The mark, the palette,
and the perceptual decay constants.
Scrutinizer's brand assets and the constants used in its perceptual model. Copy-pasteable for designers building peripheral-aware UI, and citation-ready for papers that use the same numbers.
1. The mark
The orange eye with three arrows above has been Scrutinizer's identity since the original 2007 implementation. In the Electron app it is the URL-bar toggle that turns the perceptual simulation on/off. The arrows are extramission: the eye actively projecting attention outward. Three of them echo the engine's three concentric attention zones (fovea / parafovea / periphery).
Now ships as a deterministic SVG generated by gen_scrutinizer_logo.py: flat-vector, scales crisply 16px → 512px, dimensional iris orb gradient preserved, animatable iris group (clipped to almond, so future animation can move the eyeball without escaping the eye bounds).
Raster sizes (rendered from canonical SVG; matches
scrutinizer2025/renderer/assets/scrutinizer_*_icon.png):
512 ·
128 ·
48 ·
32 ·
16 ·
toggle (64)
Dark-mode strategy
One canonical mark, one optional dark-tuned variant.
The orange (#ff9933) is the brand identity.
It has 9.27:1 contrast on the brand dark background and reads cleanly
on light backgrounds too, so the eye keeps its color in dark mode.
The only difference between
canonical and dark-tuned is the catchlight:
white in canonical (universally legible),
orange-ink (#ffb366) in the dark-tuned
version (softer on dark backgrounds where pure white can feel hot).
For app-icon contexts (Dock, Windows tray, favicon) always use
canonical. For SVG-in-HTML contexts you can swap via
prefers-color-scheme if you want.
1b. Perceptual instruments
Visualizations of what Scrutinizer does. The foveal overlay reproduces the in-app
UI bullseye (fovea, parafovea, periphery), ported from
scrutinizer2025/renderer/svg-overlay.js.
The engine sector cobweb shows the sectors the engine
samples (Blauch et al. 2026 isotropic cortical
sectors). Use these as figures, diagrams, or technical hero
images. The eye mark in §1 remains the brand identity.
Generated by muriel.tools.diagrams.foveal_overlay
and muriel.tools.diagrams.engine_sectors_overlay.
Pin sizes, themes, and parameters via the muriel CLI to regenerate.
2. Brand colors
Long-wavelength visibility orange, chosen because it survives peripheral processing (see §3). Cool ink on OLED near-black. All text-tier tokens clear muriel's 8:1 contrast minimum on the canonical background.
Full brand schema (semantic states, viz palettes, typography scale, a11y rules): scrutinizer-brand.toml via the muriel style-guide schema.
3. Perceptual decay constants
Why orange? Because red-green opponency dies faster in the periphery than blue-yellow does. The L-M (red-green) channel is foveally weighted: its sensitivity falls steeply with eccentricity. The S-(L+M) (blue-yellow) channel has dedicated wiring across the entire retina and survives ~5× further into the periphery. These constants come from the Scrutinizer model and match Mullen & Kingdom (2002) + Bowers et al. (2025) detection-threshold data.
Live demo (interactive): see Peripheral color: what designers don't know about red.
4. Foveation constants: Blauch et al. 2026 model
The cortical magnification function (CMF) parameters used in Scrutinizer's isotropic-sectors layout. Drop these into your own pipelines for sector assignment compatible with Scrutinizer's outputs.
Drop-in JavaScript:
const SCRUTINIZER_CONSTANTS = {
// Brand
accent: '#ff9933',
accent_ink: '#ffb366',
background: '#0a0a0f',
ink: '#e0e0ec',
gaze_teal: '#7dd4e4',
// Chromatic decay (red-green dies first)
sigma_lm: 120, // px @ 1080 viewport (~8° half-life)
sigma_by: 500, // px @ 1080 viewport (~40° half-life)
rg_decay_k: 0.085, // per-degree, Bowers 2025
// Foveation (Blauch et al. 2026 isotropic sectors)
cmf_a: 2.78,
fovea_radius_deg: 1.0,
ecc_scaling: 0.75,
num_rings: 50,
};5. Colorblind-safe palettes
Four named palettes that survive deutan / protan / tritan deficiencies,
shipped via muriel.palettes. Use Wong unless you
need matched-lightness families for multi-panel figures (Tol).
Wong (8 colors)
Tol Bright (7 colors)
Tol Vibrant (7 colors)
IBM Carbon (5 colors)
Import in Python:
from muriel.palettes import WONG, TOL_BRIGHT, palette.
CLI: python -m muriel.palettes --swatches out.svg.
6. Typography
Inter for everything except code, JetBrains Mono for code. System fallback stack so the site is legible even before fonts load.
--font-body: 'Inter', -apple-system, system-ui, sans-serif;
--font-mono: 'JetBrains Mono', 'Fira Code', ui-monospace, monospace;References
- Blauch, N. M., Alvarez, G. A. & Konkle, T. (2026). FOVI: A biologically-inspired foveated interface for deep vision models. arXiv:2602.03766. (Foveation model used by Scrutinizer.)
- Bowers, N. R., Gegenfurtner, K. R. & Goettker, A. (2025). Chromatic and achromatic contrast sensitivity in the far periphery. Journal of Vision 25(11):7. doi:10.1167/jov.25.11.7. (RG decay rate.)
- Mullen, K. T. & Kingdom, F. A. A. (2002). Differential distributions of red-green and blue-yellow cone opponency across the visual field. Visual Neuroscience 19(1), 109–118.
- Brown, R., Dutell, V., Walter, B., Rosenholtz, R., Shirley, P., McGuire, M. & Luebke, D. (2023). Efficient dataflow modeling of peripheral encoding in the human visual system. ACM Transactions on Applied Perception. doi:10.1145/3564605. (ECC_SCALING = 0.75.)
- Wong, B. (2011). Points of view: Color blindness. Nature Methods 8, 441.
- Tol, P. Color schemes. SRON technical note. Available at sronpersonalpages.nl/~pault/.
- IBM Carbon Design System. Data-vis colorblind palette.