Most thermal payload datasheets are technically accurate and practically misleading. Nothing is invented; the padding comes from choosing favourable definitions, quoting best-case conditions and omitting the qualifier that would make a number comparable. Here is where to look.
Key takeaways
- Twelve spec lines account for most misleading comparisons — detection range, NETD, zoom, and accuracy are the worst offenders.
- Three questions separate a real spec from marketing: under what condition, to what standard, and measured or calculated.
- A stated zoom figure is meaningless unless optical and digital are separated; hybrid figures inflate the number several times over.
- Build the comparison table yourself from conditions you define — never from the vendors’ own summary rows.
On this page
The twelve lines that mislead most often
| Specification | How it is padded | What to ask instead |
|---|---|---|
| Detection range | Johnson criteria with generous target size | Which standard, and what critical dimension in metres |
| Hybrid / digital zoom | Optical × digital quoted as one number | Optical ratio and long-end focal length only |
| Thermal resolution | Upscaled output resolution quoted | Native detector array size |
| NETD | Best-case at 25 °C, f/1.0, 30 Hz | NETD at your operating temperature and frame rate |
| Minimum illumination | Measured at widest zoom, slowest shutter | Lux at maximum zoom, at usable shutter speed |
| Stabilisation accuracy | Static bench figure, pitch/roll only | In-flight, all three axes, stated vibration environment |
| Frame rate | Sensor rate, not encoded output rate | End-to-end displayed frame rate |
| Weight | Pod only, excluding GCU and cabling | Full installed mass including all boxes |
| Power | Average, ranging and tracking off | Peak with all functions active |
| IP rating | Rating of the housing, not the connectors | Full assembly rating, including mated connectors |
| Laser range | Against a large, high-reflectivity target | Range against a 20 % reflectivity vertical surface |
| Operating temperature | Storage range quoted as operating | Operating range with full performance held |
The three questions that separate real specs from marketing
“Under what conditions?” Almost every optimistic figure has a condition attached that was omitted. Minimum illumination at wide angle is a different product from minimum illumination at 40x. Ask for the condition and half the padding disappears.
“Native or processed?” Thermal resolution, frame rate and zoom all have a raw value and a processed value. The processed value is not fake — it is what you see — but only the native value predicts what the sensor can actually resolve.
“Installed or component?” Weight, power and ingress protection are frequently quoted for the pod alone. Your aircraft carries the pod, the control unit, the cabling and the mount. Ask for the installed figure, because that is the one that determines whether it flies.
Building a comparison that survives contact with two vendors
The reason payload comparisons collapse is that each vendor answers a slightly different question. The fix is to define the questions yourself and require both to answer yours.
Write a short specification response form, send it to every vendor, and refuse to accept a datasheet in place of it. Three columns: the specification, the condition under which it is measured, and the value. Vendors who cannot fill it in are telling you something useful.
Then verify the two or three numbers your mission actually depends on with a field trial. A payload comparison decided on paper is decided by whoever writes better datasheets; a comparison decided against a known target at a measured distance is decided by the hardware. See real DRI data from six payloads for what an honest field measurement looks like.
What to ignore first
Not every impressive number matters. Hybrid zoom figures above about 100x tell you nothing about capability. Onboard storage capacity is rarely a constraint. Number of palettes is a preference, not a specification. Detection ranges for “large vehicle” targets are irrelevant if you search for people.
The specifications that actually decide mission outcomes are narrower than most datasheets suggest: native thermal resolution, native frame rate, optical focal length, in-flight stabilisation accuracy, installed mass and installed power. Everything else is secondary. The longer argument is in what payload buyers should ignore first.
Related reading
Method: choosing a payload without spec noise and evaluating payload suppliers.
Technology: how DRI figures are calculated and what zoom ratio costs.
Specs in depth: DRI explained, NETD explained and hybrid zoom explained.
A worked comparison of two real datasheets
Take two payloads that both claim “1,500 m person detection”. Payload A quotes Johnson criteria against a 1.8 m target; payload B quotes EN 62676-4 against a 0.75 m critical dimension. The numbers are identical and the products are not remotely comparable.
Now normalise them. Ask both for native detector array, pixel pitch and long-end focal length. Payload A turns out to be 384×288 at 17 µm behind 25 mm; payload B is 640×512 at 12 µm behind 50 mm. Compute IFOV: 0.68 mrad versus 0.24 mrad. Payload B resolves nearly three times the angular detail. Everything else on the two datasheets is noise.
This is why we recommend building the comparison from three physical numbers rather than from any published range. Native array, pixel pitch and focal length are unambiguous, cannot be defined creatively, and between them determine what the payload can resolve. Once you have those you can compute any range figure you like, under whatever standard your requirement uses, for both products on the same basis.
- Ask for native array size, pixel pitch and long-end focal length from every vendor — three numbers, no adjectives.
- Compute IFOV yourself: pixel pitch ÷ focal length. Lower is better and it is directly comparable.
- Derive range at your chosen task level and target size, identically for every candidate.
- Only then look at the vendor’s own range claims, and treat any large discrepancy as a question to ask.
- UAV thermal camera calibration: chambers, blackbodies and checks
- Zenmuse H30T alternatives: DJI airframe or independent payload
FAQ
Why do two payloads with the same sensor quote different detection ranges?
Because the range is calculated, not measured, and the calculation depends on the criteria standard and the assumed target size. Johnson criteria with a 1.8 m target produce figures several times larger than EN 62676-4 with a 0.75 m target — same hardware, different question.
Is a higher hybrid zoom number better?
No. Hybrid zoom multiplies optical zoom by digital magnification, and digital magnification adds no information. Compare optical ratio and long-end focal length; ignore the hybrid figure entirely when assessing range capability.
What single specification should I verify in a field trial?
Whichever one your mission depends on — usually recognition range against your actual target type. It is the specification most affected by definitional choices and the one where paper comparisons diverge most from field results.