What the painters were given

Stream: green (sourced greens, foliage, meadows)

notes/green.md · 207 lines · Notes

Given to 1 painter:

Stream: green (sourced greens, foliage, meadows)

The user noticed that last night's paintings all share one palette and one mood, while Friedrich "made many paintings that were very green." The palette had no green tube, and growth grew only bare skeletons. This stream adds the greens he is documented to have used, leaves on grown trees and a ground-cover primitive for meadows, plus a study that uses them.

1. Research (notes/research/friedrich_materials.md §9)

The key new source is the Dresden technical study of 14 paintings, Mäder et al., METALLA Sonderheft 13 (2025), p.102 [MÄD]. I checked the quotes against the text myself:

Dates come from the NGA Artists' Pigments volumes and Field 1835. Viridian, chromium oxide and emerald green are too late or too doubtful for him. No green analyses were found for Hill and Ploughed Field, Meadows near Greifswald, Morning in the Riesengebirge or Chalk Cliffs; that is a gap, stated in §9.

2. Palette (crates/paint/src/palette.rs)

New tubes, with masstone, hiding, stiffness and strength as documented approximations (sources in the doc comments):

tube masstone hiding stiff strength basis
Prussian blue #172440 0.35 0.45 3.0 transparent, very strong [AP3 pp.196–197]; strength tempered (A)
green earth #3a4843 0.2 0.35 0.3 translucent, weak, little body [AP1 p.146; FIELD p.129]; Munsell 7.5G/2.9/1.5 converted
Rinmann's green #5f8f76 0.35 0.5 0.4 semi-transparent, weak (A for the numbers)
copper green #3f7f6a 0.25 0.4 1.0 "poor hiding power in oil" [AP2 p.132]; numbers A

API:

Palette::friedrich_early_greens()   // early + Prussian blue + green earth
Palette::friedrich_1820_greens()    // 1820 + Prussian blue + green earth + Rinmann's green
Palette::friedrich_1820().with(vec![Palette::copper_green()])  // a rare tube for one passage
Palette::green_tubes()              // the two green-making tubes, to add anywhere
let st = Style { palette: Palette::friedrich_1820_greens(), ..Style::friedrich() };
let sky_pal = st.palette.only(&["lead white", "cobalt blue", "pale smalt", "yellow ochre", "red earth"]);
st.broad().mixed(&sky_pal, 0.45)    // paint the sky from its own family

The base palettes are unchanged, on purpose. I first added the greens to friedrich_early/friedrich_1820. With them, the aimed search picked the very strong Prussian blue for the smalt/cobalt sky ramp. The miss halved (0.049 → 0.023 OKLab), but recipe switches made seams at thin coats, so palette::tests::smooth_targets_make_no_seams failed (0.028 against a 0.018 limit). It would also have changed every existing sky, and his skies are smalt or cobalt [ALF; NPJ25]. So the greens live in separate palettes, and existing paintings and the golden are unchanged. The golden passed without being re-recorded.

Finding: for muted summer greens, the base palette already mixes well (cobalt blue + chrome yellow + bone black, ΔE ≈ 0.003–0.006). The greens palette adds green earth to those piles (for example "bone black 0.08 + chrome yellow 0.17 + green earth 0.75"). Prussian blue seldom wins for these targets (test green_tests::greens_are_mixed_closer). The aim code is untouched (it belongs to fixes-paint).

3. Geometry (crates/paint/src/growth.rs)

let oak = Habit::oak().grow((690.0, 570.0), 360.0, seed);
let leaves = oak.foliage((-0.55, -0.75, 0.35), seed);   // toward the sun
let mask = leaves.mask(c.frame());
let light = leaves.lit(c.frame());
for cl in leaves.back_to_front() { /* touches at cl.at, sized by cl.r, colored by cl.lit */ }
let hedge = Habit { years: 15, ..Habit::alder() }.grow(..)
    .foliage_with(&Leafing { clump: 0.05, spray: 0.5, ..Leafing::alder() }, sun, seed);
let tufts = Sward { horizon, near: h, height: 26.0, wind: Wind { lean: 0.12, gust: 0.18, period: 140.0, seed }, ..Sward::default() }
    .grow(&meadow_mask, seed);

4. Study (paintings/src/bin/study_green.rs)

A summer meadow in daylight, 1.4:1: a leafy oak, a hedgerow of young alders and beeches, far fields, a clear sky. The study's own habits:

--geom shows the geometry flat (diagnostic only).

Evidence (in [local path redacted]):

Honest judgment

At 1000px it reads as a green summer landscape: meadow, hedgerow, a tree in full leaf, cloud shadows. It is a clear step away from last night's twilight palette. It does not yet read as a Friedrich summer passage.

Next

  1. Foliage painting:
    • Leaf-shaped marks at the silhouette (hooked, jagged strokes following tilt/droop) instead of clipping.
    • Layered greens (MÄD's "up to four layers"): a warm mid layer under the dark, with sky scumbled back into gaps.
    • Light touches varied in size and shape, oriented along the twig.
  2. Growth: a mature oak habit (more years with less self-thinning, and reiteration from dormant buds), so crowns are rounder and denser. Pollard willows are worth doing for Greifswald meadows.
  3. Performance: the hedge's 21k clumps cost ~40 s at 1000px. Paint far hedges with a coarser Leafing (bigger clumps, spray 0) or a stipple over mask.
  4. Easel bindings: Skeleton::foliage, Foliage::{mask, lit, gaps}, Sward::grow and the greens palettes.
Source