Crucible Year Peter Yuill
Patterns · Visual Mapping

Visual Mapping

123456789101112ABCDEFD10 · 59%

402 paintings, every one mapped: where the ink went fully black.

11%53%share of the zone fully black

Each painting's sheet is cut into the same 72 zones and measured; the map averages the chosen paintings zone by zone. All 402 paintings are mapped, including the ones not shown on this site.

The paintings in this set.

The evidence under the average: each painting's own fully black map, in the order they were made. 402 paintings; the empty slots are paintings not shown on this site.

T_119T_132T_128T_130T_131T_127T_150T_151T_148T_149T_154T_146T_157T_159T_164T_166T_172T_179T_192T_197T_184T_200T_183T_189T_196T_193T_198T_199T_194T_213T_208T_212T_210T_225T_220T_227T_224T_223T_228T_222T_219T_231T_234T_237T_229T_235T_236T_239T_240T_246T_249T_250T_255T_252T_251T_256T_254T_257T_259T_268T_267T_260T_271T_272T_269T_265T_270T_264T_262T_277T_278T_288T_283T_282T_294T_290T_291T_276T_280T_287T_292T_281T_293T_303T_304T_305T_309T_308T_302T_318T_316T_313T_317T_314T_312T_311T_315T_310T_326T_324T_322T_321T_327T_320T_329T_333T_330T_332T_337T_331T_328T_335T_338T_336T_339T_334T_350T_343T_345T_346T_341T_342T_344T_347T_348T_349T_340T_353T_357T_355T_351T_356T_358T_352T_360T_367T_363T_359T_361T_365T_366T_364T_369T_362T_372T_371T_374T_373T_370T_377T_375T_376T_383T_384T_379T_381T_390T_393T_392T_386T_396T_391T_388T_395T_387T_397T_398T_402T_400T_399T_401T_403
1

How the maps are made.

Each painting's photograph is cut into the same grid: twelve columns along the long edge, six rows along the short one. In each cell, Python measures how much of it is ink, how much is fully black, how much is pale wash, how much belongs to the painting's largest black mass, and whether the paper was left bare. The ink is found the same way the V23 engine finds it. A sheet taller than it is wide is turned first, so every map lies the same way. Most maps come from the engine's own analysis photographs. The 143 early works without one are measured from Peter's older phone photographs, with their uneven light evened out first; those maps are noisier. Every map was checked by eye on sample works before the whole set was run. Each tile of a map is one zone, averaged over the chosen paintings. The shading is stretched to the set itself, from its dimmest zones to its brightest, on a curve that keeps the middle dark so only real gatherings glow; the legend gives the two end values. So compare shapes between maps, and read the numbers, not the brightness. A set of fewer than five paintings keeps one shared scale and shows faint.

2

What the maps cannot say.

Where the ink sits does not predict the rating: retested in October 2026 with these maps, on 176 V23 works and 136 early works, and nothing held up on later works. The flaws are where the V23 engine placed what it found, in nine parts of the sheet, so they exist only for the works V23 has read. Ratings use one scale for the whole year, V23 where it exists and V23 lite for the early works, never averaged together. Every painting counts in the averages, including the ones not shown here; those appear only as unnamed dots and empty slots.