How to Choose a Thermal Camera

Before writing this I actually pulled the official product pages, official spec tables and official manuals for five thermal cameras currently on sale, and the same pattern kept showing up: the number a listing puts in the headline is usually not the number that decides whether you can see the problem or trust the reading. A camera advertised as 480x360 is natively 160x120. Cameras labelled 8.7 Hz and sub-9 Hz can come off the same product line. And "±2°C" is almost always followed by a paragraph of conditions. This is not a shootout with charts; it is the manufacturers' own numbers placed side by side, in the order you should read them.

TL;DR: check native infrared resolution first, not the super-resolution, SuperIR or MSX figure. Second, check the conditions attached to NETD, and do not mix NETD with MRDT when one spec table lists both. Third, look at frame rate — 8.7 Hz and sub-9 Hz are not performance compromises, they sit just under a 9 fps export-classification line. Fourth, read the accuracy conditions and the split temperature ranges. Of the five, the highest native resolution belongs to the Seek Thermal CompactPRO at 320x240 (76,800 pixels), while the handheld whose official paperwork deserves the closest reading is the Klein Tools TI290, because its product page and its manual disagree with each other.

Check 1: native resolution, not the number after upscaling

FLIR's own product page lists the Flir One Pro as "Thermal Resolution 160 × 120 (native resolution) 480 × 360 (super resolution)". The same page explains what VividIR does, in the manufacturer's words: "VividIR upscales a device's native resolution by up to 3x". Put the two numbers together and you can see exactly what that 3x means — 480÷160 = 3 and 360÷120 = 3, so the image is stretched three times in each direction, which is nine times the pixel count (19,200 → 172,800). Those extra pixels are not temperature samples the detector actually took. They make the image easier to interpret; they do not separate two hot spots that were already blurred together.

The retail listing is unusually honest about it: the Flir One Pro title on Amazon literally reads "480x360 Super Resolution (160x120 Native IR)". The comparison table in the same page's A+ content lists three tiers side by side — 240×180, 240×180 and 480×360 — which correspond to native sensors of 80×60 and 160×120. Find the native figure in parentheses before you compare anything.

Check 2: NETD is one number, MRDT is another

NETD (noise equivalent differential temperature) describes how small a temperature difference the sensor can resolve, and lower is more sensitive. But the way manufacturers publish it differs: HIKMICRO and TOPDON state the test conditions next to the figure ("< 40 mK (@ 25 °C, F#=1.0)"), while Klein Tools product pages simply say "<60 mK" or "≤50 mK" with no conditions given.

FLIR's layout is the easiest to misread. The Flir One Pro spec table carries both "Thermal Sensitivity/NETD 70 mK" and a separate "Thermal Sensitivity [MRDT]" row — and that MRDT row reads 100 mK on FLIR's Italy site and 150 mK on its Spain site, while NETD stays 70 mK on both (checked 2026-10-24). Those are two different metrics. Comparing brands across the wrong row will put you out by more than a factor of two. The rule is simple: compare the same metric under the same conditions, and when no conditions are published, treat the figure as indicative rather than as a buying criterion.

Check 3: why so many cameras stop just under 9 Hz

This is the least intuitive layer. Flir One Pro: 8.7 Hz per FLIR's official page. Flir C3-X: 8.7 Hz. Klein Tools TI250 and TI290: "Maximum Frames per Second: <9 Hz". Seek Thermal CompactPRO: "Frame Rate < 9 Hz" in the official spec sheet. Four manufacturers landing on the same threshold is not a coincidence.

AXIS states it plainly in its official USA Export Fact Sheet: cameras at "≤ 9fps" fall under ECCN 6A993 in the US Commerce Department's EAR, while 30 fps versions fall under ECCN 6A003.b.4.b, with authorization from Commerce possibly required for almost every destination except Canada. A FLIR dealer's published export-conditions page gives the same split: below 9 Hz (the page cites 7.5 Hz) is ECCN 6A993, and 30/60 Hz is ECCN 6A003.b.4.b. Seek's official sell sheet prints "Subject to US EAR Export Regulations" directly in its specification block.

So 9 fps is a classification boundary, and consumer products are held under it so they can move through more markets with fewer restrictions. You pay for that at the point of use: panning a sub-9 Hz camera, or filming anything that moves, smears. Higher-rate versions of the same product line do exist — Seek's official part numbers are LQ-AAAX (iPhone) and UQ-AAAX (Android), while dealer listings show UQ-EAAX and LQ-EAAX as the "FastFrame >15 Hz" variants (dealer figures, not the official spec sheet). Confirm which suffix you are buying.

Check 4: accuracy always has conditions, and ranges are split

None of the five publishes accuracy as a bare number. The Flir C3-X reads "0 to 100°C: ±3°C; 100 to 300°C: ±3%", with the condition printed in small type: "ambient temp. 15 to 35°C and object temp. above 0°C". The Flir One Pro says "±3°C or ±5%, typical", then limits it in a footnote: 60 seconds after start-up, with the unit between 15°C and 35°C and the scene between 5°C and 120°C. Klein Tools splits the TI250 into two bands: ±2°C or 2% above 212°F (100°C), and ±3°C or 3% below it.

Range needs the same scrutiny. The Flir One Pro gives two spans: -20°C to 120°C, and 0°C to 400°C. The C3-X covers -20 to 300°C. The Klein TI250 covers -20 to 400°C. And the TI290 carries an official contradiction worth memorising: the product page lists -40°C to 400°C, while the manual (TI290-manual-web.pdf, downloaded and parsed 2026-10-24) lists -4°F to 752°F (-20°C to 400°C) and adds in the same spec block: "NOTE: Not calibrated below -10°C (14°F)". Readings below -10°C sit in a range the manufacturer itself says is uncalibrated. If you work on refrigeration lines or cold-climate envelopes, that note matters more than the bottom of the range.

Check 5: field of view, IFOV, and minimum focus distance

Resolution decides how many points you can separate; field of view decides how much fits in one frame; IFOV (instantaneous field of view, mrad per pixel) connects them by telling you how much real-world size a single pixel covers at a given distance. The Flir C3-X publishes 7.9 mrad/pixel at a 53.6° FOV. The Seek CompactPRO has a 32° FOV, the narrowest of the five, so you stand further back to frame a distribution panel. The Klein TI250 is 50°±3°; the TI290 is 55.6°±2.8° in the manual and 56°±2.8° on the product page.

Close-up ability diverges more. The C3-X publishes "Minimum Focus Distance Thermal: 0.1 m, MSX: 0.3 m" and is focus-free. The Flir One Pro is fixed focus, "Fixed 15 cm — Infinity", so anything closer than 15 cm is out. Klein publishes no minimum focus distance for either model, only FOV. If you image small components on a board, minimum focus distance becomes the bottleneck before resolution does — which is why some vendors sell a macro lens separately; HIKMICRO's official Pocket2 datasheet lists one, rated for detail down to about 100 microns.

Emissivity: a metal enclosure reading low is not a broken camera

A thermal camera measures infrared radiation leaving a surface, and materials differ a lot in how well they emit. FLIR's answer is presets: the Flir One Pro spec lists "Emissivity Settings Matte: 95%, Semi-Matte: 80%, Semi-Glossy: 60%, Glossy: 30%" — glossy metal is handled at 30%. The C3-X goes further, listing under "Measurements correction" the emissivity setting (matt/semi-matt/semi-gloss plus a custom value), reflected apparent temperature, and atmospheric compensation. Klein makes it continuously adjustable: both the TI250 and TI290 list "Emissivity: 0.01 to 0.99, Adjustable", and the TI290 manual adds that 0.95 is the default. Seek's official sell sheet puts "Adjustable Emissivity" among the key features, with the line "Calibrate for reflective surfaces when accuracy is critical".

Practically: the same brushed aluminium panel measured at 0.95 and at 0.30 can differ by tens of degrees. Set emissivity before you measure metal, glossy surfaces or glass, or you will blame the camera for your own setting.

Real drawbacks of the five models

Flir One Pro (phone-attached, iOS Lightning version): 160×120 native is on the low side here and depends on upscaling to reach 480×360; battery life is about 1 hour with a 40-minute charge, per FLIR; emissivity is limited to four presets rather than a 0.01–0.99 custom value; the Lightning version supports iPhone 14 and earlier, so iPhone 15 and later need an adapter; the English page says 1.5 m drop testing while European regional pages say 1.8 m.

Seek Thermal CompactPRO (phone-attached, iOS): 320×240 native (76,800 pixels) is the highest of the five, and the trade-off is a 32° FOV — the narrowest here, so you back up further to frame a large panel. The official spec sheet says < 9 Hz, which smears when you pan. There is no internal battery or storage; the official figure is 280 mW drawn from the phone, which drains it noticeably over a long session. The lens is manually focused, so every distance change means re-focusing.

Flir C3-X (standalone handheld): 128×96 native is the lowest of the five, and the matching 7.9 mrad/pixel IFOV is the coarsest. FLIR lists "Digital Zoom: No". Temperature range tops out at 300°C, the lowest ceiling here. Image frequency is 8.7 Hz like the One Pro. What you get back is 4 hours of battery, IP54, 2 m drop rating and a 640×480 touchscreen — but if you need to separate adjacent small components, 128×96 runs out first. The listing showed "Only 5 left in stock - order soon" when checked.

Klein Tools TI250 (standalone handheld): 19,200 pixels (160×120) is entry level among handhelds here. The screen is 2.4 inches at 320×240. There is no Wi-Fi, so images come out over USB-C only. Thermal sensitivity of <60 mK is the weakest of the five. What it does offer is 32 GB of internal storage and continuously adjustable emissivity from 0.01 to 0.99.

Klein Tools TI290 (standalone handheld): the problem here is not a specification but a conflict. The official product page says "Sensor Resolution: 76,800 Pixels"; the official manual says "Sensor Resolution: 49152 pixels (256 × 192)" — a factor of 1.56 apart — and the Amazon title follows the 76,800 figure. The same manual notes the unit is not calibrated below -10°C, against a -40°C lower bound on the product page. Weight differs too: 235 g on the page, 213 g in the manual. Charging takes 3 hours. It does have a 3.5-inch screen, Wi-Fi transfer and ≤50 mK sensitivity, but go by the manual in the box.

Limits of this article

There is no thermal camera on this desk, so nothing here is a measured temperature. Every figure comes from manufacturer product pages, official spec tables, official manual PDFs and official export documentation, plus the Amazon listings as fetched on 2026-10-24. This article contains affiliate links, and I may earn a commission if you buy through them; that does not change which numbers are quoted or how they are ordered. This machine egresses from a Chinese IP, so Amazon's currency and stock figures are unreliable — treat pricing as whatever your own checkout page shows. The dollar figures quoted from official pages (Flir One Pro listed at $429.00, reduced to $349.00; C3-X at $569.00, reduced to $499.00, the latter marked US-only by FLIR) are here to indicate tier, not to serve as a quote.

If you are building out a bench, how to read multimeter CAT ratings covers static levels, where 100 MHz stops on an oscilloscope covers the time domain, and why a bench supply's watt rating is not V×A covers supply envelopes. A thermal camera adds the temperature-distribution dimension; it does not replace any of them. For a worked example, the M.2 SSD heatsink teardown was measured with a Flir One Pro, so you can see what this class of camera actually produces in a real test.

FAQ

Q: Is a camera advertised as 480×360 sharper than one advertised as 320×240?

A: Not necessarily — find the native figure in parentheses. 480×360 is 160×120 pushed through VividIR's 3x upscale (FLIR's own wording), whereas a native 320×240 carries more than twice as many real temperature samples. Compare native first.

Q: Is 8.7 Hz painful to use, and can I get a faster one?

A: For static measurements it barely matters; panning or filming moving parts smears. Faster variants do exist, usually under a different suffix (Seek's EAAX series is listed at >15 Hz by dealers), and price and available markets can differ. If you travel across borders with one, check how export classification applies.

Q: Why do aluminium enclosures and stainless pipe read far too low?

A: Usually emissivity. FLIR's preset for glossy surfaces is 30%, while most cameras default near 0.95 (the Klein TI290 manual gives 0.95 as the default). Change emissivity, or put matte tape or paint on the spot and measure that.

Q: Phone-attached or standalone?

A: It depends on how often and where. Phone-attached cameras (Flir One Pro, Seek CompactPRO) are cheaper and pocketable, but battery, storage and power all sit on the phone — about 1 hour for the One Pro, 280 mW drawn from the phone for the Seek. Standalone units (C3-X, TI250, TI290) work independently; the C3-X is rated at 4 hours, the TI290 has a 3.5-inch screen. You pay more for it.

Q: Does higher resolution mean more accurate temperatures?

A: No. Resolution governs spatial detail, while accuracy comes from calibration and conditions. Klein's own TI250 and TI290 both sit around ±2°C or 2%, and the real differences are pixel count, screen, Wi-Fi and the bottom of the temperature range.

👉 Join Xiaomi MiMo Platform: Leading AI model platform with cost-effective inference

👉 Join Aliyun AI: Top AI products with exclusive coupons for business innovation

📌 This article was AI-assisted generated and human-reviewed | TechPassive — An AI-driven content testing site focused on real tool reviews