Commit graph

13 commits

Author SHA1 Message Date
CRT
177f06c9f7 Give the sign real ironwork, and hang the board from what it hangs on
The sign read as broken, and the reason was the same one the tankard's
handle had: `bendGeometry` applied to a `boxGeometry`. A box has two levels
in Y, and two levels cannot bend -- they shear. Both the brace and the
curl were sheared parallelograms rather than curves, and the sheared curl
is what pushed the bracket's top to 0.579 while the post it is bolted to
ends at 0.442. An arm standing above its own post is not ironwork, it is a
mistake, and it had been in every render of this model. Both are now real
sweeps from `arcBarGeometry`, with their ends placed inside the post and
inside the arm rather than corrected towards them afterwards.

Replacing the curl immediately exposed a second fault that it had been
hiding. The board's part origin sat at [0, pivotY, 0] -- the post's centre
line -- while every piece of the board and its chains is authored hanging
from the cross-bar at z = reach * 0.86, half a metre away. At rest that is
invisible. The moment the swing action runs, the assembly turns about a
point it does not hang from and the tops of the chains sweep off the bar
they are looped through. The support audit had been reporting this as a
detached board only because the sheared curl was accidentally filling the
gap; with the curl the right size, it failed honestly. The origin is now
the cross-bar and the geometry is written relative to it: same place in the
world, correct axis to turn about.

Two proportions with it. The post was 62 mm square carrying a 0.72 m board
two metres up -- thinner than the board is thick, and part of a habit of
under-sizing timber that the bench, the table and the fence share. And the
chain drop was 0.12, so the five links that connect board to bracket were a
smudge; the chain is half of what makes a hanging sign read as hanging.

Lengthening the drop grew the links from radius 0.0195 to 0.0434, at which
point they interlock in earnest and their six-sided facets, all built at
the same phase, met coplanar. Each link is now indexed differently about
its own axis -- deterministically, so the seeded stream is untouched, and
a real chain does not index its links either.

27 models, 18369 triangles, all checks pass, 0 floating cases, build clean.
2026-08-22 13:40:31 +07:00
CRT
88147f759e Give the bale back its cylinder, and its cords back their thickness
Against a photograph this was the worst model in the kit: an amorphous
pale lump with spikes in it, where the reference is plainly a drum of
straw with three cords round it.

Three numbers were doing the damage. The body is a seven-sided lathe --
too few sides to read as round on its own -- and it was then displaced by
0.045 of the height, which is 19 mm against a radius of 210 mm. A tenth of
the radius is not texture, it is deformation, and it left no axis to
recognise. Meanwhile the ends were pulled in by a third, turning the
silhouette into a cinched sack rather than a bale with the broad flat ends
the cord holds it square against. Thirteen sides, a sixth of the
displacement and a third of the end fade, and the form comes back. The
stray stalks are what should say "loose straw", and they can only do that
against a body that holds its shape -- so they are shorter too, since at
0.30 of the height they read as spikes driven in rather than stalks
working loose.

The cords were worse, in the quiet way. The ring radius was written as
`0.5 * cinch(x) + cord * 0.35`, which adds four millimetres to a
normalised 0.5: the body is built in radius-1 space and squashed
afterwards, so anything radial has to be in those units. The hoop stood
0.0044 proud before the squash and 1.85 mm after it. Three cords have been
modelled, merged, exported and invisible in every render this model has
ever produced. They are now expressed in the body's own units and sit in
the groove they pull, which is what a tie does.

27 models, 18257 triangles, all checks pass, 0 floating cases.
2026-08-22 13:30:48 +07:00
CRT
f136250707 Make the besom spray, and stop cutting its twigs to one length
The head read as a lampshade pushed onto a stick, and two numbers were
responsible.

The spray angle is derived as atan((tip - tie) / span), which at a tip
radius of 0.102 came to five degrees. Five degrees is not a besom; the
reference fans out until the spread at the sweeping end is comparable to
the length of the bundle itself, which is about twenty. The tip radius
carries that, and the slider range moves with it.

The twig lengths varied by three percent, so every twig finished within a
few millimetres of its neighbours and the bundle ended in a straight cut
across the bottom. That single edge is most of what made it look
manufactured: no bundle of cut birch ends level, and the ragged tip is
what identifies a besom on sight. Lengths now spread by about a third --
downwards only, never above the nominal span. Centring the spread on 1.0
first produced a 1.32 m broom from a declared length of 1.2, which the size
guard caught and which was a real inconsistency rather than a tolerance
problem.

Two supporting changes. The bundle ran from the binding down for 36% of the
broom against roughly half in the reference, so it is longer. And each
sheaf was a 31 mm slat, which at thirty-two of them read as a fan of laths;
they are narrower now and there are forty-six, keeping about the same mass
in the bundle with the pieces at the right scale.

One correction to the record: the head was not, as I first estimated by
eye, a fifth of the broom. Measured, it was 46% of the whole and 36% below
the binding. The eye was reading a diagonal reference against a vertical
render.

27 models, 18021 triangles, all checks pass, 0 floating cases.
2026-08-22 13:25:08 +07:00
CRT
ebc5cbec6d Close the basket's weave, fill it, and stop calling willow straw
Three defects, two of them mine.

The weave had gaps you could see the contents through. The band height was
a flat 0.11 of the model height while the spacing between rows is 1/rows --
0.167 at the default six -- leaving a 12 mm hole between every course. It
was a slatted crate, not a weave. The height now derives from the row
count so rows meet at any setting.

The fruit was invisible, and that was a regression I introduced. An earlier
pass fixed produce that hung level with the rim over a gap by resting it on
the base. That was right; what did not follow was that a basket deeper than
its contents then needs enough contents to fill it. Nine apples in two
layers topped out 12 cm below a 10 cm rim -- correctly supported, and
completely out of sight. The basket is shallower, holds fifteen, and the
layers nest at 0.95 of a diameter rather than 1.5, so the heap now mounds
over the rim, which is the only state a fruit basket is ever drawn in.

Closing the weave then exposed two collisions. Overlapping rows meet in
coplanar faces at the nodes where the undulation crosses zero -- on a
cylindrical basket every row shares one radius -- so alternate rows now sit
slightly proud of and behind each other, which is how a weaver's rod really
passes the stakes. And the rim's inner surface sat at 0.16055 against the
liner's 0.16065: a tenth of a millimetre apart, by coincidence, and the
same 22-sided prism. The rim is now the heavy rod it should be and clears
the liner.

Colour: the basket drew from the `straw` palette and rendered at hue 41,
saturation 0.59 against a reference at 29 and 0.50 -- bright yellow
plastic. Willow is a wood, so both the colour and the material slot move to
oak, which measures 7 and 0.03 away instead of 12 and 0.09.

It still does not look like the photograph, and the measurement cannot say
why. Filtering to pixels above 0.18 saturation discards precisely the grey,
weathered pixels that give the reference its character, so the statistic
reports agreement that the eye does not see. Setting the lift to 0.12 to
make willow "paler than oak" turned it salmon -- raising HSL lightness
holds saturation, so a saturated brown climbs towards peach. Sweeping the
lift over 0.00 / 0.04 / 0.08 moved the measurement by 0.01, which is the
clearest evidence that this axis is not where the remaining gap lives.

27 models, 17853 triangles, all checks pass, 0 floating cases, build clean.
2026-08-22 12:08:56 +07:00
CRT
01d00cdd26 Give the tankard a handle you could hold, and the lantern one at all
Both models had something in the place a handle belongs, and against a
photograph neither was a handle.

The tankard's was a flat box put through `bendGeometry` and then pushed
clear of the body by a hand-derived offset. Measured, the result stood
7 mm proud of a body 47 mm in radius, with the rest buried in the stave
wall -- no finger gap at all, which is why it read as a plank glued to the
side. The offset formula did not describe what `bendGeometry` does to the
geometry, and rather than deriving a better correction for a curve that
cannot be inspected, the handle is now an actual arc from
`arcBarGeometry`, specified by the two things that matter: where the ends
land and how far the belly stands off. It now stands 24 mm clear. The
proportions go with it -- the reference is a fist-sized vessel, ours was a
beaker.

The lantern's was a horizontal `bandGeometry` ring sitting on the flue: a
suspension eye, not a bail. A lantern is defined by being carried, so the
bail is the first thing the eye looks for. It is now a vertical arch whose
sweep runs about 20 degrees past a half circle at each end, which turns the
ends downwards and outwards so they cross the roof cone instead of sitting
tangent to it -- a clean half circle was rejected by the support check as
floating, correctly.

Removing the ring stranded the tongue that used to join it to the flue, and
leaving it in place produced the z-fight: its sides sat at z = +-0.00488
against the bail's +-0.00512, a quarter of a millimetre apart across the
whole apex. A bail that reaches into the roof needs no strap to hold it on.

Worth recording how that one was found. The first guess was that
`arcBarGeometry` self-intersects once the sweep passes PI. Building the
bail in isolation across six sweeps and segment counts gave 0 overlaps
every time, which killed the theory before any time went into fixing the
wrong thing.

27 models, 17433 triangles, all checks pass, 0 floating cases.
2026-08-22 11:54:54 +07:00
CRT
f64fdfd1eb Tune the oak palette against measured references, and record what did not work
With a reference photograph for every model it became possible to ask a
question the kit could not previously answer: is the colour right? Not
"does it look nice", but how far is it, in degrees and in saturation,
from the real material.

Our oak hue turned out to be correct already -- 26 degrees against 27 in
the photographs -- but saturation ran a median 0.08 high across ten wooden
models, worst on the ones that are almost entirely bare timber: the ladder
and the fence were each 0.16 over. That is the gap between weathered oak
and new pine. After the adjustment the median difference is 0.003 and the
mean absolute error drops from 0.097 to 0.055.

The same measurement said straw and linen were too yellow and too washed
out, and acting on it made them visibly worse, so that change is reverted
and the reason is written into the palette. A bale photographs as
desaturated because it is thousands of straws each shadowing its
neighbour; pushing a flat lowpoly surface to that number does not
reproduce the texture, it just makes the surface garish. The statistic was
real and the inference from it was wrong.

Two further measurements closed off the remaining colour theories. Our
saturation spread is 93% of the references', so the surfaces do not lack
variation, and the hue was never off. Whatever still separates these
models from their references is geometry, not colour -- which is where the
next pass goes.

Also adds the remaining reference shots: all 27 models now have one.

27 models, 17393 triangles, all checks pass, 0 floating cases.
2026-08-22 08:14:16 +07:00
CRT
75909a6219 Fix the log pile's axis, and finish translating the repo
The pile was the model I had been told twice was wrong and had twice failed
to find anything wrong with. A reference photograph of a real woodstack
found it in one look: ours was five long sticks lying across each other,
and the reference was forty short logs nested into courses.

The cause was a single rotation. `latheGeometry` builds along Y, and the
logs were turned with `rotateZ(PI/2)`, which sends +Y to -X -- so every log
lay along the X axis. That is the same axis the layout spaces them along.
Each log was 0.62 m long and its neighbour's centre was 0.13 m away, so
every log ran through roughly five of its neighbours. `restingHeight` was
right the whole time; it was solving tangency for an orientation the
geometry did not actually have. Rotating about X instead sends +Y to +Z,
which is the axis the solver assumes, and the pile immediately settles into
courses that rest on each other.

Nothing in the check suite could see this. Support saw one connected mass
and passed it, which is exactly what a heap of interpenetrating logs looks
like to a connectivity test.

Two smaller things the same photograph showed. The file header claims the
sawn ends are "far lighter than the bark" and that the pale discs are the
first thing you see; at +0.07 lightness they were a slightly different
brown, so the offset now delivers what the comment promises. And twelve
logs read as some firewood rather than as a stack, so the default is denser.

Also: .gitignore and .gitattributes were still entirely in Turkish. The
earlier translation sweep matched on file extension and never looked at
extensionless files. src/viewer.css tripped the scan too, but that was 80
occurrences of the CSS `var()` function, not Turkish.

27 models, 17393 triangles, all checks pass, 0 floating cases.
2026-08-21 19:34:44 +07:00
CRT
a0659e24ac Enforce "nothing floats", and rebuild the barrel from a reference photo
Two rules were missing, and each one had been quietly failing for a while.

The first is structural. Every check in this repo asked about surfaces --
winding, coplanarity, triangle counts -- and none of them could see a piece
of a model hanging in the air with nothing holding it up. A chest lid
separated from its chest passed everything we had. scripts/support.ts
voxelises the model, finds the connected components of occupied space and
requires each one to reach the floor; scripts/audit.ts runs that across
default, min, max, both sweep ends, actions and the midpoint of a morph, so
a variation cannot break what the default gets right. That audit started at
22 failing cases: the bell and the sign had no supporting structure at all
and now have a frame and a post, the lantern was in four pieces, the sack's
collar floated above its body, the basket's fruit hung level with the rim
instead of resting on the base, and the ladder's rungs were sized by a
formula that disagreed with where the rails actually were. It now reports 0.

The second is that we had no target. The feedback loop was an offline
render of our own output, which can only answer "is this broken", never
"is this right". Given a reference photograph of a cask, five defects were
obvious in seconds that three rounds of self-review had missed: four hoops
where there should be six, no rivets at all, a chime cut so shallow it did
not read, and a stave count low enough to leave a visibly polygonal
silhouette. All four are fixed here.

Two things the rivets taught us, both kept as comments rather than as
scars: a stud that tapers outward tilts its side normals back towards the
axis, which the radial check correctly reports as reversed winding -- a
real rivet head flares, so the honest shape is also the passing one. And
once seated deep enough to look right, the stud's buried inner cap falls
inside the shell the check inspects, so it is not generated at all, exactly
as bandGeometry already declines to build the inside of a hoop.

27 models, 16973 triangles, all checks pass, 0 floating cases.
2026-08-21 19:23:11 +07:00
CRT
50c7fc60b9 Translate the whole repository to English
The kit is going public, so everything a reader sees is now English:
27 model sources, the core vocabulary, meta.ts (titles, descriptions and
every slider label), the viewer UI, all script output, REFERENCE.md and
the READMEs.

Rules applied: only comments, doc comments and human-facing strings were
touched. Identifiers, numbers, logic, registry item ids, part names and
material slot names are unchanged — those are a published contract and
were already English.

Verified rather than assumed: for all 53 changed .ts files the code
skeleton (source with comments stripped and string contents replaced by a
placeholder) is byte-identical before and after. Full gate is green —
typecheck, registry conformance, ~510 model checks, GLB round trip and a
production build.

Also fixed here:
- index.html and viewer.html were missed by the translation pass because
  the file list only covered .ts/.md/.css. Both had lang="tr" and Turkish
  titles.
- The viewer formatted triangle counts with toLocaleString('tr-TR'), which
  prints 15.425 rather than 15,425.
- REFERENCE.md's model table header did not match what
  scripts/catalog-table.ts emits, so the next regeneration would have
  silently changed it. Aligned to the generator.
- Removed scripts/probe.ts and scripts/zdiag.ts; they were throwaway
  diagnostics that got committed by accident.
2026-08-21 14:35:13 +07:00
CRT
72f219d087 Süpürge, çapa ve çiti yeniden kur; kapanma/alaca parametrelerini sınırla
Üçü de render'a bakılarak elden geçirildi. Ortak sebep: her birinde
modelin KONUSU eksikti.

çit — birleşim yoktu. Dört kare çubuk ve önlerine konmuş iki lata;
derinlik/boy oranı 54:1, yani karton. Artık gerçek zıvana var: direk iki
yanak + köprü bloklarından kuruluyor, delik geometrik olarak var oluyor,
kiriş içinden geçip uçlarda karşı yüzden taşıyor. Ayrıca kiriş dağılımı
formülü kaç kiriş olursa olsun hep 0.28–0.78 aralığını dolduruyordu —
üstü ve altı doldurmak yapısal olarak imkânsızdı. 440 → 860 üçgen,
derinlik 0.09 → 0.31 m.

çapa — kaz boynu yoktu. Ağzı sapın ekseninden öne taşıyan kıvrık boyun
olmadan model "kürsü / yol tabelası" diye okunuyordu. Boyun latheGeometry
ile kurulup bükülüyor, ağız boyun ucunun teğetine hizalanıyor; ikisi de
hesapla, gözle değil.

süpürge — demet içi boş bir koni kabuğuydu: bütün levhalar tek halkada.
Üç eşmerkezli halka (6/10/16) kütleyi kuruyor. Bağlar artık demetin dış
yarıçapına oturuyor; eskiden ayrı bir tahmin formülüyle levhaların İÇİNDE
kalıyor ve hiç görünmüyorlardı.

Ölçerek bulunan iki gerçek hata:

- bendGeometry'nin etkisi geometrinin y=0'a göre nerede durduğuna bağlı.
  Orijinde ORTALANMIŞ bir gövde simetrik bükülüyor: iki uç aynı yöne
  gidiyor, siluet değişmiyor. Çapanın ağzında Z aralığı 0.0337'den
  0.0327'ye DÜŞÜYORDU — yani yorumda "sorunu çözüyor" dediğim kavis hiç
  yoktu. Tabanı orijinde kurulunca kaçış 44 mm. Yardımcıya yazıldı.
- Süpürgede açılma işareti tersti: rotateX(+f) levhanın ucunu -Z'ye
  atıyor, rotateY onu EKSENE doğru çeviriyor. Levhalar ekseni geçip karşı
  tarafa savruluyordu; demetin kabuk olmasının asıl sebebi buydu. Gözle
  değil ölçüyle yakalandı (genişlik 0.23 beklerken 0.13).

Sistemik: kapanma yarıçapı ve alaca hücresi modelin en büyük ölçüsünden
türetiliyordu. 4.89 m'lik çitte hücre 0.27 m çıkıyor, yani bütün direk tek
hücreye düşüyor ve doku sistemi hiçbir şey yapmıyordu; kapanma yarıçapı
0.69 m ile her şeyi battaniye gibi karartıyordu. İkisi de mutlak sınırlara
alındı — ahşabın damar lekesi nesnenin boyundan bağımsızdır.

Doğrulama: test yamalarının gerçek config alanlarına işaret ettiği artık
denetleniyor. Çapa yeniden yazılırken `bladeAngle` alanı kalkmıştı ama
test hâlâ onu yamalıyordu, yani o varyant sessizce varsayılan modeli
sınıyordu.

render.ts: --angles ile turntable. Tek bir 3/4 açı yanıltıcı — çit ancak
tam yandan bakınca cılız görünüyor.
2026-08-21 14:05:22 +07:00
CRT
8a982160ac wicker-basket ekle ve dokümanı 27 modele güncelle
Kullanıcının prop listesindeki son madde meyve/sebzeydi; sepetle birlikte
kapatıldı çünkü kitte zaten sepet eksikti.

Örgü, biçimi yapımdan türetiyor: yatay halkalar önce düz bant olarak
üretilip köşeleri açılarına bağlı olarak içeri-dışarı itiliyor
(yarıçap × (1 + genlik·cos(dikeySayısı·açı + faz))), ardışık sıralarda faz
π kaydırılıyor. Dikey çubuk başına iki segment yeterli — dalga tam olarak
her çubukta bir artı bir eksi örnekleniyor.

Halka başına iç yüz üretmek altı sırada ~800 üçgen tutuyor ve modeli
lowpoly bütçesinin dışına çıkarıyordu; yerine 44 üçgenlik tek parça iç
astar var.

Yeni `produce` materyal yuvası (mumsu kabuk, roughness 0.52). Meyve rengi
`hue` alanından geliyor: aynı modelden elma, şalgam ya da lahana sepeti.
2026-08-21 09:50:34 +07:00
CRT
a3300ff233 Kit'i 26 modele çıkar, doğrulamayı ve inceleyiciyi genişlet
Modeller (13 → 26): chest, bench, torch, lantern, bell, sign, hay bale,
sack, broom, tankard, book, phial, coin pouch. Kürek, çapa ve yaba
render'lara bakılarak yeniden yazıldı.

core:
- createKitModel artık tipli `actions` ve `update(dt)` taşıyor
- BuiltPart.extras: bir parça birden çok materyal yuvası kullanabiliyor
  (sandığın kapağı = meşe tahta + demir kayış, birlikte dönmek zorunda)
- BuiltPart.origin: parça başına dönme merkezi (menteşe, sarkaç)
- bendGeometry / roughenGeometry / mottleGeometry
- 5 yeni materyal yuvası: steel, brass, straw, cloth, leather, glass
- bandGeometry({ inner }) — serbest halkalar kapalı katı olabilsin

my-registry/meta.ts: katalog metadata'sı artık build.ts ve viewer için
tek kaynak. Doğrulama modelle metadata'nın uyuştuğunu denetliyor ve bu
eklendiği gün dört modelde gerçek bir sapma yakaladı.

Doğrulama (~500 kontrol):
- metadata uyumu, eylem semantiği, kare hızı bağımsızlığı
- scripts/render.ts: tarayıcısız yazılım rasterleyici. Bütün doğrulama
  geometrikti; "bu kürek küreğe benzemiyor" diyebilmek için modele
  bakmak gerekiyordu
- scripts/verify-glb.ts: dışa aktar → geri oku → karşılaştır

GLB dışa aktarımı: viewer düğmesi ve toplu CLI aynı kodu kullanıyor.

Viewer:
- parça paneli (solo/izole) ve materyal yuvası paneli (canlı renk override)
- kit görünümü: 26 model tek sahnede, ölçek ve ton tutarlılığı için
- configure() sonrası gölge bayraklarının düşmesi düzeltildi
2026-08-21 09:44:58 +07:00
CRT
98cfbab180 Initial commit: @medieval-kit registry and viewer
A lowpoly medieval procedural model library for Vibe3D, plus the model
inspector used to build it.

Registry (@medieval-kit, published as @medieval-kit/registry):
- core: deterministic mulberry32 RNG, slot-based vertex-coloured materials,
  a geometry vocabulary (box, prism, stave, band, head, flip), and the
  stable-anchor part slots the Vibe3D protocol requires
- wooden-barrel: 806 tris, 13 separate staves, iron hoops, recessed heads
- wooden-crate: 360 tris, plank rows on corner posts, forged straps
- iron-brazier: 423 tris, typed actions, per-frame flame flicker, own light

The kit declares no renderer capabilities, so it runs on WebGLRenderer as
well as WebGPURenderer. Colour variation lives in vertex attributes rather
than in materials, so each model keeps one draw call per material.

Verification (bun run verify) covers 64 checks without a browser: geometry
validity, winding by two measures (radial alignment for bodies of
revolution, signed volume for closed solids), coplanar-face detection for
z-fighting, stable root and anchor identity across configure(),
deterministic seeding, material ownership, and idempotent disposal.
2026-08-21 06:15:58 +07:00