Skip to main content

Color Science

Covert Recon Mode Validation

Color Palette — Live Calculations

WCAG 2.2 Contrast Matrix

All foreground/background pair contrast ratios. AA normal ≥ 4.5:1, AA large ≥ 3:1, AAA normal ≥ 7:1, AAA large ≥ 4.5:1.

Color Pair

Scotopic Vision Science

Why Red Light Preserves Dark Adaptation

Human scotopic (dark-adapted) vision is mediated by rhodopsin, a photopigment in rod cells with peak sensitivity near 507 nm (blue-green). Red light in the 620–700 nm range falls at the extreme low end of rhodopsin’s absorption curve, meaning rod cells are nearly insensitive to it. This allows operators to read instruments and displays without bleaching rhodopsin, preserving the ability to detect dim targets in peripheral vision.

Three Channels, Not Two

Rods and cones are not the whole retina. Intrinsically photosensitive retinal ganglion cells (ipRGCs), containing the photopigment melanopsin, form a third channel that peaks near 490 nm and drives circadian regulation rather than image formation. Its action spectrum is normative in CIE S 026/E:2018, which defines melanopic equivalent daylight illuminance (mEDI). This channel is why blue-enriched light at night suppresses melatonin — the mechanism behind consumer “night mode” features.

For an operator working at 02:00 this matters twice over: long-wavelength red is far off-peak for both rhodopsin (507 nm) and melanopsin (~490 nm), so it preserves dark adaptation and imposes minimal circadian load. We do not reproduce the melanopic action spectrum here — it is normative in a paywalled standard, and this page quotes no melanopic figure it cannot derive from published tables.

Rod Load at Equal Perceived Brightness

Comparing colours only makes sense at matched perceived brightness. Normalising by the photopic function V(λ) and taking the ratio of scotopic to photopic efficiency against 630 nm red gives the relative rod load:

WavelengthRod load vs. 630 nm
490 nm (melanopsin peak)~350×
507 nm (rhodopsin peak)~180×
555 nm (photopic peak)~32×
630 nm (this palette)

Method and provenance: computed as [V′(λ)/V(λ)] normalised to 630 nm, from the CIE 1924 photopic and CIE 1951 scotopic luminous efficiency functions at 5 nm in energy units, fetched from the CVRL database (Colour & Vision Research Laboratory, UCL) and committed to this repository at docs/data/cie/. The calculator on this page reads the same tables, so the prose and the tool cannot drift apart. Exact figures: 345× at 490 nm and 31.9× at 555 nm; 507 nm is the conventional rhodopsin peak but does not sit on the 5 nm grid, falling between 195× (505 nm) and 158× (510 nm). Values are rounded to two significant figures because the underlying tables are quoted to three or four, not because the computation is uncertain.

A correction worth recording: an earlier revision of this page interpolated a coarse table linearly across the far red, where V′(λ) decays roughly exponentially. That over-read V′(630) by 25% — 4.185×10−3 against the true 3.335×10−3 — which understated the advantage of red rather than overstating it. Moving to 5 nm authoritative anchors reduces the residual interpolation error across that tail to roughly 2%.

Intensity Dominates Colour

The single most actionable finding is not about hue at all: brightness matters more than colour. A bright red display will disrupt dark adaptation more than a dim blue-green one. Wavelength choice is a second-order optimisation on top of getting luminance low — it is not a licence to run the display bright. Dim the screen first; the palette is what protects you after that.

Why Aviation Uses Blue-Green — and Why That Is Not This

Modern military cockpits specify blue-green, not red, which looks like a contradiction until you ask what is being protected. Image intensifier tubes are strongly sensitive to the red and near-infrared — red cockpit lighting blooms the goggle. So NVIS-equipped aircraft use blue-green lighting; U.S. Navy and Army aircraft without NVIS still use red. The constraint is the equipment, not the eye.

An analyst reading a display in a dark room has no image intensifier to protect. Adopting blue-green here would import a constraint that does not apply and surrender the one property that does — minimal rod and melanopic load per unit of perceived brightness. Covert Recon Mode stays red deliberately, and on measurement rather than tradition.

MIL-STD-1472H — Human Engineering

Section 5.5.3.6 establishes requirements for dark adaptation and night vision conditions, specifying that ambient illumination shall be controllable and compatible with dark-adapted operations. Section 5.5.3.6.3 defers NVIS-compatible lighting requirements to MIL-STD-3009. This is the primary standard governing human-system interface design for U.S. Department of Defense systems (September 2020). The archived reference PDF (MIL-STD-1472G.pdf) contains the H revision.

MIL-STD-3009 — NVIS Compatibility

MIL-STD-3009 governs lighting and display compatibility with Night Vision Imaging Systems (NVIS). It defines spectral radiance limits to prevent cockpit/vehicle lighting from overwhelming image intensifier tubes. While primarily an aviation and vehicle standard, its spectral requirements inform any display intended for use in light-controlled environments.

Honest Limitation

Consumer LCD and OLED displays cannot produce monochromatic 630 nm red. They use broadband RGB subpixels, meaning even a “pure red” hex value (#cc2020) produces a spectral power distribution that includes some energy outside the 620–700 nm band. True MIL-STD-3009 compliance requires narrow-band optical filters or dedicated NVIS-compatible displays. The Covert Recon Mode palette is designed to approximate the intent of these standards on consumer hardware — it is not a substitute for dedicated tactical lighting equipment.

Standards Citations

  • MIL-STD-1472H — Department of Defense Design Criteria Standard: Human Engineering (September 2020). Section 5.5.3.6: Dark adaptation and night vision conditions.
  • MIL-STD-3009 — Lighting, Aircraft, Night Vision Imaging System (NVIS) Compatible. Spectral radiance limits for NVIS-compatible displays.
  • CIE 1951 Scotopic Luminosity Function V′(λ) — Commission Internationale de l’Éclairage. Standard observer for scotopic (rod-mediated) vision, peak at 507 nm. Tabulated values here are from Wyszecki & Stiles (1982), Table I(4.3.2), as distributed by CVRL (scvle.csv); the CIE 1924 photopic companion is vl1924e.csv. Both are committed at docs/data/cie/ so this page’s arithmetic is reproducible without a network fetch. The normative publications are CIE 018:2019 Table 2, DOI 10.25039/CIE.DS.gr6w4b5g (scotopic) and Table 1, DOI 10.25039/CIE.DS.dktna2s3 (photopic); the committed CVRL files reproduce them exactly at every 5 nm point used here, confirmed by independent retrieval from four sources.
  • CIE S 026/E:2018 — System for Metrology of Optical Radiation for ipRGC-Influenced Responses to Light. Defines the melanopic action spectrum (peak ~490 nm) and melanopic equivalent daylight illuminance (mEDI). Normative source for the circadian channel; not reproduced on this page.
  • WCAG 2.2 Contrast Requirements — W3C Web Content Accessibility Guidelines. Success Criteria 1.4.3 (AA) and 1.4.6 (AAA) for text contrast ratios.
  • ISO 9241-3 — Ergonomic requirements for office work with visual display terminals. Display ergonomics including luminance contrast and color requirements.
Straight talk about your data.

We use two cookies, both essential:

  • _csrf — Prevents cross-site request forgery. Required for form submissions. Security-only.
  • _dns_session — Only exists if you choose to sign in. No account required to use DNS Tool.

We log your IP address for two reasons: rate limiting (so nobody abuses the service) and security (identifying malicious actors and complying with legal obligations). We check source geography for analysis accuracy — DNS responses vary by region, and knowing which resolver answered from where makes the science better.

No tracking cookies. No analytics cookies. No ad networks. No data brokers. Our code is open-core — the application framework is publicly available under BUSL-1.1 with timed Apache-2.0 conversion. Verify it yourself.

Public analyses are archived by a third party. When a scan is not marked private or run as /dev/null, its report URL is submitted to the Internet Archive, which creates a permanent public record outside our control. That is deliberate — it gives every published analysis tamper-evident provenance independent of us — but it is not reversible, so it is stated here before you scan rather than after.

If you create an account and want out, account deletion removes your login and scan history. Public domain analyses remain available because they contain only public DNS records, already hashed. Full details: Privacy Policy.