illumination op• Data kinds: none → table (an op determined by its arguments alone — it takes no image or data input)
• Call: import illumdesign; illumdesign.illumination_design(surface='glossy', defect='topographic', slope_deg=10.0, part_size_mm=50.0, camera_height_mm=300.0) (or opsoptics.get("illumination_design"))
Rank the standard light families for a surface / defect pairing (`table`).
Candidates: low-angle ring (dark field, elevation 20°), high-angle ring
(bright field, 70°), the elevation that :func:lighting_sweep finds best,
dome, coaxial and (for `defect="edge"`) backlight. Each is scored by the
simulated Michelson contrast of the stated defect — `topographic`: a
smooth facet of *slope_deg* (a dent wall, a bump); `scatter`: a rough
patch (chipped edge, pit, fine scratch); `pigment`: an albedo patch at
half the surround; `edge`: a silhouette — **multiplied by the uniformity
of the background radiance the camera sees** across the part (min/max over
the centre and the four edge midpoints). That second factor is what
separates a robust choice from a fragile one: coaxial light on a glossy
part gives a huge negative contrast exactly on axis (the glare) and almost
none a few millimetres away, so its background uniformity is poor and it
ranks below a dark field whose background is uniformly dark. Irradiance
uniformity is reported too. The result lists the candidates best first
with their numbers, the `recommended family, and rule_of_thumb` — the textbook
choice (a smooth facet → coaxial bright field on a glossy finish, else the
ring elevation that mirrors it into the camera; scatter → dark field; pigment →
dome; edge → backlight) so a disagreement between simulation and rule is
visible rather than hidden. Two things the numbers say that folklore does
not: a *smooth* 10° facet does not light up in dark field (it mirrors
the low light away from the camera) — the "dark field shows scratches"
rule is about their rough flanks, the `scatter` class here; and for that
class a large coaxial (bright-field) source often scores *higher* than
dark field because the rough patch appears dark on a uniform glare with
contrast near 1 (the wafer / glass inspection practice). The model does
not score sensor saturation or the glare's dependence on part flatness,
which is why the dark-field rule survives on the shop floor; the table
shows both so the choice is made with the numbers.
Every optics op validates its input before computing (nothing slips through silently):
• Units are baked into the argument name — _mm / _um / _deg / _mrad. Confusing mm with µm does not crash; it yields a plausible-looking wrong answer, so the name prevents it. Nothing here guesses the unit from the magnitude.
• **Strings raise ValueError** — float('50') succeeds, so an unparsed configuration value would slip through as a length (measured: thin_lens('50', '200') returned a plausible 66.667 mm). bool is refused too, as the implicit promotion True == 1.
• **complex / masked arrays raise ValueError (real-valued slots only; silently dropping the imaginary part or peeling off the mask is refused). NaN/Inf raises ValueError on every input.**
• Division by zero and its relatives are refused by name: focal length 0, radius of curvature 0, refractive index <= 0, a fully opaque aperture (all zeros, so the normalisation is 0/0), a PSF whose sum is <= 0, a Stokes vector with S0 = 0, and an object sitting at the front focal point (the image is at infinity).
• Only two ops return a non-finite value, and both state it as a contract: depth_of_field returns far_mm = inf beyond the hyperfocal distance (that is what the hyperfocal distance means), and gaussian_beam returns wavefront_radius_mm = inf at the waist (the radius of curvature of a plane wavefront). Both also return a finite companion (far_is_infinite / curvature_per_mm). **Any other silent NaN/Inf is detected internally and raises ValueError** — "float64 overflowed" and "the answer is infinite" are different claims, so the first is never returned wearing the face of the second.
• Size caps: generated grids are capped by optics.MAX_GRID (4096); supplied fields/PSFs/apertures by optics.MAX_FIELD_ELEMENTS (2^24); ABCD element chains by optics.MAX_SYSTEM_ELEMENTS (1024); Zernike by MAX_ZERNIKE_TERMS (512) / MAX_ZERNIKE_ORDER (40) / MAX_ZERNIKE_BASIS (2^25). This closes, fail-closed, the paths where a small argument triggers a huge internal allocation (measured: n_max=40 × 4096² needs 108 GB).
• Physically impossible states are refused too: a Stokes vector with degree of polarisation > 1, negative transmittance, negative intensity, and invalid Zernike indices such as n-|m| odd.
• mv_illumination_practice — 照明の実務知識 — 波長・偏光・点灯方式・外光・安全
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• illumination_design_demo — py -3.11 examples/illumination_design_demo.py
table as input)abcd_matrix · wavefront_stats · paraxial_trace · seidel_coefficients · spot_stats · tolerance_analysis · wavefront_from_opd · spot_diagram
illumination)light_source · irradiance_map · illumination_uniformity · defect_contrast · lighting_sweep
*Provenance: illumdesign.py — OPTICS operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.