Optical literature synthesis - 31 May 2026

The literature points to the next EmberScope optical branches.

Fast freeform LWIR precedents support the custom reflective path, but they also show why the current F/5 proof geometry is a package baseline rather than a prototype prescription.

Review next branches Open current optical eval Open mechanical screen

Dated synthesis (31 May 2026) — one recommended branch has since been run.

This synthesis recommended testing faster LWIR branches, including a four-mirror reflective family. The four-mirror branch has since been fully optimized (10 June 2026, see the optics hub): it closes image quality at F/5 with conic surfaces only. The literature case for faster systems (F/1–F/2) is now the live question, because diffraction at F/5 limits single-pixel energy regardless of prescription quality.

Design decision

The next prescription pass should test fast LWIR branches, not keep polishing the slow proof geometry.

F/1.5-F/2 Fast uncooled-LWIR regime repeatedly used by the closest freeform precedents.

18-50 mm Useful first focal-length range for balancing swath against centimetre-scale GSD.

3 or 4 Mirror-count families worth evaluating before locking the compact reflective baseline.

150 mm Payload box still has to include baffles, window, detector, filter, focus, and service margin.

Evidence read-through

The strongest papers agree on compact reflective optics, but they make fabrication and stray light part of the prescription.

Evidence	What it shows	What EmberScope should do next
Duveau alpha-Z LWIR	Three-mirror, 18 mm, F/1.5, 18 deg x 24 deg-class uncooled LWIR prototype with diamond-turned mirrors and 5-axis housing.	Create a fast alpha-Z branch and include non-sequential housing and baffle stray-light checks.
Fuerschbach phi-polynomial LWIR	F/1.9, 30 mm pupil, 10 deg full-field freeform reflective imager coupled to an uncooled microbolometer.	Create a narrow or mid-field high-GSD branch with explicit freeform and alignment variables.
Freeform assembly paper	As-built LWIR diffraction-limited performance depends on real datums, shim/focus strategy, and secondary mirror compensation.	Treat alignment features, focus compensation, and monitored field points as prescription fields.
Yuan airborne LWIR spectrometer	F/2, 8-12.5 um airborne optics tie optical design to calibration, thermal background, and laboratory NEdT testing.	Keep window, filter, calibration, and thermal-state allowances inside optical readiness gates.
Bauer/Papa/Rolland roadmap papers	Three- and four-mirror freeform systems have structured solution spaces with volume, etendue, stop placement, sag, slope, and metrology tradeoffs.	Run branch-level specification sweeps before local tuning of any single pretty geometry.

Next optical branches

Four candidate families have enough evidence to deserve a fair screen.

Fast alpha-Z wide-search

Start near F/1.5 with 18-25 mm focal length. This is the strongest wide-swath uncooled-LWIR precedent.

Fast three-mirror TMA

Use the Fuerschbach and Druart branches for F/1.5-F/1.9, 30-50 mm focal length, and better GSD.

Narrow high-GSD TMA

Test whether a 75-100 mm class branch is needed for centimetre-scale hot targets despite reduced swath.

Four-mirror reflective

Try one stop-placement and pupil-control branch before rejecting four mirrors on complexity alone.

Sampling pressure

F-number and focal length drive both weak-target signal and pixel sampling.

Choice	First-order consequence	Design reading
F/1.5 at 10 um	About 36.6 um Airy diameter, or roughly 2.2 pixels on 17 um pitch.	Fast enough to stay in the detector-energy discussion for weak small targets.
F/5 at 10 um	About 122 um Airy diameter, or roughly 7.2 pixels on 17 um pitch.	Useful package proof, but too slow to treat as the final fire-camera prescription.
18 mm focal length	About 9.4 cm GSD at 100 m AGL with 17 um pixels.	Good search swath, marginal for centimetre-scale targets unless altitude is low or contrast is high.
34 mm focal length	About 5.0 cm GSD at 100 m AGL with 17 um pixels.	Promising midpoint; at F/1.5 the entrance pupil is still about 23 mm before packaging allowances.

Prototype-readiness gates

The literature changes what must be visible before a candidate can be called prototype-ready.

Each candidate needs detector-energy, OPD or MTF, and stray-light checks beside the existing ray-trace metrics. A clean spot table is not enough if the housing, window, detector, or stop geometry creates an unmodelled thermal path.

The same record must carry sag, slope, freeform departure, material, metrology, datums, focus compensation, and expected alignment motions. Those are design variables for a drone payload, not paperwork after the optical form is chosen.