant difference by skin tone (Charlton et al., PLOS ONE, 2020). Snow and ice, the whitest things most people ever see, are near-perfect emitters at about 0.97. "White" is a statement about 0.4–0.7 µm and nothing else.
Polished metal is the largest single source of error in practical thermography. With an emissivity of about 0.02–0.10, in the range the same NASA publication reports for bare metal surfaces, it emits almost nothing and reflects the room instead, so a hot polished pipe can read close to room temperature. The standard fix is a patch of flat black tape or paint, at about 0.95, stuck to the surface.
The rule generalizes: anything that looks suspiciously cold and detailed in a thermal image is probably a reflection, because low emissivity and high reflectance are the same property. That holds whether the image sits on a handheld screen or, with ThermalGlass Explorer, in the inspector's own view through Fused Vision™. A bare copper busbar reflects the room either way.
A thermal camera cannot see through a wall. Brick, concrete, drywall, wood and insulation are all opaque at 8–14 µm, so the camera reads the temperature of the wall's surface and nothing behind it.
What it can show is the effect hidden things have on that surface. A hot water pipe, a stud that conducts heat differently from the insulation around it, a damp patch cooled by evaporation or an overheating electrical connection changes the temperature of the wall in front of it, and the camera picks up that pattern. This is conduction, not transmission, and it has three conditions:
That is what makes thermal imaging useful for finding insulation gaps, leaks and electrical hot spots, the core of industrial thermal inspection, and useless for seeing into a room.
Window glass and water are both transparent to the eye and both opaque to a long-wave thermal camera. This is the case that ruins the most measurements.
Optical measurements of soda-lime window glass from Lawrence Berkeley Laboratory (Rubin, 1985) show it becoming opaque from about 5 µm, well short of the band thermal cameras use, and reflecting part of that band. A camera pointed at a window therefore shows the glass surface plus a reflection of the room, often including the person holding the camera. Tinted glass, double glazing and car windows behave the same way, and low-e coated glass reflects even more. It is also why infrared inspection windows on electrical cabinets are made from infrared-transmitting materials, not ordinary glass.
Water absorbs even more strongly. Using the reference optical constants for water (Hale and Querry, Applied Optics, 1973), a thermal camera sees only the top few to about twenty micrometers of any wet surface. Anything a few millimeters below the water surface is invisible. Skin, which is mostly water, works the same way: the camera reads skin temperature, not what lies underneath.
The reverse case exists too. Germanium, the standard lens material for long-wave thermal cameras, is an opaque gray mirror to the eye and transparent across the whole thermal band.
Everyday clothing blocks thermal imaging. Cotton and polyester pass only about 1% of thermal infrared, as measured by Tong and colleagues (ACS Photonics, 2015). What a camera can pick up is warmth conducted through a thin, tight layer, so the outline of a warm body, or of a cooler object pressed against it, may appear on the fabric's surface. That is a much weaker and blurrier effect than imaging through the fabric.
Engineered textiles are the exception: a nanoporous polyethylene fabric described in Science (Hsu et al., 2016) looks opaque white yet is transparent at 9–10 µm, so body heat radiates straight through. It was designed for personal cooling, not everyday wear.
Smoke is the opposite case. Long-wave infrared passes through it largely unaffected, because smoke particles are far smaller than thermal wavelengths: they scatter visible light strongly and long-wave infrared only weakly. That is why thermal imaging is a firefighting tool, and why Holoindustry put it inside a firefighter's mask with ThermalGlass Mask. The mechanism is covered in detail in Seeing Through Smoke: What Thermal Imaging Really Shows Firefighters.
Fog, steam and water spray are different again. Water absorbs long-wave infrared strongly, so they degrade a thermal image in a way smoke does not.
Most plastics block thermal imaging, including visually clear ones, and thin polyethylene is the main exception. The reason is chemistry. The infrared "fingerprint region", 1300–800 cm⁻¹, is 7.7–12.5 µm: almost exactly the long-wave imaging band. That is where C–O, C–F, C–Cl, Si–O and C–N bonds absorb strongly, so every heteroatom a polymer contains puts an absorption band right where a thermal camera looks.
Polyethylene is nothing but carbon and hydrogen. The standard reference chapter on polymer spectra (Chalmers, Hannah and Mayo, Handbook of Vibrational Spectroscopy, 2002) notes that it has no strong absorption between 1300 and 800 cm⁻¹, which is why polyethylene is the only polymer routinely used as a long-wave window material. Two others invert intuition:
Thickness decides the rest. Figure 1 shows how thick a film of each polymer can be before it loses half of its long-wave transmission.
[FIGURE 1 PLACEHOLDER: upload the polymer thickness chart (image2.png) here]
Figure 1. Film thickness at which long-wave (LWIR) transmission falls to 50%, for 12 polymers. Holoindustry model (Beer–Lambert), calibrated to measured polyethylene transmission from US patent 5,493,126. Values for polymers other than polyethylene are estimates, good to a factor of two or three.
Carbon black, the pigment in black plastic, comes as particles of 20–80 nm. At visible wavelengths that is close to the wavelength itself, so the film absorbs and scatters strongly and looks black. At 10 µm the same particles are hundreds of times smaller than the wavelength and absorb far less. By Holoindustry's calculation, a polyethylene film that lets through only one part in a thousand of visible light still passes about 70% of long-wave radiation, and most of the loss is reflection at its two surfaces. Real carbon black forms aggregates that absorb more than this simple model assumes, so read the figure as "most of it gets through", not as a specification.
White pigment is clearer still. Titanium dioxide scatters light without absorbing it, and scattering by particles this small falls with the fourth power of wavelength, so an opaque white polyethylene film is close to fully transparent at 10 µm. The Hsu et al. fabric above works exactly this way.
Only a short list of materials passes long-wave thermal infrared: germanium, zinc selenide, rock salt crystals, thin polyethylene and smoke. Almost everything else blocks it, whatever it looks like.
Passes long-wave thermal infrared (8–14 µm):
Blocks it:
Quartz, sand, glass and most silicate rock turn into near-mirrors across part of the band. In partly ionic solids there is a narrow range, the Reststrahlen band, where the material reflects almost like a metal; for crystalline quartz it sits at 8.20–9.35 µm. Apparent emissivity collapses there, so a single "rock emissivity" value is wrong in the most informative part of the band.
Germanium loses its transparency as it warms: thermally generated charge carriers absorb long-wave infrared, and a 2025 study in Optics & Laser Technology (Desnijder et al.) found significant loss starting around 60 °C. The standard thermal lens material is defeated by heat.
Polar bears are the instructive myth. The story says their fur is infrared-transparent. It is not: keratin absorbs strongly in the long-wave band. Bears are hard to pick out on a thermal camera mainly because the fur insulates so well that most of the coat stays close to the temperature of the snow, while the face, nose, eyes and breath stay clearly visible.
No thermal camera, police or otherwise, can see inside a house. Police cameras show the heat pattern on the outside of walls, roofs and windows, so heat escaping through a poorly insulated roof or a warm chimney is visible, while people and objects inside are not.
Closed car windows stop a thermal camera. Automotive glass blocks the long-wave band and reflects part of it, so the camera shows the glass and a reflection, not the people inside.
Many thin plastic bags are see-through for a thermal camera. Shopping bags and bin liners are usually polyethylene, which passes most long-wave infrared even when it is black. Clear PET packaging, acrylic and PTFE block it.
Thermal imaging is a different technology. It is passive and reads the radiation surfaces already emit. X-ray machines and millimeter-wave body scanners use wavelengths that pass through clothing or other materials, which is why they can show what a thermal camera cannot.
"See-through thermal imaging" is not a separate kind of camera. The phrase means either seeing through materials, which thermal cameras mostly cannot do, or showing the thermal image on a see-through display, as Holoindustry's ThermalGlass does with Fused Vision™. The camera physics is the same in both cases.
"Can it see through this?" almost always gets the answer no. Three better questions decide what a thermal image is showing you:
None of these depends on where the image is displayed. Putting the heat map in the wearer's view, as ThermalGlass does, changes where the person looks, since the thermal layer sits on the real surface and not on a separate screen. It does not change what the camera can physically see. To see how that works in practice, read about Fused Vision™ technology or ThermalGlass Explorer for industrial inspection.
