Methodology: how we score thermal cameras

How the ranking is calculated

Each device is scored on twenty-four criteria grouped into five families: sensor and image, optics and spatial resolution, measurement and reliability, usability and reporting, ergonomics and ruggedness. Scores are recalculated on every display from the published specs.

The scoring scales are fixed. Adding a device to the database never changes another device's score, which lets you cite a score and find it again later.

Devices are only compared within their own Segment.

Weightings by use case

Use case	Sensor	Optics	Measurement	Usability	Ergonomics
Building diagnostics and energy audits	30	20	25	15	10
Electrical panels and switchgear	25	25	25	10	15
Electronics repair	30	30	25	10	5
Leaks and moisture	35	15	25	10	15
Mechanical and industrial maintenance	25	20	30	15	10
Solar (photovoltaic)	30	25	25	15	5
Automotive and electric vehicles	25	25	25	10	15

These weightings reflect a technical judgment call, debatable and published for that very reason. The comparator lets you replace them with your own.

Missing data

No value is ever estimated. When a manufacturer does not publish a spec, the corresponding criterion drops out of the calculation and lowers the confidence index.

This index is weighted by the use case being viewed. A missing criterion costs in proportion to its weight in that profile: not publishing a camera's weight hurts a lot for fieldwork and barely at all for solar work. Confidence is therefore specific to each use case, and a device well documented for building work can still be excluded for electronics repair for lack of information about its macro lens.

Three display levels follow from this:

• Above 75%: the score is displayed normally.
• Between 55 and 75%: the score is preceded by the ≈ sign and shown in gray. This is a partial estimate, accompanied by the list of missing specs.
• Below 55%: no score is displayed.

This choice is deliberate: a discreet manufacturer should not receive a bad score, which would suggest a bad product. It should receive a degraded score, or no score at all.

Every product page also shows a raw, unweighted rate: the share of specs the manufacturer actually publishes, across all use cases. It describes the product's documentation and does not depend on any scoring scale.

There is one exception. On criteria that specifically concern the manufacturer's candor, such as the native thermal resolution, not publishing counts as zero rather than being neutralized. Without this, staying silent would pay off more than telling the truth.

What we have not done

The listed devices have not been physically tested. The values come from manufacturer documentation, user manuals, and tests published by third parties, all identified on each product page along with the date they were checked.

There are no user reviews, no editorial score, and no "editor's pick" selection.

Illustrations

All images on this site are original illustrations generated from our own data. No product photograph belonging to a manufacturer or a stock photo library is reproduced.

Price range

Every product page can show an indicative price reference, from € to €€€€€, established from public prices cross-checked on a given date, excluding promotions and any specific retailer. For some industrial or rare models, in the absence of a public price, the bracket relies on a distributor's offer or a similar catalog price. This is not a selling price: it changes, it varies from one retailer to another and depending on configuration (lens, frame rate, kit), and it can be missing for some devices. We fill in this information gradually rather than estimating a value for devices that don't yet have one.

Symbol	Price range	Approximate positioning
€	under €200	Entry level / smartphone
€€	€200 to €500	Enthusiast / routine diagnostics
€€€	€500 to €1,000	Mid-range
€€€€	€1,000 to €2,000	Professional
€€€€€	over €2,000	High-end / expert

The positioning shown is a rough market indicator, not a ranking: a device's comparison segment depends on its specs, not its price.

This price range plays no part in any score. This site's ranking relies exclusively on technical specs, whether a device costs €30 or €8,000.

Thermal camera selector: exclude, rank, explain

The thermal camera selector starts from a need expressed in plain language and translates it into verifiable rules. It uses no data beyond the catalog and changes no scoring scale: the use-case score it displays is exactly the comparator's.

1. Exclude unsuitable devices: conditions, not deducted points

A device is excluded if it fails to meet one of the stated conditions: actual form factor read from the record (standalone camera, smartphone module or thermal phone, fixed installation) rather than inferred from the segment, budget compared to its price range, measurement use requested but explicitly unavailable (non-radiometric images, no report generation), temperature measurement requested for what is an observation device, discontinued model (unless the visitor includes them), a published minimum focus distance greater than the working distance, or a target smaller than the measurable area at that distance (more than twice smaller than the smallest visible detail when only spotting). Fixed-installation cameras are only offered when the visitor selects that form factor.

A device is placed "to check", neither excluded nor ranked, when essential data is missing: form factor not specified, price range not determined while a budget is set, radiometric export or report generation not documented while requested, field of view not published (spatial resolution cannot be calculated), minimum focus distance not published for work under 30 cm, macro lens working distance not published when the result depends on it, or documentation insufficient to publish a use-case score. Between 30 cm and 1 m, an unpublished focus distance is merely flagged. We never estimate a missing value.

The calculation

The suitability score is 60% the use-case score and 40% a resolution score at the stated distance. This second score compares the target size to the size resolved by the device, i.e. the recalculated IFOV multiplied by the distance. When the visitor needs to read temperatures, the resolved size is multiplied by five. A 3× IFOV target is a common minimum spot-size approximation; the selector deliberately uses 5× as a conservative rule that leaves more margin for focus, alignment, optics and thermal contrast.

Target ÷ resolved size	Resolution score
4 or more	100
1 to 4	60 to 100, on a logarithmic scale
0.5 to 1	0 to 60, only for spotting without temperature measurement
under 1 when measuring, under 0.5 when only spotting	device excluded

Any distance, target or budget

Visitors can set the working distance, target size and budget to “any”. If the distance or the target is set to any, no distance, focus or target-size condition is applied, and the score becomes the use-case score alone (instead of 60 % score and 40 % resolution). The ranking is still made by level of use. It is less precise: each result says so and shows its measurable area at several reference distances, limited to its minimum focus distance when published. A device kept for a specific distance is always kept when the distance is set to any. General-purpose use starts with any distance and target, because no single value represents it.

For work at 10 cm or less, the macro lens is only taken into account when its spot size (or smallest measurable area) and its working distance are published, and when that distance matches the stated one within ±20 %: a macro designed for 8 mm is never counted for work at 10 cm. Having a macro does not exempt the standard lens from the focus check. The selector offers 1, 2 and 3 cm distances to assess a device with its macro. If the lens is sold as an option, the result says so.

2. Rank eligible devices, one level at a time

Each segment has its own scoring scale: the same specifications do not earn the same points in S1 and S5. Retained devices are therefore ranked by suitability score, then by documentation confidence, only within the same segment, shown as a level of use: occasional use (S1: phone modules, thermal phones and entry-level cameras; S3: visual thermometers and compact imagers), regular use (S2: general-purpose handheld cameras), professional use (S4) and expert use (S5). Fixed-installation cameras form a separate group. No ranking mixes two scales. In the comparator, sorting by score likewise groups devices by segment. The "best thermal camera for…" guides apply exactly these rules to the typical situation of each use case (table below).

3. Explain every result

Each retained device comes with three sections computed from its data: why it fits (use-case score, area resolved at the stated distance, native sensor, margin when the target is at least four times larger than the resolved area, documented radiometric export and reporting); its limits (distance beyond which the target becomes smaller than the resolved area, then the exclusion distance, reliance on the macro lens, NETD above 80 mK, frame rate of 9 Hz or less, dependence on a smartphone, discontinued model, partial score); and missing data relevant to the current search. These explanations come with every retained device, including in the table, where they expand row by row. They come from fixed rules published here, not written case by case.

Default proposed values

Each use case pre-fills a typical distance and target size. These are editorial reference points, shown and adjustable, not measurements.

Use case	Distance	Target size	Example
Building diagnostics and energy audits	3 m	10 cm	thermal bridge, insulation defect
Electrical panels and switchgear	1 m	2 cm	terminal, connection
Electronics repair	10 cm	2 mm	component on a board
Leaks and moisture	2 m	10 cm	pipe path, damp area
Mechanical and industrial maintenance	1 m	3 cm	bearing, shaft
Solar (photovoltaic)	5 m	15 cm	faulty cell
General-purpose or other use	any	any	common hot spot

General-purpose use applies an equal 20% weighting to each of the five families, just as the comparator's sliders allow. Observation devices do not receive a score in this measurement selector. They now have a specialized ranking based on detection, sensitivity, endurance and ruggedness.

For inspection use cases U1 to U6, devices classified as observation are not suggested, even when the visitor only wants to spot a heat difference: this method cannot establish their suitability at the selected distance and target size. Selecting observation shows a general list filtered by form factor and budget; distance and target size are not applied. Use the specialized ranking to compare observation devices.

Limitations

A suitability score depends on the visitor's answers: it cannot be compared from one search to another or from one level of use to another, and is not a general ranking of devices. It accounts for neither the exact price, nor availability, nor any hands-on handling of the devices. Documented software or reporting does not guarantee the specific functions you need (file formats, report content, compatibility): check them with the manufacturer.

Commercial relationships and ranking

To date, this site has no affiliate links, no commercial partnership, and no payment tied to the listed devices.

If paid links are ever introduced, they will be flagged next to each link concerned and on this page. They will never enter any ranking formula: the calculation uses only technical specs.

Specialized rankings: four independent methods

Profiles U1 to U6 remain reserved for measuring instruments. Firefighting cameras, thermal drones, observation devices and optical gas imaging cameras use separate scales; their scores are never mixed into the main comparator.

• Firefighting and rescue: thermal safety, visibility, ruggedness, handling and transmission.
• Drones: radiometric data, aerial resolution, mission, platform and software. Complete aircraft and payloads are separate.
• Observation and hunting: sensitivity, detection, angular resolution, endurance, ruggedness and optical aids.
• Gas detection (OGI): gas sensitivity (30%), resolution (20%), cooled detector and NETD (20%), onboard quantification (10%), ATEX certification (10%), ergonomics and battery (10%).

As in the main comparator, an unpublished specification is excluded from the calculation. Weighted confidence controls display: a full score from 75%, a partial estimate from 55 to 74%, and no score below 55%.

Scoring-scale changelog

Every change to a scoring scale is dated and logged. Without this, the promise of stable scores would be worthless.

• v3.6 — 2026-10-22: working distance, target size and budget can be set to “any”. Without a distance or target, the selector ranks devices on the use-case score alone, by level of use, and shows their measurable area at several reference distances. General-purpose use starts with any distance and target. Use-case scores are unchanged.
• v3.5 — 2026-10-21: nine documented catalog additions and six specialized additions. New automotive profile U7 (25/25/25/10/15; base A and electrical B internal scales). U1–U6 weights, anchors and scores unchanged. Fixed-camera guide has no new numerical ranking.
• v3.4 — 2026-09-29: selector and comparator fixes. Temperature measurement now requires the target to cover at least the whole measurable area (threshold 1 instead of 0.5); the 0.5 threshold now only applies to spotting without measurement. The macro lens is only used when its published working distance matches the stated distance within ±20 %, and it no longer exempts the standard lens from the focus check; 1, 2 and 3 cm distances added. Rankings grouped by segment (level of use) in the selector, guides and comparator, because each segment has its own scoring scale. Explanations available for every table row. Use-case scores are unchanged.
• v3.3 — 2026-09-28: selector v2.0 in three steps. Form factor is now read from the record’s "form" field (instead of being inferred from the segment), which separates thermal phones and fixed-installation cameras; new requirement on measurement use (radiometric images, report); results and guides ranked separately by form factor; every result explains why it fits, its limits and the missing data. Weightings, scoring scales and use-case scores are unchanged.
• v3.2 — 2026-09-24: added six smartphones with a built-in thermal camera, placed in segment S1 alongside smartphone modules: Cat S62 Pro (discontinued), Blackview XPLORE 1 Pro, Ulefone Armor 27T Pro, Ulefone Armor 28 Ultra Thermal, Ulefone Armor 29 Pro Thermal and AGM G3 Pro. New rule: when the native resolution is not consistently published, it is shown as not published rather than estimated; the “native pixels” criterion then drops out of the calculation, the score becomes partial and a “Native resolution not published” label is shown. The advertised resolution is kept as the displayed resolution. The catalog now contains 179 devices from 38 brands and 323 sources.
• v3.1 — 2026-09-24: added six documented references: FLIR T865, Guide PS610, Guide PS400, Fluke Ti450 PRO (discontinued, kept for the used market), Seek Thermal Compact and TOPDON TC001 Plus. The catalog now contains 173 devices from 34 brands and 303 sources. Resolutions obtained through super-resolution or software processing remain separate from native pixels. Scoring scales are unchanged.
• v3.0 — 2026-09-24: launched three independent specialized rankings: firefighting and rescue, thermal drones (systems and payloads separated), and observation and hunting. The six measurement profiles and their scores are unchanged.
• v2.8 — 2026-09-21: added seven documented references: HIKMICRO Mini3, Fluke TC01A, Seek Thermal Nano 300, Guide MobIR ES2+, FOTRIC TK7, Optris PI 400i LT and FLIR A70 Smart Sensor Advanced. The catalog now contains 167 devices from 34 brands and 286 sources. Software-enhanced resolutions remain separate from native pixels, and configuration options are not presented as standard equipment.
• v2.7 — 2026-09-21: added five documented references: testo 860i, TOPDON TC005, FLIR ETS320, Optris Xi 410 LT ETH and Pulsar Telos LRF XL50. The catalog now contains 155 devices from 34 brands and 252 sources. The Pulsar remains excluded from measurement rankings because it is not radiometric.
• v2.7 — 09/24/2026: removed the “HIKMICRO M15 Explorer” record. Its source, a mislabelled retailer page, actually described the Explorer E20 smartphone camera, already listed; the real HIKMICRO M15 is a 4G trail camera with no thermal sensor. The address is kept as a guidance page. The catalog now has 167 devices. The entry is archived in data/retires.json. The specialized rankings were extended (firefighting and rescue, drones, observation and hunting) and their displayed scoring weights aligned with the actual calculation.
• v2.6 — 09/17/2026: added 12 devices to the catalog (FLIR E52/E54/E76/E86, Fluke Ti480 PRO/TiX580/PTi120, testo 883-2, HIKMICRO M30, Chauvin Arnoux C.A 1954 DiaCAm2, Guide PC230, Bosch GTC 600 C), all verified against manufacturer or distributor datasheets. The catalog now has 150 devices and 247 sources. Corrected an identity error: the device cataloged as "testo 883" is in fact the testo 883-1 (reference 0560 8830); its ID and name have been corrected accordingly, with no change to its specs.
• v2.5 — 09/16/2026: added the thermal camera selector and its suitability score (60% use-case score, 40% resolution at the stated distance). Use-case scores and scoring scales do not change.
• v2.5 — 09/16/2026: removed a duplicate entry for the Testo 872s, which had no source of its own and contradicted the manufacturer's own datasheet. The catalog now has 150 devices. The entry is archived in data/retires.json.
• v2.4 — 09/16/2026: price range filled in for 149 of 150 devices, from a survey of public and distributor prices (manufacturers, specialist resellers, price comparison sites). Two devices remain ND pending a reference check. No effect on scores.
• v2.3 — 09/16/2026: price brackets expanded from four to five tiers (under €200, €200 to €500, €500 to €1,000, €1,000 to €2,000, over €2,000). Devices whose documented price does not allow a choice between two new brackets revert to ND pending verification. No effect on scores.
• v1.1 — 09/14/2026: the resolution-honesty criterion now scores 0 when a manufacturer advertises interpolation without publishing the target resolution. Previously the criterion dropped out of the calculation, which favored opaque manufacturers.
• v1.1 — 09/14/2026: storage tier set at 90 points for internal memory of 64 GB and above.
• v1.1 — 09/14/2026: the confidence index now excludes criteria marked NA from the denominator.
• v2.0 — 09/14/2026: six use-case profiles instead of five. The "night observation" profile is removed from the measurement comparator (it belongs to monoculars) and now has an independent specialized ranking. Added "electronics repair", "mechanical and industrial maintenance" and "solar (photovoltaic)". The former "electrical" profile is split: panels and switchgear on one side, boards on the other.
• v2.0 — 09/14/2026: added two macro criteria to family B (capability and minimum measurement size), with internal weights per profile.
• v2.0 — 09/14/2026: drop-resistance anchors become specific to each segment. A 20 g module and a 380 g camera are not judged on the same scale.
• v2.1 — to do: the "micro_balayage_optique" value of the interpolation-method field needs renaming. It now covers two different techniques that genuinely add information: optical micro-scanning and testo's multi-shot super resolution. Both stand in contrast to single-image resampling, which adds nothing.
• v2.2 — 09/14/2026: the confidence index is now WEIGHTED by use case. A missing criterion costs in proportion to its weight in the profile being viewed. Not publishing a camera's weight hurts a lot for the fieldwork profile and barely at all for solar.
• v2.2 — 09/14/2026: confidence becomes specific to each use case. A device well documented for building work but incomplete on its macro lens remains scored for building diagnostics and is only excluded for electronics repair.
• v2.2 — 09/14/2026: three levels instead of two. Above 75%: full score. Between 55 and 75%: score shown as a partial estimate, with the list of missing specs. Below 55%: no score.
• v2.2 — 09/14/2026: the Confidence column in the Calculations tab remains a raw, unweighted rate, identical for all use cases. The site additionally calculates a per-use-case confidence. The two intentionally coexist: the raw rate describes the product's documentation, the weighted confidence decides whether a score is displayed.