The workshop notebook
About thirty painters in this project each painted an original picture in Friedrich's manner, from scratch and alone. Each kept notes. Alice, the viewer, reacted to the pictures. None of it reached the next painter. This notebook gathers what they tried, what happened and why, by what a painter works on. It has no recipes and no verdicts on which picture came out ahead.
How sure. [firm]: several painters, or painters and Alice, saw the same thing. [fair]: one clear case or a cause shown by changing one thing. [thin]: one remark, or my reading.
Sources. [AR] main:notes/round6/alice_review.md (her descriptions only). [CT] main:notes/craft_trees.md. [SB] main:notes/sketchbook.md. [coast], [mountains], [winter]: origin/amnesia-coast:notes/fresh2_coast.md and its mountains and winter siblings (round 2). [e3-free] … [e4-near]: main:notes/amnesia3/easel3_.md and amnesia4/easel4_.md (rounds 3–4). A branch name (r7-arm1 … r13-tree3) means that branch's notes file. Other main files by path under notes/: [lab/sky], [lab2/rock], [l3_green] (loops/), [lake] (paint1/evening_lake.md), [texture] (round7/), [halo] (loop1_halo.md), [grain] (loop2_grain.md), [dry_rims], [twigfix] (fixes/*/README.md), [winter_port] (round7/), [fresh] (fresh_painters.md, round 1) and [glitch], [edges], [time], [drying], [varnish], [pencil_check], [round6], [workflow], [amnesia3]. [s:X] is worktree X's session log.
1. The ground and underdrawing
- The warm ground shows wherever paint is thin. In every round the red-brown ground came through thin sky, sea and snow as orange flecks, slivers or torn scallops, mostly at the dry ends of long strokes and in blended passes [mountains; coast; r9-arm1; r9-arm3]. Alice: "those orangey, earthy, browny lines, look horrible" [AR]. What held: a thin dead color first in the passage's own tones, a second pass of a different stroke length and judging coverage at full size [mountains; lake; SB]. A flat one-color underpainting showed through a sea as "bristle-textured blots" [coast]. [firm]
- The ground's warmth, used on purpose, helped. "The red ground pricks through the pale blue as a fine warm speckle" [r11-study1]; glints at dusk "read as dusk warmth," the same flecks elsewhere "as dirt" [s:r7-arm1; s:r9-arm1]. A warm ground also pulls pale cool colors warm: a lilac-gray came out khaki and near-white snow went pinkish, and two winter painters lightened the ground for pale pictures [r10-arm2; r11-fable; r11-study2; r11-astra]. [firm]
- The pencil can't be seen. A light search line was invisible at working size on the warm ground; only a firm dark restatement read [e4-green; r8-arm1; r13-tree1; SB]. "That stage contributes nothing you can see" [r10-arm1]. Painters placed things from written positions or overlays, "which are not the drawing" [e4-green]. It was slow too: "the single most expensive stage after the sky, for something a painter does in minutes" [r13-tree2]. [firm]
- Drawing followed the painting. "A painter draws it; I compiled it" [e3-near]. "Anything singular (a stone, a man, a crescent) I could draw"; numerous things were the hard part [e4-free]. Redrawing a dolmen after the sky was down "felt like a real painter's second thought" [e4-free]. [firm]
- Outlines Alice took for pencil were paint. She asked about "remnants of the pencil?" around rocks [AR]; with the drawing removed the edges looked the same, and the lines came from contact glazes, snow caps stopping short and the varnish [pencil_check]. The ground's own texture could print through as well: horizontal "wood-grain" from a ground laid in level strokes (a tool fault, since repaired) and "a crackle net" where a soft brush touched only its ridges [grain; r13-tree2]. [fair]
2. Skies
- The sky everyone trusted. From round 2 on nearly every painter laid the sky thin in long near-level arcs, fused it, stippled into the wet and once dry added a finer stipple barely lighter than the field, denser toward the glow [coast; winter; mountains; e3-free; r11-study1..3; r12-tree1..3]. Round 2's three each called it their strongest passage: "a luminous grain with no stroke direction" [winter]. Round 1's skies had been ruled strokes, "the same sky three times" [fresh]. [firm]
- To Alice the texture read as compression. "It almost looks like JPEG artifacts" [AR]. Lossless and compressed crops looked the same: "the crunchy mottle is in the paint (the stipple over the weave)" [AR], mostly the stipple's "blotches with sharp scalloped rims" [texture]. She wanted more texture, not less: "we're just not doing the texture right yet" [AR]. Stipple itself is not the fault: Friedrich stippled skies, mist and distant hills (notes/research/friedrich_materials.md, [NG p.56]) and round 2's skies were stippled; how to make it read as paint is still open. A wet sky of unequal thin bands, no stipple, read to her as "less digital and more bad painter" [lab/sky; AR]. [firm]
- The fault that stayed: a correct, quiet gradient. "Close to a digital gradient at viewing distance," the stipple "only visible up close" [r12-tree2; r13-tree3; SB]. A sky whose light was written as a smooth color formula looked to Alice "maybe a little too perfect," a "computational light simulation" [AR]. Why it stayed: the sky was the slowest thing to repaint, so "the sky got the fewest iterations" [r13-tree3]. [firm]
- Stipple has to sit close to what's under it. Much lighter than its field it "reads as salt" [coast]; stippled band by band from fixed mixes it made "hard stripes," fixed by touches taking their color from the paint beneath [r7-arm1; r7-arm3]. Blending a fresh broad pass lifted it into "cloudlets" or rust streaks down to the ground [r8-arm3; r8-arm2]; long strokes that ran dry left holes: "coverage … doesn't mean covered" [r9-arm3]. [firm]
- Clouds sit only if laid into the wet sky. Banks painted inside outlines on a dry gradient came out as "two cigar-shaped cloud cutouts pasted on it"; laid unfenced into the wet sky they "sit in it" [lab/sky]. Wet-stippled clouds vanished under the blender [r10-arm3]. First shapes read as "white 'worms'," cigars, "cartoon sprites" [r10-arm3]. What held: thin level streaks, warm only underneath, edges lost [coast; e4-free]. Alice on one sky: every cloud was "drawing the exact same moon" [AR]. [firm]
- The glow stays the lightest thing. Three trunk studies first made the snow lighter than the sunset, "a digital gradient, not evening," and all three darkened the snow a step below the sky [r11-study1..3]. [firm]
3. Distance and air
- Distance comes from value steps. Each nearer range a step darker and cooler, mist at each foot: "it came from values, not from detail" [mountains; SB]. [fair]
- Carry the sky back over a crest. Stroke ends notched hard crests into "a mesa"; sky taken from just above and laid back over it "turned the crenellated mesa into a soft dome": "the move a painter makes, and it worked" [e3-green]. A stenciled ridge was "the filled selection" [edges]; Alice: "Still the boundaries" [AR]. A dark brought into a wet sky "broke into a lace of pale beads," while over the dried sky it gave "a clean found ridge" [lake]. [firm]
- Never stipple a pale veil over dark land. "Over dark ones every dot stands alone" [mountains]; "pale salt on the darker strip" [r11-fable]; stippled mist read as "granite" [mountains; winter]. A thin veil brushed level and fused worked [mountains; r11-fable]. [firm]
- The far band of woods was weak nearly everywhere. "A flat dark hump" [r10-arm1], "a hedge wall" [r12-tree1], "more diagram than distance" [r12-tree2]; an evenly stepped skyline was "a perfect sawtooth," seen only at full size [r10-arm2]. What helped: paler, lower, broken into stretches with open country between, in unequal groups, the tops re-cut with a small brush into "a believable serration" [r13-tree1; r11-astra; r9-arm3]. Groves and field trees became "identical balls with identical shadows, scattered like stamps" until set in mixed groups [e4-green; e3-green; SB]. [firm]
- A near-black wood won't recede by glazing. Pale columns read as "white ghosts or smoke"; a veil into open dark "vanished." Body-paint haze after the dark dried, then separate firs in rows paler toward the air, finally showed "air and sky between them" [l2_near; l4_near; l5_near]. [fair]
- Far buildings. "The spire carries it" [r10-arm2]. Many small strokes into an open band made gables "stacked like pagodas"; one firm shape per roof on a dried band read [r9-arm1]. [fair]
- Guard the horizon. Where snow and glow matched, a wood set on the planned horizon "floated in snow" until the painter probed for the real meeting [r9-arm3]; snow spilling above it "silently ruins a horizon, the most important line in a Friedrich" [r12-tree3]. [fair]
4. Snow, ground and grass
- The foreground was every round's weak spot, by the painters' own account. "Friedrich's foregrounds are dark but drawn" [coast]; "his foregrounds reward close looking everywhere; mine do in patches" [e3-green]; the other five easel painters said the same [e3-free; e3-near; e4-free; e4-green; e4-near]. All six painters of rounds 12–13 called the snowfield empty, named what was missing ("a track, a stone, a stake") and none added it [r12-tree1..3; r13-tree1..3]. One who added a fence last: "it took ten minutes, which says where the next hour should go … not more passes on the sky" [r11-fable]. [firm]
- Snow recedes only when laid out on the ground. Drifts placed as blotches read "like a lake mirroring clouds"; bands narrowing with distance receded [r10-arm2; r11-study3]. The stroke size still never changed with depth [r11-study1; r11-study3]. [firm]
- Shade laid on afterward looks pasted on. Over dry snow shadows came out as "crisp blue outlined scribbles ('wire')" [r9-arm3], glazed as "a separate pasted-on path" [r11-study3], stippled as "a strip of dark clover" [r10-arm3]. What held was mixing the shade into the snow as it was laid, "the way a painter mixes the shaded grass, not shades the grass afterwards" [r10-arm3; r11-study1]. Stippled cool hollows became "blue puddles" and "purple slabs": "the contrast of the touch against what's under it decides" [r12-tree2; r12-tree3]. [firm]
- New snow never matches old snow. A drift over a trunk's foot went "a pale oval, then … a blue puddle, then a pale rectangle"; shadows mixed from the planned snow color came out lighter than the real snow, leaving "pale halos round the stones" [r9-arm1; winter]. What held: paint taken from the field right beside the spot and drifts laid as rows of strokes along their surface, "how a painter would do it," where a filled shape was "a hard white box" [r9-arm1; r13-tree2; r13-tree3]. [firm]
- Snow lying on things beads. "Stipple works as a field texture; as a thing lying on a thing it beads" [r11-fable]; round white touches "all looked like wool or fog" [r9-arm2]. Crests cut against dark bark scalloped into "a cotton edge"; long flat overlapping strokes, broken in places, held [r11-study1..3]. A bank against a stone read as one bank only when deep to windward, nearly gone in the lee and colored from the field [l1_near; l2_near]. [firm]
- Thick snow reads as illustration. Relief light made body snow look like "crumpled paper" [r9-arm1]; "more 'clean illustration' than thin oil" [r10-arm2]. Tracks without foreshortening were "a flat pale ribbon (a second river)" [r9-arm1] and later ones "too evenly spaced" [r11-fable]. [firm]
- Grass went on last, and placement was the whole problem. Everyone flicked dry grass upward over finished ground, as Friedrich did. It failed by evenness: "a uniform barcode of short upright dashes," "confetti," "a digital sprinkle," "statistical, not looked at" [e4-near; r8-arm3; r12-tree2; r10-arm3]. What held: patches where wind thins the snow, clumps larger near, long curving blades each leaning differently: "the barcode became a meadow" [r12-tree2; r8-arm3; e4-green]. Alice: "still very digital"; "the foreground grass just stops" [AR]. [firm]
- Small plants vanish or turn into diagrams. A thin thistle over grass of its own value is invisible in the preview; transparent greens vanish on green; built part by part a plant is "a diagram of one" [e3-green; e4-green]. [firm]
5. Water and ice
- Darks first, lights brought to them wet. Laid as one dark with the mirrored sky brought to it, a headland and its reflection "read as one dark shape" [lake]. In the water lab the layered version "reads as a render. Every edge is found," its reflection "a knife-sharp outline"; laid wet and together, "the reflection dissolves into the water" [lab/water]. Alice: the layered bank "abnormal," the wet one "more bad painter," still "not terribly water-like" [AR]. [firm]
- Calm water wants whole level strokes. Drifting strokes made "cloud-like blotches," a flat brush "brick marks" [r7-arm3; e3-free]. Flipped fir silhouettes put spires in the lake instead of "the flat polygon of the headland" [lake]; reflections in rows of broken level pulls held, and a level blend over them "left a ghostly rectangle" [r7-arm2]. One soft light line couldn't be made: blended it became two lines "with a darker trough between," unblended "a single uniform hard wire" [lab/water]. [firm]
- Frozen water reads as a road. A brook "is still somewhat 'road'" [winter]; a river stayed "one even band with a perfectly continuous edge" until snow tongues dragged from both banks, sheen and cracks made it ice [r9-arm2]. A pond was "a little too neat an ellipse" [r10-arm2]; shore touches "read as road markings" [r11-astra]. [firm]
6. Trees (shortened from [CT], which has the full record)
- What the record says of Friedrich. He drew particular trees in pencil in "small strokes and hooks," reused studies for decades and painted trees over the finished sky; no text says how he painted bare twigs [CT]. Alice, at a real Friedrich winter tree: every branch carries "the same amount of remnant snow," and "so much texture," probably many small strokes [AR]. [fair]
- The small wood is where trees failed. Thousands of even twigs read as a "spiderweb," sprays "in the air with no visible parent," "brooms," "a fairly uniform haze" [CT]. Alice: "branches in the middle of nowhere"; yes, "the brooms"; "twigs hang in the air" [AR]. Twigs laid as a tone read as "fur or cotton wool" [lab/baretree]. Twigs forking "at the same fractions and similar angles" [r11-study3] drew "trees only do that when they're in extreme distress" [AR]. [firm]
- What held: few limbs, each one movement. Round 2's oak was sparse and broken, each limb pulled from where it leaves its parent, a smaller brush set into the wet end of the last [winter; CT]. Straight runs turning at knots read as oak; smooth curves as "smooth noodles," "rubbery" [r8-arm2; winter; r11-study2]. Where the brush was changed or ran dry, joins showed as light bands and dotted chains: "a painter would reload" [r10-arm2; r8-arm2]. Twigs hanging in the air were partly a brush fault, since repaired [twigfix]. [firm]
- Crowns in leaf. Over 40,000 even touches read as "confetti," "textured more than … a lit volume"; a few big masses read as "broccoli" [lab/foliage; lab2/lime]. A lime laid dark first, then hooked leaf touches back to front colored by their light, with a cool veil away from the sun, drew from Alice "the big trees and bushes finally look nice" [r10-arm3; AR]; a lay-in cut hard to the crown gave "a hard, broccoli-like silhouette" until the leaf touches made the edge [r10-arm3]. [firm]
- Firs need a dark mass under the needles. Hatching alone read as "fishbones," "palm fronds," "comb teeth" [r8-arm2; r10-arm2; r11-astra]; evenly spaced firs were "a row of Christmas trees" [SB]. [firm]
- Pollards: painters and Alice disagreed. Five painters drew a fat bole with thin upright rods from one head, and one wrote that the rods "worked well" [r9-arm2]. Alice: "like the brush you use for shaving"; "I don't think those trees even exist" [AR]. An old pollard has a gnarled head and thick branches of several ages [CT]. [firm]
- Snow on limbs, trunks and feet. Separate white touches beaded into "white pearls" and "polka dots"; a continuous ridge on the real upper side read as snow on wood [CT; r11-fable; r11-study3]. Backlit trunks came out "one mid-dark brown all the way up"; Alice on one: "the tree's shading simplistic" [r11-study2; AR]. A round stroke end at the foot left a bulb or "boot"; snow laid over the foot last, with an uneven edge, planted the tree [CT; r12-tree1]. [firm]
- Everyone painted the same oak. From the same research all six late painters painted a centered stag-headed dome; Alice saw one as "too symmetrical" [CT; AR]. Their critiques asked for accidents: "one long limb reaching out and one gap where a limb fell" [r13-tree2]. [firm]
7. Rocks and buildings
- Rocks come out as other things. "Macarons, loaves, haystacks, beetles, soap bars" in round 1 [fresh]; later "a mushroom capstone," "a hay bale," "pillows," "a stack of bricks," "pies on white plates" [e3-free; e3-near; e4-green; e4-near; r8-arm1]. One painter called a dolmen "the focal point and the weakest-drawn thing in the picture" [r10-arm1]. [firm]
- What made stone. "A high shoulder, a broken back and a crack running off at an angle made it a stone" [SB]. A pure oval "reads as an egg or potato" [coast]. Angular planes with only the top taking the sky light, strokes following the planes, only the shadow side blended [e4-free; e3-green; SB]. Roughening too fine gave "pumice," black specks "pepper" [coast; SB]. [firm]
- A rock sits when shadow side, foot and cast shadow are one dark. Finished inside its outline with a glazed cast shadow, a rock "looks pasted on"; with the three laid as one dark across the rock-ground line, "B sits" [lab/rock]. Alice still saw the shadows as wrong ("super wrong," "a different bad shadow"), and of a rock built from three lit planes said the close crop "looks actually pretty good," the shadow "probably still very wrong" [AR; lab2/rock]. [firm]
- Backlit stone goes pale. A tor's tops came out "pale beige: loaves"; "contre-jour painting is all rim" [mountains]. Snow caps read as "stickers," "a fried egg," "little cakes (white top, dark flat underside)"; snow put wherever stone faced up made "tiger stripes" and even landed inside a covered chamber [s:r11-flash; winter]. [firm]
- Cracks and weathering. Drawn into open paint cracks came out "dashed, like stitching"; left as unpainted channels, "an even, ruled line"; one pointed stroke with a light lip, lost halfway, held, and "the eye goes to it" [e3-near; SB; lab2/rock]. Streaks from ledges and clustered lichen turned mottle "into something that reads as weather" [e3-near]. The "dots on the lit part that look like a digital anomaly" were the varnish pooling at stroke edges, a tool fault since repaired [varnish; AR]. [firm]
- Buildings. A ruin's weathering went "rusty leopard spots" until stone-only colors were used [winter]; a dolmen chamber read as "a hut with two windows" until the glow showed only in slivers [e4-free]; rock on a straight base went architectural [e3-near]. [fair]
8. Figures and animals
- The small figure from behind is in nearly every picture, usually the only hard silhouette, rimmed toward the hidden sun [coast; winter; mountains; lake; r9-arm2; r10-arm1]. First versions were toys: "a bell-shaped sack in a cowboy hat," "a pawn," a hat "a mop with a bow tie" [r10-arm1; r11-fable; s:r9-arm3]. What read: a coat narrowing at the shoulders and flaring at the hem, one leg in stride, a stick, a hat painted as part of the body, feet sunk in snow [r9-arm2; r10-arm1; lake]. [firm]
- Figures show faults only at full size, and later passes paint over them. A sea veil left a man's "coat … pale gray and his cap … gone," and the preview hid it [e3-free]. On flat ground a standing figure's head sits on the horizon, so painters sized figures by eye [e3-free; r11-fable]; at the right scale they were "too small to carry the sentiment" [r10-arm2]. Alice: "The figure disappears into the snow; that should not happen"; "a rendering glitch around the figure" [AR], partly a tool fault near thick paint and partly a veil cut around the figure with a margin [halo]. [firm]
- Crows: a real silhouette or nothing. Perched crows "read as pods and the fliers as specks" until given an upright body, beak, tail and bent wings [s:r9-arm1; r8-arm1]; dragged arcs made "hearts" [winter]. Against a dark trunk they vanished; against sky they were "blobs with beaks" at full size [r12-tree1]. Crows placed by eye floated when the tree was reworked [r9-arm3]. [firm]
9. Small details
- The first try at any small detail repeats. "The first try of nearly every small detail … came out as a regular, repeated shape: dots, arrowheads, confetti, scallops, lumps. The fix was always the same: fewer, irregular spacing, varied size, patchy" [r13-tree1; e3-near; r12-tree2]. The irregularity did the work; fewer was one painter's way to it, and texture itself was never the fault (§15). [firm]
- Moons. A dragged arc gave "a blunt banana," a filled shape "a small moth"; a crescent filled with short touches so the horns point, or one pointed stroke swelling along the lit limb, turned toward the sunk sun, read [mountains; winter; r8-arm1; s:r8-arm1]. A glazed glow drew a hard line across the sky [winter]. Alice: "the moon has a weird halo"; "the moon does not look like that ever" [AR]. [firm]
- The brush won't come to a point by itself. Tips came out as "a rounded peg or a bean"; "a real sable round comes to a point when lifted" [mountains; winter; r10-arm1]. [firm]
- Particular things. A fence with a sagging rail was "the most Friedrich-like particular thing in the picture" [r11-fable]; Alice on another: "not right, very digital" [AR]. A net read as "a flag" [coast], a boat as "a puck" [r9-arm2], a cross "like a sword" [r9-arm3], reed beds "as domes" [r7-arm3]. Tiny things carry at full size and vanish in the preview: yarrow read as "scraps of white paper" [mountains; r10-arm3]. [fair]
10. Edges and how things sit in their surroundings
- Alone on dry paint a thing looks pasted on; laid wet against its neighbors it sits. Every lab found it: cloud cutouts, the knife-edged reflection, the pasted rock [lab/sky; lab/water; lab/rock]. The old habit of drying everything before each passage made cut-outs [time; round6]. Timing decides the edge: a hill painted into open sky dragged into "a pale, broken band," after lunch less and the next day sat on top crisp, "a cut-out" [time]. [firm]
- A stencil makes an edge no brush makes. "No brush lays that; a selection filled with paint does" [edges]. Edges decided by the light held: found against the glow, soft along most of a contour, lost into a neighbor of close value; but soft over dry paint stays "a thinning, broken fringe" and a truly soft edge needs wet paint [edges; SB]. A soft mask gave "gaps, not thin paint" [r11-study2]. [firm]
- Nothing sits behind anything unless arranged. "There is no 'behind' for gestures, only order and masks" [r13-tree2]; "every new pass needs every earlier motif subtracted" [e3-free]. A land glaze "drew a visible horizon line straight across the dark willow trunks"; broad strokes filled any hole smaller than about two brush widths [r9-arm1]. [firm]
- Halos come from margins. A veil cut around motifs with a slightly larger outline left a ring that reads as a glow; cutting exactly, or blurring the veil and then cutting, removed it, and touches just outside in the surroundings' color closed the rest [l3_green; halo; SB]. [firm]
- Anything that rings a whole shape becomes an outline. A ring shadow "reads as a cartoon outline"; light on every side of a stone was "like frosting"; a lit rim on both sides of a trunk, "a pale outline" [e3-near; e4-free; lab/baretree]. Light only on faces turned to the light held [e4-free]. A drawn contact line "read as a seam," since "a stroke's centerline isn't its edge" [r11-study1]. [firm]
- Carry past the line and cut back. "A painter would carry the paint past the line and cut back with the next passage" [winter]. Laying the neighbor's color back over an edge made it "lost and found" [r7-arm2; r9-arm2]. Thin dark edges over pale sky dried with "a thin light-brown rim … the most 'digital' thing in the picture" [r9-arm2; r12-tree3]. [firm]
- Feet are where things look pasted on. A cross's foot under snow of exactly the right color still left "a straight horizontal cut … (correct but reads as pasted)" [r8-arm1]. Grass laid over roots and stone feet was the most common cure [e3-free; e3-green; e4-free]. [firm]
11. The paint surface and texture
- Wet paint behaves by the hour. A light into an open dark plowed "a camouflage mottle," into a setting dark it was "soft and leafy," into a tacky one "flat poster slabs"; a dark over dry sky was "a clean dark" [lab/foliage; lab2/oakleaf]. Pale paint set fast and blacks slowly, so a dark passage could still be open beside a tacky sky [drying]. Hidden wet paint was the commonest surprise: a rock "went semi-transparent," a river plowed, "the drying state is invisible until it bites" [e3-near; e4-near], much of it later traced to an older model of wet paint [round6]. [firm]
- Body paint over dry paint kept its outlines, "a lacy net of loop outlines," shadows "like glass tubing": a tool fault, since repaired [r8-arm3; r9-arm2; dry_rims]. Painters at the time went over dry paint only with stipple or glaze [r8-arm3]. [firm]
- Glazes. A glaze waits for everything under it to dry, so "a shadow can't be modeled wet-into-wet with the thing that casts it" [r8-arm1; r9-arm3]. Glazes fading toward nothing drew hard lines and rectangles [winter; coast]. Even glazes read as gradients: "a painter would glaze with a brush, unevenly" [r12-tree3]. The final veil darker toward the edges, after Friedrich's advice, "gathered the picture onto the glow" [lake]; heavy, it was "a photographic vignette with a visible oval" [e4-free]. [firm]
- Thin paint shows its other color. Dark twigs came out "golden-brown and transparent," light touches over cool darks dried orange, thin edges showed "red rims" [r9-arm1; coast]. Mixing each mark for how it looks over what is under it and setting out only a passage's few colors fixed it [r9-arm1; r11-astra; r10-arm2]. [firm]
- Stipple is a surface, not a scatter; sparse it read as "pollen" or "frost" [e3-near; e3-green]. Big flats and fluid washes left paint only at the rims, "fish scales" [r7-arm3]. Strong raking light drew "worms" and "swirling impasto," and painters kept it low [e3-free; e4-free]. [firm]
- Cracks. The early crack pattern read as "crazed glass" and "a dark rectangular grid," and all three round 2 painters made finer ones [mountains; winter; coast]. Later cracks read over pale snow "as ruled hatching" [r8-arm2]. Alice: "we do want the age stuff," but "still too neat, still too digital"; in real Friedrichs she saw cracks that "almost flow, running down a little to the right" (Monk by the Sea), clusters random "in a sort of human random way" and an amount that varies by painting, where "ours is one uniform net" [AR]. The crack widths have since been repaired (the first cracks now open widest); direction and clusters are still open. [firm]
12. Composition
- Left alone, painters paint the same picture. Free painters chose "twilight, a crescent moon, the evening star, a figure seen from behind and a dead oak" again and again [amnesia3; e4-free]. These are Friedrich's own motifs; what Alice noticed was the sameness from painter to painter, not the motifs. Alice saw that a painter "without any extra instructions around themes, converges on roughly the same painting" (pond, sky, trees reflecting, foreground grass) [AR]. [firm]
- The shared design: low horizon, big quiet sky, dark near stage, empty middle, far band, light low and behind: "everything important is dark against the lightest band of sky" [r10-arm1; mountains; coast]. "Contre-jour needs few decisions" [r8-arm1]. [firm]
- Safe, centered, even. "Too centered and too symmetric" [r12-tree3]; "a Friedrich arrangement, but without his tension" [r13-tree1]; "closer to a pleasant summer view than to Friedrich's charged emptiness" [r10-arm3]. Even spacing crept into crests, skylines and uprights [r10-arm1; r10-arm2]. Alice: "an emptiness that doesn't do anything for her"; "it doesn't come together" [AR]. [firm]
- Squint first. The value view showed "the rock, sky and middle floor all at the same value," and darkening the surroundings "changed the whole picture's weight in half a second" [e3-near]. Framing makes its own faults: keeping a crown inside the frame bent its edge shoots into brooms, "not from a tree" [r13-tree2]. [firm]
13. Working method, time and looking
- The preview and the full-size picture are different paintings. The full-size crops "caught most of the above," while in the preview faults "read only as vague wrongness" [r11-study3]; "a fix judged at one width can be wrong at the other" [r9-arm1]. The preview is for composition and value [SB]. Alice later asked to "paint, look and deliver at one size" [AR]. [firm]
- Saved stages made detail work possible and trained a bad habit. Resuming a full-size crop from a late saved stage took seconds instead of minutes, "what made iteration possible" [mountains; r13-tree1]. But any edit marked every stage stale, so painters overrode the check by reflex, and it bit: "a stale boat," "the old shadow" [r9-arm2; r11-study2; r12-tree1]. Crops are close but not exact, since long strokes carry paint from outside the window [workflow]. [firm]
- What is slow is the sky and the ground, not the tree. A full render took a couple of minutes, mostly priming, pencil and sky; a tree of thousands of strokes took seconds [r12-tree1; r13-tree3; r11-fable]. So the sky was painted once, and sky, horizon and snowfield got the fewest looks [r13-tree3]. [firm]
- The painter's clock. With hand time counted, one fir wood came to about 53 hours against half an hour for the sky [time]; the lake painter finished a passage "rather than resting in the middle of the wood" [lake]. A view coloring open, setting, tacky and dry paint showed what was still wet [time]. [fair]
- Finding a cause meant stopping stage by stage. One rectangle took five stopped renders [coast]; a crackle net took four crops with layers switched off: "a painter would have seen it at once on the easel" [r13-tree2]. [firm]
- Look after every step; undo freely. "Undo is the palette knife" [e4-free]; a third of passages undone is "normal" [SB]. "A program written in one go would have had the mushroom capstone and the lawn-green foreground, and I wouldn't have known" [e3-free]. [firm]
- Measure when the eye doubts, and expect the tools to move. A "pale" rock was simply lighter than its turf [SB]; sizes given as half-widths doubled boulders [e3-green]. A round 2 picture repainted on a later engine came out with thinner marks, which Alice saw in its trees [winter_port; AR]. [fair]
14. What painters did that a person would not, and the reverse
Painters here, unlike a person:
- Placed marks by positions read off a small image, "programming with a picture as the debugger" [e4-green; e3-near]. [firm]
- Finished each thing alone over dry paint and laid tens of thousands of like touches; their notes say what a person would do: "merge some, lose some and let one or two big ones dominate" [lab/foliage; time]. [firm]
- Answered a fault with more of the same, stenciled every edge and subtracted every figure from every glaze by hand [CT; edges; r9-arm1]. [firm]
- Mixed from the plan instead of the canvas, so shadows came out lighter than the snow under them [winter; coast]. [firm]
- Placed snow by argument, and got it on limb undersides and inside a covered chamber [r13-tree1; r11-study3; s:r11-flash]. [firm]
- Saw a mark only after a render, and gave the slowest passage the least attention [r11-study2; r13-tree3]. [firm]
A person does, and painters mostly did not:
- Reloads when the brush runs dry; paints snow over a tree's foot and cuts back without thinking [r8-arm2; r11-study2]. [firm]
- Brushes a shadow wet with the thing that casts it and keeps the blender inside the passage [r8-arm1; mountains]. [firm]
- Paints one particular tree or stone and adds the particular thing: all six late painters named the missing track or stone and none painted it [r13-tree2; CT]. [firm]
When painters acted like painters, it paid: cutting sky back over a crest, darkening the surroundings, waiting for paint to set. "Paint has consequences here, and time is a real tool" [e3-green]. [firm]
15. What comes back in every section
- Evenness reads as digital: grass, twigs, snow beads, stipple bands, skylines, tracks and cracks all failed first by being even. What helped was unevenness (clusters and gaps, varied size and spacing), not fewness: Alice saw "so much texture," probably many small strokes, in a real Friedrich tree, and found an empty picture's emptiness did nothing for her [r13-tree1; SB; AR].
- Pasted on or sitting depends on when a thing goes down (wet against its neighbors or alone on dry paint) and where its color comes from (the paint beside it or the plan) [lab/*; time; r9-arm1; winter].
- Two sizes, two pictures: nearly every fault was found late in a full-size crop [SB; r11-study3].
- The painters saw their faults before anyone else, but their fixes stayed inside the same approach, and the particular things they asked for went unpainted [r12-; r13-; CT].
- Still open: a sky with structure that reads as paint; foliage whose shade isn't one slab; cracks with direction and clusters; a twig mass that reads as tone without turning to fur [SB; round6; AR; CT].