COTS baseline comparison - 22 May 2026

Off-the-shelf cameras help validation, but they do not replace the custom optical head.

The EmberScope comparison separates fast buy-and-fly learning from the actual payload question: whether a compact, raw, radiometric camera can keep centimetre-scale sampling at useful survey altitudes.

Decision summary

The custom optics case rests on controlled sampling and integration, not on the absence of COTS thermal cameras.

4.0 cm Current custom-head GSD proxy at 80 m with an 18 degree horizontal field.
10.4 cm GSD proxy for a 45 degree, 640 pixel COTS thermal camera at 80 m.
17.2 cm GSD proxy for a 69 degree, 640 pixel COTS thermal camera at 80 m.
>1 m Lepton-class 95 degree, 160 pixel sampling at 80 m, before blur or fill factor.

Position

COTS hardware should be bought for learning, not mistaken for the final optical answer.

Vue Pro R, Zenmuse XT2, ICI 8640, thermoMap, Duet T, Lepton, and Boson-class products all help bound the problem. They can support calibration rehearsal, negative-scene collection, operator review, and safe small-hot-target surrogate testing.

None of the easy fixed-lens drone-camera branches simultaneously gives the current custom target's narrow survey field, raw radiometric access, platform independence, package control, window/baffle ownership, and future-proof detector path.

Baseline comparison

The fixed-lens COTS branches trade speed of deployment for coarser sampling or platform lock-in.

Baseline Role Radiometry Mass / power hook Optics hook 80 m GSD Judgement
Lepton 3.1R SWaP floor Radiometric model, but 160 x 120 at 8.6 Hz About 0.9 g and 150 mW operating Fixed 95 degree HFOV 109 cm Good for electronics prototypes; not a primary small-hot-target survey camera.
Boson+ custom head Preferred custom path Radiometric 640 x 512 branch with raw/pre-AGC integration path About 7.5 g bare core before carrier and interface margin Purpose-built 18 degree HFOV target 4.0 cm Best current route to controlled sampling and package ownership.
Vue Pro R 640 Legacy buy-and-fly validation camera Radiometric JPEG and 14-bit TIFF paths About 92-113 g; about 2.1 W in cited material 45 degree 13 mm or 69 degree 9 mm fixed lens examples 10.4-17.2 cm Useful field surrogate if available; too wide and too legacy for the final payload basis.
Tau 2 / ThermalCapture Legacy radiometric reference 14-bit radiometric recording path exists About 105 g camera class and <4 W in vendor material Fixed lens choices; 45 degree class common About 10.4 cm Available-equipment surrogate only; do not start a new payload around Tau 2.
Zenmuse XT2 Integrated DJI gimbal branch Radiometric FLIR thermal sensor with R-JPEG/TIFF/SEQ paths About 588 g for a 640 x 512, 13 mm unit in reseller specs DJI gimbal ecosystem, 45 degree 13 mm class About 10.4 cm Good operational reference; platform tie, gimbal mass, and fixed optics are the issue.
ICI 8640 P Strong lab/flight surrogate 14-bit raw/radiometric output with USB and SDK path About 72-75 g without lens and <1 W; lenses add mass 24.8 degree 25 mm or 17.6 degree 35 mm current lens examples 5.5 cm or 3.9 cm Closest COTS camera to the custom sampling problem, but packaging remains project work.
thermoMap / Duet T Mapping-system comparator Radiometric TIFF/R-JPEG class mapping outputs thermoMap cited at about 134 g and 5 W; Duet T is ecosystem-priced 9 mm thermoMap or 45 degree Duet T fixed-wing workflow 10.4-17.2 cm Useful mapping precedent, but not a direct compact custom-head substitute.

Near-term use

A COTS surrogate can still move the field program forward while custom optics mature.

Validation scenes

Collect rural no-fire backgrounds, roads, rocks, machinery, water, buildings, people, livestock, dust, and sun-warmed ground.

Calibration rehearsal

Exercise the warm and ambient field references, window checks, drift flags, and metadata required by the calibration strategy.

Small-hot-target surrogates

Run safe target trials at 35 m, 80 m, and 120 m so fill factor and operator-review issues are visible before custom hardware arrives.

Pipeline packets

Use COTS data to test raw-frame capture, geotagging, candidate patches, review labels, and replay manifests without claiming final payload performance.

Evidence base

The comparison uses current product pages plus the existing RFS detector and survey checkpoints.