Monet 1864–1926 — Color Kernel (Merged)

A color kernel reading of Monet

Russell Parrish · Artist Influencer · artistinfluencer.com · Images: WikiArt CDN + Wikimedia Commons · deduplicated

What This Shows

Two measurements extracted from 341 unique Monet paintings across six decades, plotted against time. Left panel: how broadly distributed hue is across each painting, not how vivid but how wide the angular spread of color decisions is across the full hue wheel. Right panel: how often strong color edges occur where no luminance edge runs, color making structural marks that light is not making.

Both computed in OKLab, a perceptually uniform color space. The system is behavioral, not volumetric; it measures how Monet used color, not how much. The two metrics are structurally independent: hue entropy captures palette breadth; CBI captures color acting structurally. Neither predicts the other.

He did not simply intensify. He expanded to a peak, then decoupled color from light. Those two things happened in different decades.

Five Periods

Early ≤1871	HE 0.623 · CBI 0.121	Narrowest hue palette. Color anchored to subject and light; 3% of paintings reach individual CBI ≥ 0.20. n=37.
Impressionist 1872–80	HE 0.673 · CBI 0.155	Hue field opens with outdoor light. High-key fraction peaks (0.388), the most genuine whites and tints of any period. 27% reach CBI ≥ 0.20. n=94.
Series I 1881–90	HE 0.720↑ · CBI 0.143	Career peak for hue breadth. Wide subjects (ocean, limestone cliffs, storm light) pull across more of the hue wheel than enclosed riverscapes can. CBI dips below Impressionist: breadth was subject-driven. 19% reach CBI ≥ 0.20. n=123.
Series II 1891–1905	HE 0.622 · CBI 0.180↑	Palette narrows as subjects concentrate; CBI peaks. Composition froze; fog and mist did the structural work (top CBI: Charing Cross Bridge, Rouen Cathedral Magic in Blue, Houses of Parliament Fog Effect). 36% reach CBI ≥ 0.20, the highest rate of any period. n=59.
Late Giverny 1906+	HE 0.690 · CBI 0.167	Broader hue than Impressionist, darker tonal ground (high-key 0.216 vs 0.388). Color partially re-anchors to luminance. 25% reach CBI ≥ 0.20. n=28. (9 of 28 are Venice 1908, mean HE 0.762; without Venice, Water Lilies cluster ≈ 0.656.)

Two Arcs, One Career

Top row: Hue Entropy (↑ = career peak at Series I). Bottom row: CBI (↑ = career peak at Series II).

Early

0.623

0.121

→

Impressionist

0.673

0.155

→

Series I

0.720↑

0.143

→

Series II

0.622

0.180↑

→

Late Giverny

0.690

0.167

high_key (tonal lightness): peaks Impressionist, declines every period after: 0.326 → 0.388↑ → 0.349 → 0.320 → 0.216

What the Axes Measure

Left Panel

Hue Entropy

Shannon entropy of a chroma-weighted 36-bin hue histogram. Chroma gate (C ≥ 0.02) excludes near-neutral pixels. High = color spread across the full hue wheel. Low = palette concentrated in one or two hue families. Career range: 0.26 to 0.94.

Right Panel

CBI

Chromatic Boundary Independence. Strong chroma-gradient pixels where luminance gradient is weak: top 15% chroma edge vs. bottom 30% luminance edge. Color drawing lines that light doesn't draw. Individual threshold ≥ 0.20 flags meaningful color autonomy.

Navigation & Controls

Scroll wheelZoom toward cursor, each panel independent

Click + dragPan within panel

Double-clickReset zoom and pan

HoverThumbnail, title, year, HE, CBI, high-key fraction

Period togglesIsolate periods; stats bar updates to visible data

Genre togglesFilter by subject

Fixed / rescaleRescale axes to the visible data range

Tooltip Metrics

Metric	What it measures
`hue_entropy`	Angular breadth of color use, normalized 0–1. Career median across 341 paintings: 0.695. Six near-monochrome outliers (fog, snow) sit at or below 0.35.
`CBI`	Color autonomy from luminance. 22.6% of all paintings reach ≥ 0.20. Series II has the highest individual rate (36%); Early the lowest (3%). Late Giverny 25%.
`high_key`	Fraction of pixels with OKLab L ≥ 0.72: tints and near-whites, absolute threshold. Rises Early→Impressionist (0.326→0.388), then declines each period without reversal, bottoming in Late Giverny (0.216).

What to Look For

Peak breadth and peak autonomy in different decades. Series I (amber) clusters highest on the left panel. Series II (orange) clusters highest on the right. Toggle each period alone and compare where it sits on each axis.

Series II: narrows and frees simultaneously. HE drops from 0.720 to 0.622 while CBI rises to 0.180. The top CBI paintings, Charing Cross Bridge (0.295), Rouen Cathedral Magic in Blue (0.293), Houses of Parliament Fog Effect (0.273), are all fog scenes. When luminance is suppressed, color must carry the edge.

Late Giverny: between the two peaks, not above them. HE 0.690 and CBI 0.167 both sit below their respective series-period peaks. What Late Giverny holds distinctly is the darkest tonal ground: high_key 0.216 vs every other period above 0.32.

high_key falls every period after Impressionist. The Impressionist decade was the most genuinely light-filled moment of his career. The surface of the Water Lilies reads as luminous; the pixel measurement does not agree.

What the Measurements Show

Palette breadth and color autonomy peaked in different decades and are structurally unrelated (r = −0.016 across 341 paintings). The common reading of Monet's career as a single intensification arc is not what the data shows. There are two arcs, on different schedules, with different causes.

Hue entropy peaks in Series I (0.720). The coastal campaigns of the 1880s (Étretat, Belle-Île, the Mediterranean) demanded greens, cold blues, and warm ochres simultaneously. Ocean and limestone pulled the palette wider than Seine riverbanks ever required. The breadth was subject-driven, not intentional expansion.

CBI peaks in Series II (0.180). When Monet suppressed light, color was forced to carry the edge. The most structurally autonomous color of his career came from his flattest light: Charing Cross Bridge in fog (0.295), Rouen Cathedral in blue mist (0.293), the Houses of Parliament through Thames haze (0.273). The series repetition froze the composition. The weather did the structural work.

Late Giverny sits between the two peaks, not above them. HE 0.690, CBI 0.167, both below their respective series-period highs. What Late Giverny holds distinctly is the darkest tonal ground of the career: high_key fraction at 0.216, against the Impressionist peak of 0.388. Tonal brightness declines after Impressionism and does not recover. The Water Lilies pond dominates the pixel ground. The light the eye perceives is an inference, not a measurement.

Mean chroma rises slowly across the career: 0.038 early, 0.043 at Series I, 0.044 in Late Giverny. He did not become dramatically more saturated. What changed was breadth, then autonomy, then darkness, in that order, in different decades, for different reasons.

Dataset & limitations. 341 paintings deduplicated from WikiArt CDN (324) and Wikimedia Commons gap-fill (17). Late Giverny is the thinnest period: 28 paintings across 1906–1925, with the 1909–1913 years and much of the Water Lily series underrepresented; major holdings are in museum collections that restrict digital reproduction. The Late Giverny HE figure (0.690) is Venice-weighted: 9 of 28 paintings are Venice 1908 (mean HE 0.762); without them the Water Lilies cluster averages HE ≈ 0.656. Direction of findings is consistent with what is known about Monet's late work; exact magnitudes should be treated as provisional. Cataracts diagnosed 1912, operated 1923. The measurement records what the paintings show. It does not say why. CBI values are measured at thumbnail resolution (200px long edge) and carry a slight upward bias, diminishing toward zero for high-CBI works. The fog paintings cited are minimally affected. The ≥ 0.20 individual threshold was calibrated against full-resolution images; at thumbnail resolution the equivalent threshold is approximately 0.23, meaning period CBI rate percentages are slightly overstated in absolute terms. Relative ordering across periods is not affected.

Light to Field