A thermal camera finds the automotive faults that a scan tool cannot describe: a dragging brake, a blocked radiator core, a seized injector, a resistive earth. This page compares nine imagers on the numbers their own listings publish, and it is blunt about the difference between a real detector resolution and an upscaled one.
Warm It Up
Set Emissivity
Compare Sides
As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Cameras were compared on the detector resolution each listing states, any enhanced or super resolution figure quoted alongside it, stated thermal sensitivity, display or host device, and the extra imaging modes the manufacturer names. Where a listing quotes only an upscaled figure the table says so. Figures are the manufacturers' own and we publish no temperature measurements of our own.
A dual-lens imager from a workshop diagnostic brand, with five blend modes.
Two cameras on a phone attachment, from a brand that lives in workshops.
One of the few listings here that states a real detector resolution.
No pick matches that combination yet. Try a wider budget or a different measurement.
| Product | Form | Stated detector resolution | Enhanced or other imaging figure stated | Display | |
|---|---|---|---|---|---|
| FOXWELL RT280 PRO | Handheld, dual lens | Not stated | 512 by 384 TISR, 2 MP visible camera, 5 image modes | Own screen, size not stated | View |
| TOPDON TC002C Duo | Smartphone module | Not stated | Dual camera system, thermal sensor plus visual camera | Uses the phone screen | View |
| BSIDE handheld imager | Handheld | 240 by 240 | 15 thermal palettes | Own screen, size not stated | View |
| AMPBANK M256 | Handheld | 256 by 192 | 512 by 384 TISR, 3D thermal analysis | Own screen, size not stated | View |
| FLIR E50 | Handheld | 240 by 180 | MSX blending, 15 degree lens, Class II 1 mW laser | Own screen, size not stated | View |
| Raythink EX10 Pro | Handheld | Not stated | 320 by 240 super resolution in real time, 40 mK sensitivity | 2.8 inch | View |
| HSFTOOLS HP96 | Handheld | Not stated | IntellFault scene recognition using deep learning | 3.5 inch touch screen | View |
| Testo 860i | Wireless phone imager | Not stated | Bluetooth and Wi-Fi to the Testo Smart App, USB-C cable supplied | Uses the phone screen | View |
| Robofinity InsightDrive | Fitted to the vehicle | Not stated | Thermal imaging night vision system with recording | Not stated | View |
Resolution figures are quoted exactly as each manufacturer publishes them. Where a listing quotes only TISR or super resolution, that is an upscaled output and the native detector figure is recorded as Not stated. Only three of the nine publish a native resolution. Nothing here is a measurement of ours.
Four things separate these nine:
TISR and super resolution are interpolated outputs. A 256 by 192 detector quoted as 512 by 384 TISR still has 49,152 real measurements. Compare native figures or compare nothing.
Only the Raythink listing states a figure, 40 mK. Sensitivity decides whether a slightly warm connector separates from the harness beside it, which is most of what electrical fault finding on a vehicle is.
Under a bonnet everything is warm and nothing has edges. The FOXWELL and TOPDON units carry a second visible camera so a blended image shows which component the hot spot belongs to.
A phone module means the phone goes where the camera goes. A handheld keeps the expensive screen in your hand and the lens in the engine bay, which is often the difference that matters.
The ones defined by a temperature difference rather than a fault code. A brake that drags runs hotter than its opposite number. A radiator with blocked cores shows cold stripes across a warm face. A dead cylinder shows a cool exhaust runner at the head. A resistive earth or a failing connector shows a hot spot in a harness that should be uniform. A wheel bearing on its way out runs warmer than the other three. In every case the diagnosis comes from comparison, not from an absolute reading.
No. TISR and similar super resolution labels describe an output image built by interpolation from a smaller detector, usually combined with slight movement between frames. The AMPBANK listing is the honest example here because it states both figures, 256 by 192 native and 512 by 384 TISR. The FOXWELL listing states only 512 by 384 TISR, so the number of real measurements is unknown from the listing alone.
Because a polished or chromed surface has low emissivity, meaning it radiates poorly and reflects strongly, so the camera mostly sees the temperature of whatever is being reflected. A bare exhaust manifold, a chrome trim piece or an aluminium heat shield can read far colder than it is. The workaround is to read an adjacent painted or dull surface, or apply a patch of matt tape and measure that.
For most comparative diagnosis, yes. The TOPDON TC002C and the Testo 860i both put the image on a screen you already carry, and the images go straight into the job record. The practical limits are physical rather than optical: the phone has to go wherever the camera goes, a hot engine bay is not a good place for a phone, and a dropped phone costs more than a dropped imager.
No. Thermal cameras read the surface temperature of the first opaque object in front of them and nothing behind it. What they can do is show heat that has conducted through to the surface, which is how a dragging brake shows up on the outside of a wheel and how a blocked core shows on the face of a radiator. That is conduction becoming visible, not the camera seeing through anything.
The problem with a pure thermal image on a vehicle is that everything under a bonnet is hot and nothing has edges. A blended image solves it by laying the outline from a visible camera over the heat data, so the warm spot you are looking at is identifiably the alternator rather than an orange blob. FOXWELL states a built-in infrared camera alongside a 2 megapixel visible light camera and five image modes covering thermal, visual and combinations of the two, which is the feature that makes a workshop image worth showing to a customer.
FOXWELL is a diagnostic tool brand rather than a test instrument brand, and that shows in the framing of the product. The RT280 PRO is presented as a troubleshooting camera for efficient fault finding, which is the right mindset for automotive work where the job is comparative: this side against that side, this cylinder against its neighbour, this connector against the one next to it.
The specification needs reading carefully. The 512 by 384 figure quoted on the listing is TISR, which is an upscaled output rather than the number of detector elements, and the native resolution is not stated separately. Neither is thermal sensitivity, temperature range, accuracy or emissivity adjustment. Those matter, and the manual is where they live.
A phone module is the form factor that gets used, because the screen is already in your hand, the images land straight into the job file and there is no second device to charge. The TC002C Duo adds the thing most phone thermal modules lack, which is a second visible-light camera in the module itself, so the blended image does not depend on borrowing the phone camera and dealing with the parallax between two lenses several centimetres apart.
TOPDON matters here for the same reason FOXWELL does. This is a company whose other products are scan tools and battery testers, so the software is built by people who expect the camera to be pointed at a wiring harness and a brake rotor rather than at a wall cavity. For a technician who already uses their phone for fault codes and photographs, it is the lowest-friction way to add thermal.
The listing is unusually quiet on specification. There is no stated detector resolution, no thermal sensitivity, no temperature range, no accuracy figure and no statement about emissivity adjustment. A phone module also has practical limits under a bonnet: the phone has to go where the camera goes, which is not always somewhere you want a phone.
Among nine listings that mostly quote enhanced or super resolution figures, BSIDE states 240 by 240 as the resolution of the camera. That is a square sensor format, which is unusual, and it gives 57,600 pixels of real data. It is not a high number by current standards, but it is a number you can reason about, and a stated native figure is worth more than a larger upscaled one when you are deciding whether a camera will resolve a single injector at arm's length.
Fifteen palettes sounds like a marketing count and is more useful than it appears in automotive work, because the right palette depends on what you are looking at. Iron and rainbow palettes exaggerate small differences, which is what you want across a row of exhaust ports. A grey palette preserves shape, which is what you want when the customer has to recognise the part.
Everything else is unstated: no thermal sensitivity, no temperature range, no accuracy, no emissivity adjustment, no frame rate and no lens field of view. At this price the honest description is a comparative tool. It will show you that one wheel is hotter than the other three. It will not give you a defensible temperature.
Short answer: the FOXWELL RT280 PRO is the pick for a working bay because it pairs an infrared camera with a real 2 megapixel visible camera and five blend modes, which is what makes a thermal image legible under a bonnet; the TOPDON TC002C Duo is the best phone module, with both cameras in the attachment and workshop-oriented software; and the BSIDE handheld is the value pick and one of only three listings here that states a native detector resolution rather than an upscaled one.
Twelve listings were supplied and three of them are not automotive thermal cameras. The UNI-T UT-Z002 and the Seek Thermal Nano macro lens are both accessories: lenses that bolt onto a camera you already own, stated on their own listings as compatible with specific imagers only. Neither takes a picture. The FLIR MR160 is a moisture meter with an 80 by 60 Lepton imager built in to guide the pin and pinless moisture readings, which is a building diagnostics tool with no vehicle application. Adding any of those to reach twelve would have padded the table with things that cannot do the job, so the count is nine and every row is a camera you could point at a vehicle.
This is the single most important thing to understand before comparing these nine. A thermal detector is a physical array of microbolometer elements, and the count of those elements is the only figure that describes how much real temperature data the camera collects. TISR, super resolution and similar labels describe an output image assembled by interpolation, often combined with slight sensor movement between frames, and the resulting file is larger without containing proportionally more measurement. The AMPBANK listing is the honest one: it states 256 by 192 infrared resolution and 512 by 384 TISR separately, so you can see both. The FOXWELL listing states only the 512 by 384 TISR figure and the Raythink listing states 320 by 240 super resolution, so the native counts are unknown from the listings alone. Three of the nine publish a native figure: BSIDE at 240 by 240, AMPBANK at 256 by 192 and FLIR at 240 by 180. How to compare thermal camera resolution works through the arithmetic.
Automotive faults rarely present as a large temperature difference across a small object; they present as a small difference across a comparatively large one. A connector with a marginal crimp might run a few degrees above the rest of the harness. A partially blocked radiator core shows as a stripe a few degrees cooler than its neighbours. Whether a camera separates those is decided by thermal sensitivity, usually quoted in millikelvin, not by pixel count. Only the Raythink EX10 Pro listing publishes a figure here, stating 40 mK. Every other listing on this page is silent on it, which is a significant gap given how central it is. What NETD means on a thermal camera explains what the number describes and why a lower one is better.
Point a pure thermal camera into an engine bay and the result is a warm orange field with a few brighter areas and no identifiable components. That is a real limitation, because the value of the image depends on knowing which part is hot. FLIR calls its version MSX, FOXWELL supplies five image modes from a 2 megapixel visible camera and TOPDON puts both cameras in the phone module. The technical reason to want both lenses in the same housing is parallax: two lenses a few centimetres apart see slightly different geometry, and at the short working distances of under-bonnet inspection that misalignment is visible. A module that carries its own visible camera aligns the two by design.
Vehicles are full of the surfaces that thermal cameras handle worst. Chrome trim, polished aluminium, bare stainless exhaust, foil heat shielding and machined metal all have low emissivity, which means they radiate poorly and reflect strongly, so the camera reports the temperature of whatever is being reflected rather than the part itself. A stainless manifold can read cooler than the painted block beside it while being hundreds of degrees hotter. The practical fixes are to read an adjacent dull or painted surface, to stick a patch of matt tape on the part and read that, or to set the emissivity value correctly if the camera allows it. None of these nine listings states whether emissivity is adjustable, which is a real omission. Why shiny metal can mislead a thermal camera and how to set emissivity cover both sides of this.
The Robofinity InsightDrive is on this page because it is genuinely automotive and genuinely thermal, but it is not a diagnostic instrument. It is a thermal night vision system that mounts in the vehicle and records, described on its listing as neither an ordinary dashcam nor a regular night vision device. It exists to show a driver a pedestrian or an animal beyond the reach of the headlights, not to show a technician a warm bearing. It stays in the table because the page is about automotive use and that is an automotive use, but a workshop buying a diagnostic tool should look at the other eight.
Work comparatively and let the vehicle warm up first, because a cold engine hides almost every fault worth finding. Take a reference image of the good side before you look at the suspect side. Keep the working distance consistent, since the area each detector element covers grows with distance and a small hot spot averages away. Avoid reading through glass, which is opaque to the wavelengths these cameras use, so pointing one at a windscreen measures the glass. And treat every absolute number as provisional unless emissivity has been handled, because on a vehicle it usually has not been.
Cameras were compared on the detector resolution stated by the manufacturer, any enhanced, TISR or super resolution figure quoted alongside it, stated thermal sensitivity, the presence of a second visible-light camera, display or host device, and the imaging modes each listing names. Upscaled figures were never recorded as detector resolution. Three products supplied with this group were excluded as off-category and the reasons are given on the page. We ran no laboratory tests and publish no temperature measurements of our own. Amazon data supplies images, availability and links only.