One paragraph in the asset pipeline so the next fresh session does not
rediscover it: prompts ask for magenta beds ON PURPOSE, _fine.key's grey
repaint defeats them unless bed=True is passed, and the solid white paddles
were exactly that interaction.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's batch, delivered in dependency order. "The opponents look pretty
harsh, almost already pre-zombie" — so BOTH tennis opponents got a new
identity: warm skin, human eyes, an easy near-smile, rolled anchor-first so
every later sheet inherits the face. The male's bounce pair also ADOPTS the
female's palm-out posture (his old crouched receiver stance was the wrong
pose for bouncing a ball — her better one is now the shared convention), and
the pre-serve ball was re-measured onto the drawn palms: BOUNCE_IDLE's oDx
0.45 -> 0.24 / oApex 0.55 -> 0.72 for his close carry, with her wider reach
riding TOURS (oDx 0.48 / oApex 0.73) the same way the player's pair already
does.
"parts of the body as well as the inner of the racket are solid" — the beds
were the cutter's own doing: _fine.key repaints enclosed magenta pale grey
(protection against accidental magenta strings), which defeated the beds we
now draw magenta ON PURPOSE. key()/main() grow a bed=True switch (the
famroll convention, used by _runroll and the refresh cuts); the run cycles'
beds went from ~70% to ~44% opaque — strings, not paddles. Enclosed pockets
of a roll's off-spec background (one sheet came back on blue) are scrubbed
by corner-colour distance too.
Both RUN CYCLES re-rolled off the friendly anchors: his landed 8 clean
frames (racket +0.05..+0.30, one side); hers took four rolls — the gates
caught a white background, a crossing racket, near-dup neighbours — and the
keeper drew a 10-frame cycle, from which the best 8 were chosen by
maximising the minimum adjacent stride (29.7%, smoother than his 22.7).
o_run7/8 return to the SPR list.
AND THE CHASE STOPS DITHERING (Oliver: "switching between a left and right
movement increasingly faster the nearer the ball comes"): intercept()
re-solves every frame and the earliest-reachable point jumps around its
solution as the flight shortens, flipping his run direction each time. The
CPU now commits to its stored target unless a fresh solve moves it by more
than 0.35 m. SPRV 121, main.js v244, 66 tests, every opponent frame verified
200 in the live game.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Her run cycle carried a DETACHED SHOE floating behind her in every frame —
a cut that kept two blobs per figure — spotted while fixing the men's
flicker. Four rolls to a keeper: roll 1 closed on a duplicate (8->1 at
4.4%, gated), roll 2 came back a grid, roll 3's grid was re-laid into one
row locally but its racket flipped sides frame to frame (-0.42..+0.31, the
exact "flashing left and right" the warning exists for), roll 4 landed:
eight frames, racket steady at +0.05..+0.17, one face, no luggage. SPRV 119.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver: "the serving and running of the opponent during normal tennis still
feels odd. which spritesheet was used?" Two sheets, two defects. The run
(o_run1-8, _refcheck/run_o.png via _runroll.py) FLICKERED IDENTITY — frames
alternated a tanned living face with a grey red-eyed one, because the
identity reference bakes in o_toss/o_smash and THOSE were the second defect:
the serve pair came from the legacy opponent_toss regen batch in a chibi
register (big head, heavy outline) that matched nothing around it.
Fixed in dependency order. The serve pair re-rolled pose-style (o_idle1 the
identity anchor): the batch's kit gate rejected all three tries but try 3
was right by eye — the "bright" channel it tripped on is the white-filled
string bed the run frames already carry — so it was installed on the montage
verdict, which is the pipeline's own rule. Then _pose_ref_run_o recomposed
off the cleaned frames and the run re-rolled: SIX frames, one face, one
racket side (+0.18..+0.24), real strides. The model drew 6 and a clean 6
beats a fought-for 8 (the player's cycle is already 6) — o_run7/8 deleted,
the SPR list trimmed, render's run8/run6/run4 probe does the rest. All six
verified 200 in the live game. The women's cycle (a floating shoe behind
her in every frame) is still rolling and lands separately. SPRV 118.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two phone verdicts. "easy is now way too easy" — its first cut was tuned
before HARD taught us which dials bite, and with reaction 0.62 and a 0.62
quality cap it had become a training dummy. EASY is a mild step below MEDIUM
now (flub 0.18, quality 0.76, reaction 0.54, speed 0.9): beatable without
being able to fall asleep on it. And the quit chip sat exactly on the BLOOD
panel's corner in the splatter modes — it rides 40px down now, below the
pips' strip, one position with one meaning in every mode. Verified live over
a zombie run: pips and door stacked, nothing touching.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver benchmarked HARD against the game's own roster: "the runner in zombie
mode is way faster than the regular player in hard" — and the numbers agreed.
The RUNNER covers 6.15 m/s (racketSpeed x rung x foe.speed) with no reaction
gate; HARD's classic CPU did 5.59, only after its 0.22 s read, and ambled
back to centre at HALF speed between shots — that recovery jog was the free
points. HARD now moves 4.3 x 1.5 = 6.45 m/s, and DIFFS grows `recover`: the
between-shots factor, 0.85 on HARD where the shipped 0.5 stays pinned on
MEDIUM. Wrong-footing him still works; strolling him side to side no longer
does.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
"all tennis screens are missing the ability to exit the game and return to
the menu — right now I need to force close the app." A quiet x chip rides the
top-right corner of every run (the scoreboard owns the left; the desktop key
hint shifts inboard). Tapping it — or Escape on desktop — freezes the world
with the existing paused flag and asks LEAVE THE MATCH?: BACK TO THE MENU in
the menu's crimson, PLAY ON underneath. Quitting is the death path minus the
death — restartRun, refreshSplash, splash — and banks nothing: an abandoned
run never happened, statistically. Verified live both ways: PLAY ON resumes
the frozen rally, quitting lands on the splash with the chip retired.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
"the hard mode is still too easy — make it really challenging." The dial that
was missing is aim: the CPU rolled a random lane on every shot, so a third of
everything came politely back to your racket regardless of the other six
numbers. DIFFS grows `smart` — the probability the CPU plays the OPEN court,
the corner away from P.x — and HARD sets it to 0.8: four shots in five now
make you run. EASY and MEDIUM keep random aim (MEDIUM pinned by test), so
the shipped classic is untouched below the top chip.
The rest of HARD tightened in the same pass: reaction 0.22 s, speed 1.30,
reach 1.85 m, quality floored at 0.70 under a cap of 1.0, serve 1.15, flub
one ball in a hundred. It reads your shot almost instantly, covers what it
reads, and places its reply where you are not.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
"using the difficulty in classic tennis I don't feel any difference" — and he
was right, verified live: G.diff arrives intact, the dials were just the
wrong ones. flub/cap/speed barely register: the quality cap trims a
distribution whose 0.45 floor never moved (EASY and HARD averaged 0.59 vs
0.70), and ±12% on aiSpeed never binds because the CPU pre-plans its
intercept. What a player reads as SKILL is how fast the CPU reacts to the
shot (aiReaction — the config's own comment says corners are winners because
of it), how long its racket is (aiReach), the FLOOR of its shot quality (a
raised qMin means every ball comes back with pace), and the serve — none of
which the first cut touched, and the serve was hardcoded 0.6-0.95 besides.
DIFFS carries seven dials now with real spread: EASY whiffs one ball in
three, reads late, swings short and serves soft; HARD reads in 0.28s, swings
1.75m, floors quality at 0.62 with a cap of 1.0, and serves 10% hotter.
MEDIUM is pinned dial-for-dial to the shipped TUNE values by test, so the
middle chip remains exactly yesterday's classic. All G.diff-guarded, so the
racket zombie and gore's opener keep their own numbers.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver named the wizard's wrongness precisely: a pre-highlighted card that
also navigates is "a checkbox and a button combined in one". The border said
"this is a setting", the tap said "this was an answer", and the screen left
before the setting-reading of it could cash out.
So the two roles live on different screens now. The question screens (PICK A
GAME, DIFFICULTY, the zombie pair) carry NO selection: every poster and chip
is full-bright and equal — the dimming went too, it read as disabled — and
tapping answers and advances. The remembered run survives only as a small
LAST tag in the card's corner: an indicator, not a selection. The final
screen keeps its acid borders honestly, because SERVE UP is a real confirm —
the rule is "a border means SERVE UP will use this; a bright poster means
tap me to answer". Question screens rebuild on entry so LAST follows what
was actually played, and the first card clears the header by the margin the
old h3 rows used to give (Oliver's padding note).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
"the mode setup is really overwhelming now" — three posters plus a sub-row
plus player tiles plus court meters was a wall. Discussed, Oliver picked the
wizard with tap-advance over next-buttons and the accordion: PICK A GAME
(three posters, now 148px tall — alone on a screen they get to be posters),
then the mode's own second step only when it has one (classic: DIFFICULTY
chips stacked tall; zombie tennis: ARCADE and THE TOURNAMENT as full posters;
gore skips straight through), then MATCH SETUP holding player + court +
SERVE UP with the mode riding the summary footer.
Tapping selects AND advances — one decision per screen is the wizard's whole
promise, and a NEXT under a single row of choices is a second tap for nothing.
Back walks the chain the way it came (who-and-where returns to the second
step only when the mode has one). Every step comes up pre-selected from last
time, so a returning player is three taps from serving. Three new stages on
the intro's existing data-stage switcher; splash Enter goes to 'mode' now.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's restructure: mode first, then what the mode asks — classic deals
EASY / MEDIUM / HARD chips, ZOMBIE TENNIS deals two sub-mode posters (ARCADE,
the exhibition ladder as was, and THE TOURNAMENT), and SPLATTER GORE stands
alone as the third top card. config.MENU is the grouping, MODES stays the four
run modes, and pickedMode stays the run-mode id — start(), stats and the
summary never learn the menu got taller. The sub-row sits inset under the
posters; sub-cards are the poster at 2/3 height centred on their scene.
DIFFICULTY is the one real gameplay addition: config.DIFFS scales the three
dials the classic CPU already reads (flub, quality cap, aiSpeed), MEDIUM
pinned by test to the shipped classicFlub/classicQualityMax so the middle chip
IS yesterday's game. Rides on G.diff next to G.classic, so restarts keep it.
AND THE FEEL SPLIT OLIVER REPORTED IS REAL: "in classic tennis I tend to miss
balls way more often" — the match's three contact helpers (matchReach 1.35,
the matchAssist floor, the short-ball chase) were all G.match-gated, so
classic played with a 35% smaller racket and no rescue on the dying drop
shot. rallyFeel() = match || classic now carries all three; gore's opener
keeps the old prologue feel on purpose.
The ARCADE poster took two rolls: the first came back a flat sticker line-up
("very constructed, not artsy"), the second anchors on the zombie poster
itself as a MOOD reference — one challenger lit at the net, the queue receding
into purple haze. Rolls in _refcheck/modebtn. The cup's stats mapping also
joins MODE_MATCH (it banks through endRun as MATCH and read back as GORE).
SPRV 117, main.js v233, 66 tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The bottom-anchored button shows the lower ~60% of a poster, and the classic
scene was composed sky-first — the serve lived in the top half and the crop
took his head ("the classic player head is cut"). Fixed in the ART, not the
CSS: one edit-mode roll on the existing image, the player shrunk and moved
down to stand fully in the lower half, sky and crowd untouched. The other
three scenes already composed bottom-heavy and keep their crop. SPRV 116.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
"the subtitle is barely readable" — cream at 80% over a pixel crowd was too
polite. White at full opacity, one size up from the court rows, and a solid
1px black outline all round plus the soft shadow, so the mode's one-line rules
read on every scene including the classic crowd, dimmed or lit.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver, off the first live look: a little higher, and align the art at the
bottom. 78 -> 96 px min-height, background-position center bottom — the scenes
now lead with their court instead of their sky, which suits three of the four
(the gore court's blood and net especially). The classic serve's head goes
over the top edge with the bottom crop; his scene was composed sky-first. A
per-mode position or a re-roll fixes that one if it grates.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's design call for the setup screen: the mode buttons should POP — a
bright full-bleed background symbolising each game mode, the description
written over it. Four scenes rolled through the _famroll OpenRouter route,
each anchored on its mode's own sprite for style, no text baked in:
CLASSIC a sunny sports-poster serve under a blue summer sky
ZOMBIE the turned opponent at an acid-green floodlit night net
CUP the champion's silhouette looming over the gold trophy
GORE the horde pouring over the net into a blood-red sunset
Installed as mode_*.png at a third of the 2K roll, quantised to 128 colours
(2.1 MB of menu art became 265 KB; pixel art quantises for free). The button
is the banner now: art + scrim gradient as one background stack, caption
bottom-left with a text shadow, and the unpicked modes sit desaturated and
dimmed so the chosen one is the thing that glows — the menu's own "one bright
thing means selected" rule, applied to pictures. The sprite-tile row and its
`tile` field are gone from MODES; raw 2K rolls kept in _refcheck/modebtn for
re-cuts. SPRV 114 -> 115, main.js v229. Verified in the browser: draw, select,
summary line, no console errors.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver: the tennis-family sprites are optimised for movement, not gore — the
card should lead with the zombie. Each family already anchors on a horde body
(_famroll.FAMS reads its character from one), so _heads.py now cuts the icon
from THAT sprite instead of the fam_ sheet's tennis stance: the face on the
card and the monster on court are the same person, in the gorier print.
No resize, no requantise on the way: drawCard scales the PNG itself with
smoothing off and only reads the aspect, so the crop ships as a native-size
window onto art that is already flat pixel art. The top-down head detector
survives for the four bodies it works on; seven get hand boxes, because the
horde bodies carry PROPS — the keeper's ball launcher and the physio's
stretcher ride above their heads, the director holds a walkie-talkie level
with his, and the brute's and champion's shoulders are wide enough that a
full-silhouette crop at head depth came out 2.2:1 and starved the card title
of room. SPRV 113 -> 114.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's concern, after the un-chunking: a fresh session walked straight into
_regen.py and diagnosed from there, twice — first the legacy prompts, then the
model's rolls — while the real bug sat in the cutter, provable by putting any
fam_ sheet next to the sprite cut from it. The family-pipeline section already
demoted _regen.py, but 270 lines below the "Asset pipeline" heading, and the
session memory that routed the last investigation pointed at _regen by name.
So the routing is now the FIRST thing under "## Asset pipeline": a
which-tool-for-which-job table that names _famroll -> _zsheet for families,
marks _regen.py as the single-sheet remaster tool it still is (bosses, players,
crowd, weapons — those jobs keep it alive, so it stays), and states the
diagnostic rule that would have saved two days: when installed art looks worse
than the _refcheck sheet it came from, the suspect is the CUT, and a re-cut is
free. _regen.py's own header says the same, for whoever opens it first anyway.
The "Only ink votes" lesson joins the family-pipeline section with the numbers:
0.0% of ink blocks uniform at b=11 against 99.2% of empty ones is how a
detector proven on single-row sheets quietly broke on 9-row geometry — the
threshold never changed, the data under it did.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver put two pictures side by side — z_rufall3 as the fam_ru sheet drew it,
and z_rufall3 as it shipped — and they were not even close. The sheets were
never the problem: every one of the eleven family rolls came back FINE, and the
pipeline pixelated seven of them on the way down.
_fine.block_size() scores b x b blocks by uniformity, and it scored ALL of
them. key() zeroes the background, a zeroed block is uniform at every b, and on
a 9-row family sheet the background is ~85% of the bounding box — so the
detector's 0.82 bar was cleared by the background alone, at any block size it
tried. Measured on the ru sheet: 0.0% of ink blocks uniform at b=11, 99.2% of
empty ones, overall 0.821. Detected block: 11. _zsheet then honestly snapped
428 px figures down to 39 art px and scaled them back up — the pixelation was
MANUFACTURED there, not drawn. br survived at 0.775 (its ink coverage is 4
points higher) and the whole per-family chunkiness table — br fine, ch/ru/sh/st
at 6 px blocks — was ink-coverage noise around one threshold.
The fix is one line where every caller routes through: background-only blocks
get no vote. Then every family re-cut from the sheets already on disk — no new
rolls, no tokens — and all eleven came down at art block 1. br's re-cut is the
regression check: byte-identical result. Dimensions drift at most 3 px, so
nothing on court changes size. SPRV 112 -> 113 so phones drop the year-long
cache on the chunky art.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's tournament freeze was the stale latch by a THIRD route:
matchFallWillPlay kept the deathmatch /4 while cup matchPoint divides by
4*need, so it promised a fall on a no-crossing point — whose synchronous
matchReady left the after-set latch scoping a dead rally. No more ordering
arguments: the latch is set BEFORE the resolve, the swallow exempts the
in-progress resolution via G.matchEarly, and every path that arms a rally
clears it in matchReady. willFall now mirrors the cup divisor exactly and
knows a mid-round game keeps its body. Regression test drives the full QF
sequence (no-crossing points, the fall, the mid-round game, game two).
And the two 'not really' reports: the trail's 0.26 alpha was tuned on the
gore palette and vanished on the bright lawn — 0.38 now, gate at 13 m/s;
oppBright 1.18 -> 1.3.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The fourth mode: four rounds up the dead hierarchy (CUP_ROUNDS), each round
drawing a rung from its band, the champion always the final. It rides the
match machinery whole — cupDraw feeds the drawn rung through d.opp so every
ladder dial scales untouched; the body's pool spans the round (need-scaled
pool and divisor, so wounds carry between games); cupGameWon holds the same
body mid-round, kills it on the round, and past the final cupWon banks a
NOT-dead match run under CHAMPION OF THE DEAD CLUB. Blood carries; round one
is a single game; deathmatch stays as the arcade run. Full-run state test.
Sidequests: the ghost trail is pace-qualified (TUNE.trailSpeed) so both
tennis modes streak — pow>0.8 was only ever reached by the gore slingshot;
opponent sprites draw ~18% brighter via a load-time baked copy
(render.brighten, TUNE.oppBright) with no PNG touched; the cup scoreboard
line names the round instead of lying with deathmatch arithmetic; the rack
and the apron demo no longer draw under a trophy screen.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Once the player's winner is past the net its landing is closed-form knowable
(predictLanding is exact), so a ball that will bounce in beyond the body's
provable reach resolves while still in the air — the collapse starts as the
ball flies past, roughly half a second earlier than the bounce-1 call, which
stays as the fallback for anything too close to call. Same conservative
margins, same willFall gate, same decided-latch discipline.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver, three refinements in one breath: the frame PROGRESSION quicker
(drop 0.06+0.25 -> 0.05+0.15 — against a live rolling ball the slower drop
read as mush), the prone hold KEEPS its bench-tuned 2 s ('the lying itself
can stay that long'), and the rise row is cut ('the opponent respawns
anyway') — done() re-marks the body in its wound skin regardless, so rising
first was a respawn with extra steps. The rise frames stay in the sheets
for the bench. Wound-beat test updated to the new contract.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's retest: 'the fall never happens, the game does not continue, the
ball is gone.' Root cause: the decided latch was set AFTER resolveInterlude,
and on a point with no stage change matchPoint re-arms the next rally
synchronously — so the latch outlived the rally it scoped and swallowed
every resolution after it.
Two changes: matchFallWillPlay() gates the early resolution to points that
actually put a collapse (or the game-winning kill) on court — an ordinary
point keeps its ordinary bounce-2 ending, so the ball never vanishes
mid-roll — and with only those paths admitted, the after-set latch is
provably safe: both defer matchReady until their beat ends. A regression
test drives the exact soft-lock sequence through the real update loop.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver: 'the zombie falls after the ball is lost although we know this
earlier — it could play out as ball lost because of the falling.' The match
point now resolves at bounce ONE when the winner is provably beyond the
body's reach (closed-form flight vs its own chase dials, full swing reach
plus 0.6 m of doubt — borderline balls wait for the ordinary bounce-2 call).
startFall keeps the ball rolling for an early-decided point, and the
d.decided latch swallows the rally's later resolutions so it can never bank
twice. Verified in-pane: collapse starts ~0.3 s after the bounce, ball rolls
past the prone body, score banks exactly once.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Five of Oliver's six (the sixth is a play-mode discussion):
- THE DOUBLE FAULTS WERE FRAME QUANTIZATION: the bounce registered a full
integration step after true contact — ~0.75 m of travel at serve pace — so
serves aimed half a metre safe were judged long. Classic enforces the
service box, the match only the court, hence 'zombie tennis does not'.
The bounce now rewinds to y=0 before anything reads the position.
Measured in-pane: 30/40 faults on good taps before, 0/40 after (mistimed
and smudged taps also 0/40; aimed hard serves keep their scatter risk).
Also: the strike flick dead zone is 44 px so a smudged tap stays a tap,
and the aim base clamps 0.18 m inside the lines — scatter, not the aim,
is what faults.
- The mop beat holds the walk-on until the boy actually LEAVES the court
(ordered off at beat end, 6 s cap).
- A match game win that restores blood shows the duel's WOUNDS TREATED
popup; full-blood wins stay silent.
- The opponent walk-on card holds 4.0 s (was 2.4).
- Card heads are cut from the 4K sheets at 144 px (assets/_heads.py,
z_<fam>head.png, HEAD_CROP stays as fallback) — the old 65 px crops were
the 'very low res' heads.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver, after the facing fix: still flashing, female worst. Film-strip
diagnosis in the pane: with facing pinned the racket STILL alternated —
frame 4 of the female cycle tucked it across the body, and half the frames
of both player cycles lost or shrank the hoop to cut fragments (overlapping
figures sliced by the even-spaced valley cuts).
Both cycles re-rolled with three new prompt rules that now live in
_runroll.py: the racket stays right of the midline in EVERY figure (named
again for the airborne pose, where it kept vanishing), and the figures
never overlap (no more sliced hoops — a debris-stripping pass was tried
and removed: after the recrop a real racket is indistinguishable from a
fragment, prevention beats cleanup). The model returns 2-row grids at
times; a coherent band is cropped and cut. Installed: racket-side spread
+0.01..+0.05 across all frames of both cycles (was a full side swap).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver, women's tour: 'the racket she is holding flashing left and right
very fast' during auto-run. The rolled cycles carry a real racket, and the
mirror flip teleports it across the body — the per-frame steering flutter
the old racketless art could hide. earnFlip() requires an auto-steered
direction to persist TUNE.faceFlip (0.13 s) before the body mirrors, at all
three steering sites (player assist, tennis CPU, match opponent); manual
input stays instant. Measured over an 8 s live rally: 4 player flips, 2
opponent flips — every one an actual turn.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver: 'the motion seems very flashy and unnatural'. Two causes, both
measured: drawChar centres a frame by its total width while the racket's
reach varies across the cycle, so the BODY teleported sideways up to 28
source px on a frame swap (the female player was worst) — every frame is now
padded until its lower-third centroid sits at the frame centre (residuals
±0.2 px, the normalisation lives in _runroll.py's cut). And the hardcoded
4.5 cycles/s was tuned on 2-4 near-duplicate frames; on eight real phases it
was a strobe — rolled sets stride at TUNE.runHz (2.4), old art keeps 4.5.
In-pane: body wobble down from ~36 device px per swap to 3.2.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's four picks, each rolled off its own settled BOSS frame through the
family pipeline (rows in _famroll.py) — the rung and the chapter boss are the
same person. matchFoe decoupled from bossArt's cycling on his call: the match
ladder is 12 with the champion capping rung 12; chapters keep their ten.
The champion refused to bulk through four rolls (words lost, then a new
pair-reference in char_ref only half-won), so his mass is stature: OPPONENTS
rows carry an optional world height h read by zombieH — champion 2.1 m, the
Bulk 2.0 m — and the serve aims at the body actually drawn. HEAD_CROP measured
for all four; ramp test extended to 12 rungs.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The old run frames were generated one at a time and it showed: 91/110/92/98 px
of body-width drift, duplicated poses drawn differently, no racket. Each
character (player m/f 6 frames, opponent m/f 8) now comes from ONE 4K call
anchored on a composed identity reference of its settled frames — the broken
run art stayed out of the reference on purpose. Stride-gated, cut with _fine,
judged by eye per roll. drawChar probes run8/run6/run4/run2.
Side effect: the male opponent's old run art faced LEFT against the
art-faces-RIGHT convention — he had been moonwalking right since install.
Fixed by drawing the new rows to the convention.
Pipeline lessons recorded in PROJECT.md (the model negotiates the count; a
back view has no direction; the racket clause goes early and loud).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Three mechanical causes of 'no serve agency', measured then fixed:
- the 0.12 m perfect snap was ~0.3 s of hang time (now TUNE.apexSnap 0.05,
and the gold band reads the same knob);
- the fixed mid-box target depth capped every good serve at ~36 m/s in the
loft loop — player serve depth now rides quality, flush aims 0.3 m inside
the service line (ladder: 49 / 52 hard / 36 mistimed / 30 bad);
- the strike flick was splatter-only — in tennis and the duel it now sways
placement (horizontal) and buys pace (vertical, serveHard + serveHardSpin,
the extra dive IS the clearance), paid for in scatter against the real
fault machinery. Tap stays the safe second serve.
Phone verdict pending — see the PROJECT.md session log.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver's batch, the code half:
- The tennis serve lost its ghost fingers (idle tap demo + apex rider deleted
with their counters; the rally-return hint survives).
- The toss rises over the tossing HAND, measured off the toss frames' raised
fingertips (TOSS.dx/oDx, women's tour override) — not over the head.
- An uncaught toss returns to the bounce stance instead of flashing the smash
frame the player never earned.
- Between match games the court belongs to the mop boy: 5 s (TUNE.matchMopT),
full job allowance, a hurry speed between stains, and the far-court reach
limit lifted for the match (his stains are exactly where opponents die).
The next opponent walks on only after — d.ending guards the empty court.
- The walk-on card wears the opponent's own head: config.HEAD_CROP, straight
crops measured off the family sheets, no new art.
- The match intro is DEATH MATCH, counting the real blood pool.
- Daniel is benched: every ump_* personality line routes through aside()'s
early return; officiating calls keep speaking. Personality pass later.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The flattened body curve stays (farK 0.25 — 'really good, helped a ton'); the
broadcast lean was 'distracting' and its knobs sit at 0 like the aimCam pair.
The plumbing stays for the T panel. PROJECT.md: farK is now a settled feel
decision.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Oliver: the far-side opponent is barely visible in portrait, and the lens was
already at its measured ceiling. TUNE.farK (0.25) grows far bodies ~20% while
the court keeps the true projection — the 16-bit tennis trick, anchored so the
near player is untouched. TUNE.farCamY/farCamZ add a broadcast lean while the
ball is on the far half, pivoted about the player's feet (fOff) so the pull
room the apron math promised never moves. Both live in the T panel; both
awaiting the phone verdict — see the PROJECT.md session log.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two notes off the bench, both about the fall.
THE BLEEDING DID NOT STOP. bleed()'s wound anchors are ABSOLUTE WORLD HEIGHTS —
the headless neck spurt is pinned at y = 1.5 m, the torn arm at 1.15 — so a
corpse lying flat through the whole 2 s floor hold kept spraying from where its
neck used to be, two metres above itself. Oliver: "the dynamic bleeding must
stop during fall as this looks like blood fountains midair." The line calling it
already said "open wounds bleed every frame they're STANDING" and never enforced
it; now it does. Measured on the headless stage: 20 particles emitted while
walking, 0 during the fall. Only the body G.matchFall owns is suppressed —
everything else on court bleeds as before.
THE SLIDE NOW FINISHES BEFORE THE RISE. "The slide must be finished before the
rise" — a body pushing itself back onto its feet is planted, and still drifting
while it stands up reads as the court moving under it. fallSlideMix goes
[.10 .30 .45 .15] -> [.12 .35 .53 0], the rise's share redistributed across the
first three IN PROPORTION so the total travel is unchanged: 0.06 / 0.26 / 0.41 /
0.000 m, still 0.77 m end to end.
Watch fall3: it carries 0.41 m over its 2 s hold now. The ease-out front-loads
it so it should still read as a skid that settles rather than a creep, and if it
drags, share moves back into fall2.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
All of it found from the sprite bench, and most of it was the same mistake
twice: a correction applied on top of art that already handled it.
THE HEADLESS BODY SHRANK. "Every headless mode shrinks the whole opponent a
little" — measured, 34.0 px healthy / 33.8 arm off / 27.8 headless on all EIGHT
families, a flat -18%, and the FEET shrank 16-19% with it. A body that loses its
head does not get smaller feet. drawZombie carried `if (st === 2) h = 1.52` to
take a head off the top, and a headless sprite is still a FULL-HEIGHT frame,
because the generator draws a gushing neck stump whose blood plume fills the
space the head occupied: 18.3% of z_rw2's frame sits above its shoulders against
10.7% above z_rw1's head. So the art had taken the head off, the per-frame PNG
scaling had handled the height, and the override took it off a third time.
Wrong on the fallback skins too (z_cw2 is a SHORTER png, which the scaling
handles): its body drew at 1.30 m against the healthy 1.78 m. Gone. The HP bar
and the aim mark read config.zombieH directly, so they had been floating above a
shrunken body for as long as it existed; they re-align for free.
THE WOUNDED BODIES WALKED BACKWARDS, and this is not bench-only — racketAI and
updateBench both set face as the sign of x movement, so a match did it too. The
racket is what measures it, Oliver's own test: "if this racket is facing to the
right I would assume that he also walks to the right." Its string bed keys out
to an enclosed hole (or is painted pale on the master), so the hoop locates
itself. The healthy stride is drawn racket-RIGHT (+0.38..+0.42) and every wound
stride racket-LEFT (-0.39..-0.45): THE WOUNDED BODY IS DRAWN FACING THE OTHER
WAY, on the master and therefore in the reference and therefore in all seven
families. One flip rule cannot serve both rows — fixing the healthy walk
inverted the wound walk, which is the second thing he saw. The flip is derived
from the frame name now, and only for the WALK frames (w1/w1b/w2/w2b), because
they are the only ones whose orientation tracks travel: the wound forehand
measures racket-right like the healthy one, and a serve or a fall puts the
racket overhead where it says nothing about direction. Checked against 18 frame
names including the tour-suffixed ones.
AND _famref.py PINS THOSE SIDES. That branch is only correct while the art stays
this way, and this is the THIRD art inconsistency this session that reached the
court because nothing measured it. A re-roll that mirrors a row now fails the
reference build pointing at drawZombie's artFacesRight, instead of shipping.
A DEAD WRITE, found on the way: racketAI set rz.facing, and zombies render off
zb.face. `facing` is the PLAYER's field and uses the OPPOSITE convention (a char
flips on facing < 0, a body on face > 0), so the match opponent's turn was never
driven by its chase at all — it fell through to the generic sticky-face line.
THE PRONE FRAME LAY ON ITS RACKET. "The player basically levitates over the
ground held by his racket." Every frame is bottom-aligned on the ground point,
and in the fall3 art the lowest ink is the racket hoop thrown out beside the
body — z_shw1fall3 is 249x85 with the body's underside at y67 and rows 68-84
racket, so the torso floated a fifth of the frame up. TUNE.proneSink drops it by
that fraction, putting the body line where fall2 leaves its knees. The shadow
made it worse and is suppressed for the whole beat: the contact ellipse is sized
off DRAWN WIDTH, and a prone frame is three times wider than a standing one, so
the widest darkest shadow in the game was painted under the one body not
standing on it.
THE FALL CARRIES THE RUN. "When the opponent runs to the ball from left to right
and then falls, the player must still move to the right while falling." The
velocity is INHERITED (racketAI and updateBench both record zb.vx), so a body
that went down standing still does not slide and one caught at a sprint slides
furthest. It was exponential decay first, which put 93% of the travel in the
first two frames — "the visible slide should also extend to the 3rd and 4th
frame" — and no retuning fixes that: stretch a wall-clock decay far enough to
move the prone body and it creeps for the whole floor hold. So the slide is a
DISTANCE BUDGET shared across the frames (TUNE.fallSlideMix), eased out within
each so it still decelerates. x is set absolutely from the start, so the total
cannot drift with frame rate. 0.05 / 0.22 / 0.35 / 0.11 m — the landing skids
and the rise drags itself the last of the way.
THE BENCH GOT A FALL BUTTON to tune all of this, and it plays the REAL beat:
same FALL table, same tickMatchFall, because a bench that re-implemented the
timing would be refining a second copy of it. matchReady() is skipped while the
bench owns the court (it runs setupPoint and re-places the body — a match doing
match things inside an art test), the walk stop stops walking during a fall, and
startFall reads the table LIVE via fallLen() so tuning the numbers cannot leave
the 'over' beat expiring mid-collapse.
FALL is 0.06 / 0.25 / 2.0 / 0.45 after a session in the bench. fall1 and fall2
are deliberately the same pace (0.87 and 0.88 m/s of slide at a 1.54 m/s entry)
so the collapse reads as one movement decelerating into the ground; the 0.07 s
fall2 it replaced gave each drawn frame ~4 display frames and read as a cut.
THE LENS, portrait camZ 48 -> 49. The three camera numbers are not independent:
holding the near player's drawn size fixed means camF = s_player * (camZ - zMax),
holding the court's on-screen depth fixed means camY = depth / (s_player -
camF/camZ), so 49 gives 795 and 11.23 and horizon re-derives to 43. Far opponent
33.1 -> 33.8 px, player unchanged, court band within a pixel. One metre is
deliberate: at camZ 55 the umpire is already clipped off the right edge and by 70
the court reads top-down, and it narrows the player's lateral room because
xVisible clamps him to what fits. A figScale experiment that scaled every body
1.2x got further (far body 40.6 px) but is NOT in here — it needs the vertical
contact constants to scale with it (ball-vs-body is gated on ball.y < 1.9, which
sits just above the tallest chest at 1.856 m, so the roller man would have gone
invulnerable) and Oliver reverted it pending a real playtest.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver walked the sprite bench and reported the same two faults on every family:
the healthy walk loses its head, and the wounded walk's feet do not move. Both
were in the ONE FILE every zombie tennis character is drawn from — and the family
pipeline had copied it faithfully, which is the pipeline working.
WHERE IT CAME FROM. That sheet was not composed from the sprites. It was a
GENERATED RE-DRAW of the club pro, made in nine separate 4K calls, one per row
band, then stacked (gen4k.py, in a session scratchpad, untracked and now gone).
Nine calls returned nine scales, and call 1 came back with the shuffle frame
decapitated. Measured against the sprite it was drawn from, that one figure
scaled 2.341 wide and 1.996 tall where all thirty others were proportional to
three decimals: 33 px missing off the top, width untouched. The local input to
that call was exact at 2.000x. The sheet lost the head, not the sprite.
THE FIX IS TO STOP PUTTING THE REFERENCE THROUGH THE MODEL. _famref.py composes
it from the 31 z_r* frames on disk — one scale, applied once, NEAREST — and costs
no tokens. That kills the head and the five scales together, and it explains two
things in _zsheet.py that read like fudges: the fall1 re-anchor and the walk-frame
levelling exist only to undo per-band scale drift. The levelling is also what
turned a headless z_sh2 into a headless AND 13% oversized one, by reading the
missing head as a height error and scaling up to compensate.
THE GATE WAS MISSING ON THE INPUT. _zsheet.py checks what comes BACK from the
model — band count, figure count, stacked bodies, edges, palette, heights — and
nothing had ever checked the sheet handed TO it. _famref.check() re-reads the file
it just wrote with _zsheet's own detection and refuses one where any figure's
width and height scales disagree by 2%, or where a stride pair is frozen.
THE THIRD DEFECT NEEDED NEW ART. z_rw1b had the same legs as z_rw1 (15.1% against
the healthy pair's 51.1%), so the club pro has slid through both wound stages
since the day he shipped, and every family inherited it. _stride.py draws the
missing twin, and what made it work after words had failed for as long as they
were tried is SHOWING the edit: image 2 is the healthy pair side by side, the same
character performing exactly the change being asked for. The headless twin then
took the ARM-OFF pair as its example — a wounded body is a closer picture of the
edit, the same chaining the wound ladder already uses. Its gate measures three
things and prints all three: legs moved, body above the hips did not, racket kept
its bed. The thresholds are CALIBRATED on pairs that already ship, not chosen —
my first guesses rejected z_e1/z_e2 and z_g1/z_g2, both accepted art. And a twin
FILLS the racket bed instead of keying it out: the first rolls came back with 959
px of enclosed transparency against their base's 0, which would have flickered the
racket face empty/filled once per step.
AFTER, on the installed sprites — leg difference within each stride pair, where
the wounded pairs used to sit between 4.8% and 19.7%:
r 51.1 / 68.1 / 36.7 ru 55.9 / 59.1 / 38.7
br 40.7 / 42.0 / 26.8 gk 41.1 / 44.6 / 30.6
sh 59.4 / 71.4 / 43.9 lj 38.4 / 45.5 / 27.4
bb 56.5 / 67.1 / 43.2 st 51.6 / 66.1 / 35.7
All seven families re-rolled off the rebuilt reference, every play row with its
head, and all 24 family/wound-stage states alternating a real pair — checked by
instrumenting drawImage on the bench, not by eye. br and lj sit lowest on the
headless pair and are worth a re-roll if they ever read as stiff; lj needed one,
its first roll came back frozen on all three pairs with the headless row 31 px
short.
AND THE REFERENCE IS TRACKED NOW. It lived in gitignored _refcheck/, so the thing
the entire pipeline rests on was unreproducible for as long as it existed. It is
assets/_pose_ref_family.png, next to _pose_ref_serve.png, and the scripts that
build and consume it are in the repo instead of a scratchpad.
Also in here: the SHAMBLE BOB, which was a 1 px lift on a wall clock applied to
every body every frame whether it was moving or not — a held pose bobbed forever
and a corpse bobbed while it lay on the ground. It only happens while walking now,
and it steps off ground covered, which is the same rule the strides follow.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
CLASSIC TENNIS, third time asked and this time actually found. Both cues open
with a mode gate and I checked the wrong two features twice: drawApproachRing
(the closing circles that show when to tap) and drawApexBand (the serve bar
with the gold perfect window) each began `if (!G.match ...)`. So the mode whose
whole pitch is "if the core tennis is good everything else is good" was the one
mode with neither. They run on `G.match || G.mode === 'tennis'` now — the gore
mode stays out on purpose, there is no rally there. A test reads both function
bodies and fails if either is ever gated on the match alone again.
(The two I kept pointing at — the landing ring and the placement ring — really
are shared. They were just not the ones being asked about.)
THE BOUNCING WAS IN THE ART, and measurable: the six generated families came
back with the shuffle frame 30-44 px shorter than the stance it walks out of,
against 9 px on the hand-made racket family. The game draws every frame at its
own PNG height, so the body rose and dropped once per step. A walk cycle is one
drawing with the legs moved and must be one height; _zsheet levels the stance
and the stride pair now and leaves the forehand its own, exactly as the serve
and the fall keep theirs. All seven families re-cut: spreads are 1-8 px.
THE BENCH GETS BUTTONS. Arrow keys are not a control scheme on a phone, which
is the device the walk has to be judged on — so every family and every wound
stage was unreachable from the place the problem shows. Three on-screen pads:
BODY, WOUND, POSE, the two-way ones split left/right. Geometry lives beside the
drawing and input hit-tests the same function, the rule the rack established.
THE FALL is quicker still (0.05/0.07/2.0/0.30): drop, lie there, get up.
Three umpire lines pulled — ump_wound_you, ump_felled, ump_wound_opp2. Not
funny enough to keep. Files deleted, call sites collapsed to the two that stay.
60 tests green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THE MENU BUG, and it explains three reports at once. cStart's handler was
`addEventListener('click', start)` — and addEventListener hands a listener the
MouseEvent as its first argument, so `start(modeId = pickedMode)` never saw its
default: modeId was an EVENT OBJECT. `G.classic = modeId === 'classic'` was
therefore always false and the zombie branch never fired. Picking CLASSIC or
ZOMBIE TENNIS on the setup screen did nothing at all; SERVE UP always started
the gore mode. A default parameter only applies to `undefined`, and a click
handler always passes something. `() => start()`.
That is also why "in normal tennis i dont see the ball circles and the serve
hint bar": there was no normal tennis to be in. Both are drawn unconditionally
in render() and the tennis struggle hints are already mode-gated correctly.
Verified through the real menu after the fix: CLASSIC gives mode=tennis with
no bodies on court, ZOMBIE gives match=true against the SHAMBLER in its own
body. Every other bare handler in the codebase takes no parameters — checked.
THE MOP BOY had the same class of bug one layer down. sendBoy was called and
mopWorthIt() refused: it tests `d.z >= COURT.netZ - 2`, the NEAR half, because
in the gore mode everything dies on the near court. A match's blood is all on
the FAR side, where the body stands and bleeds through four wound stages, so
the boy could never qualify. The near-side bias stays; the match is the
exception.
DANIEL was being dropped, not missing. Two causes, both mine: the asides were
queued at 0.7-1.5 s and matchPoint fires the score call at 0.4 s with a clip
over a second long — voice() drops anything that would start while he is still
talking (one man, one microphone). And asideEvery is 55 s, a whole match game.
Delays moved past the score call; matchAsideEvery is 15 s, which fits a mode
whose events are a limb and your blood rather than a wave of bodies.
THE FALL: quicker down, longer on the floor (0.09/0.11/2.0/0.30). And it was
genuinely SHRUNKEN — the generator draws fall rows smaller than standing rows
(the brute's buckle came back 190 px against a 224 px stance) and _zsheet's one
global scale preserved that faithfully, so the body shrank the moment it
started to fall. A fall row is anchored on its own buckle now, which must draw
at standing height. All seven generated families re-cut: 224/153/98/181 against
the hand-made racket family's 224/154/85/179 — the convergence is the check.
THE FLICK: 0.42 made going out impossible, which fixed the 8-in-10 bug and
overshot it ("too easy. i must be able to play out"). 0.62 puts risk back where
it belongs — ~15% out at quality 0, 5% at 0.75, and never on a perfect flick.
The test now asserts that shape: bounded, monotonically falling, zero at 1.
59 tests green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THE FLICK, again. "i can barely reach the extremes of the court." It could
reach 3.3 m against a 4.115 m line — safe, and too safe. TUNE.flickWidth is
3.95 now (96% of the way to the corner, 17 cm to spare) with flickReach 0.42,
which keeps the guarantee intact: reach x timing + scatter stays inside the
line at EVERY quality, worst case 4.06. 20% more reach, still impossible to
send out by aiming.
A FORWARD FLICK IS A SMASH. Up-screen is the far court, so flicking into it
asked for depth and nothing else — a direction with no consequence behind it.
It buys pace now (smashBoost, scaled by timing). Deliberately NOT distance:
solveShot still lands it on the same spot, so the shot arrives flatter and
sooner rather than longer. A smash you can hit out on a whim is just the old
bug wearing a hat.
THE FALL HOLDS. "the opponent should fall and then lie there for a quick
moment, the player stands up too quick." fall3 goes 0.30 -> 0.62. A corpse
that touches the ground and is instantly upright reads as a stumble.
THE CARD. Ten opponents, six of them NEW bodies off the one-sheet pipeline
(shambler, ball boy, steward, runner, groundskeeper, line judge), with the
club pro's own body and the brute woven in as 7 and 8. It is a ladder of
STAFF: the further you get, the further up the club the thing across the net
used to be. Narrative and difficulty are the same list read twice, and a test
walks all ten — whiff must fall every rung, names must fit the board's 8-char
slot, speed may zigzag by character but never regress.
Steeper per rung (speedPer .08->.11, pacePer .035->.05, flubKeep .82->.76):
"in general this is too easy now". With aimed returns no longer sailing out,
the player stopped handing over free points and the ladder had to earn them
back. The opening got SOFTER at the same time — 0.72 where the old first
opponent was 0.85.
THE UMPIRE WAS MUTED IN THIS MODE and nobody noticed. aside() refused during
any interlude; the match runs on a permanent one. So the single mode that is
literally a tennis match he is sitting over was the one mode he never spoke
in. Same bug shape as sendBoy, same fix. Six dry lines via say -v Daniel: two
on taking a limb, two on your blood, one when there is nearly none left, one
on felling the thing. asideEvery still gates him.
...and sendBoy had the identical bug, so the mop boy now comes out between
games to the FAR court, where one body has been bleeding through four wound
stages. Between games only, never mid-point.
The line judge took two rolls: the first cut to 6 figures in its last row and
_zsheet.py refused it rather than mis-slicing. That refusal is the feature.
59 tests green. Verified in the browser: SHAMBLER on court in its own body.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THE FLICK. "when returning a ball with a spin it seems too sensitive, 8/10
times the ball lands out." Arithmetic, not feel. The singles line is 4.115 m;
an aimed shot targeted a flat ax*3.3 and THEN added err() on top — up to
+-2.4 m at quality 0 — so a flicked return landed as wide as 5.7 m, and the
only clamp in the path sat at +-5, outside the court. An UNAIMED return
targets 0 + the same scatter and is therefore always in: aiming was the thing
losing the point, which is backwards for the mode's one act of skill.
TUNE.flickReach scales the aim width by the same quality the scatter shrinks
with, so the two can never sum past the line — worst case 1.49+2.4 = 3.89, and
a perfectly timed flick still gets the full 3.3 with nothing left to scatter.
The test asserts that sum across the whole quality range rather than sampling
rallies. Not named `aimReach`: that key already exists further down TUNE (the
slingshot's 22 m aim ray) and the duplicate silently won until a test caught it.
The placement ring carries `q` now and is the flick's receipt — tight and hot
when your timing earned the full width, wide and dim when you barely got a
racket on it. Without it the rule is invisible: the shot just goes less wide.
THE FALL (handoff item 4). A point that takes a limb knocks the body down and
it drags itself up short of it. The three collapse frames wear the stage it
STILL WAS and the rise wears the stage it has BECOME — that order is the whole
beat, and it is what turns a popping wound stage into something the point did.
Own clock beside the horde's warning, for the reason this file keeps relearning
(a beat that moves the phase under a latch strands it); the 'over' beat it sets
outlives the fall so it cannot expire into setupPoint() mid-collapse. Only
fires when the stage actually moved — a quarter off a full pool is 0.75, so
point one has no limb to explain and a second of collapse for nothing is a beat
the player waits out.
THE OPPONENT SERVES (handoff item 7), and it was mostly switched off rather
than missing. setupPoint already placed a non-player server behind its own
baseline and armed aiServeT; update() already tossed and struck for O. Both
were gated on `G.mode === 'tennis'` and the match runs in splatter under a
permanent interlude — one condition. Genuinely new: service alternates at each
game boundary (the match calls advancePoint directly and the flip lives in
tennis's pointTo, which it bypasses — the same shape as the G.serveNum bug),
the zombie takes O's mark so picture and ballistics agree on where it stands,
and the serve poses are named onto it because O is invisible here.
racketStage() moves to config: drawZombie and the fall must not answer the
same question twice, which is how the boss's skins and its blood drifted apart.
Docs: PROJECT.md gains THE ZOMBIE FAMILY PIPELINE — the measured API contract
(4K is a parameter and the default is 1K; n is 1 so "return both" is not a
thing; aspect is an enum so pad to it), why magenta stays despite green being
the documented key, and the two things _zsheet.py must do globally.
Verified in the browser: brute in the match, z_brfall2 -> z_brw1rise on a
wounding point, and bounce -> toss -> serve off the opponent's own baseline.
58 tests green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Nine sheets meant nine chances for the character to drift, and all day it took
every one of them — a brute that was three different brutes, a corpse that came
back a living man. Asking for the WHOLE family in one image removes that by
construction: one roll, one light, one palette, 31 figures.
What was missing was the way back down. _fine.py cuts a single row; a family
sheet is nine. _zsheet.py is that step and nothing else — it reuses _fine's
keyer, block detection and column cutter, and adds the two things that have to
be GLOBAL or the family breaks:
SCALE taken once from the play row's standing frame. Per-band scaling would
make the serve strike the same height as the idle stance, and the game
sizes a sprite off its own PNG height. Measured after: serve 319,
prone fall 83 — the racket family's are 307 and 85.
PALETTE taken once from the largest figure anywhere, so a wound row cannot
quantise onto slightly different colours than the row above it.
The enclosed magenta rule is INVERTED here, deliberately. _fine.py repaints
magenta trapped inside a figure because the generator used to fill string beds
with the backdrop colour by accident. This sheet asks for exactly that and gets
it, so the bed cuts to transparency and the racket is an open hoop.
_regen.py keeps three fixes found on the way: margins as a PROPORTION (an
absolute pixel margin means nothing to a model that picks its own resolution —
it clipped 3 rolls in 4), reference sheets that fill keyed-out string beds
instead of showing the model a magenta racket, and --verify, which asks the
question no gate here ever asked: is the art already on disk what it claims to
be. It found 27 of the brute's 31 frames dead — z_brw1fall was one drawing
pasted four times, 0.011 apart where a real sequence is 0.53.
The brute is live: ZFAMS gains 'br', the KEEPER wears it. Verified in #bench,
healthy and arm-off, 54 tests green.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Left over from the rename: the handoff still described opponent 2 as "rung N"
and pointed at matchRung, and three comments in gameplay.js still called the
opponent curve a ladder. One word per thing, everywhere now — OPPONENTS, WOUND
STAGES, a sprite FAMILY — with a note in the handoff saying which word replaced
what, so the next reader is not left translating.
(The barrel weapons keep their own "rungs of the same ladder" line: that one is
about the WEAPON payout order and was never part of the confusion.)
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A day went into sprite consistency and most of it was spent solving the problem
in the wrong place. Written down so the next session starts from the answer
instead of rediscovering it.
The boundary everything else follows:
the generator, GIVEN REFERENCE IMAGES -> identity, pose, composition
the pipeline, DETERMINISTICALLY -> exact colour
the gates -> layout accidents only
With the proof, because it is the part that is hard to believe until measured:
the brute's serve sheet came back with his build, his garments and his DARK
TONES pixel-identical to the reference and his big pale shirt two shades
brighter. The model was looking straight at the character. Every call
re-renders and re-rendering re-lights. No prompt fixes that.
And the rule underneath all of it — A REFERENCE BEATS A DESCRIPTION, at every
level. The racket zombie's sheet was made from the character as an IMAGE plus
the poses as IMAGES and landed first time; the brute's three sheets each
re-derived the character from the same paragraph and produced three different
brutes. Corollary worth having: every corrective sentence in a prompt is
compensating for a reference you did not supply. When a prompt grows a third
"do NOT" clause, add an image instead.
Also written down: correct-then-check, calibrate-before-you-gate (with the
numbers every threshold was calibrated on), the five gates and what each one
actually measures, six traps that each shipped something silently before they
were caught, and three things still open — including that `changed` passed a
headless brute that still had its head, because it knows something moved and
not what.
The brute is PARKED: out of ZFAMS, 31 frames on disk, nine batches in the JSON,
one line away from live. Its healthy eleven are good; its wound stages are not.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Re-running --variant to rewrite a spec re-seeded EVERY placeholder from the
master, and placeholders are copies of the lean red-shirted man the poses came
from. That silently replaced a finished brute's bounce, toss, strike and four
fall frames with racket-zombie art. I restored z_br1..4 by hand afterwards and
did not notice the other seven were gone — so three subsequent passes were
regenerating sheets that already existed and had been good since this morning.
Recovered from 0e92c72, and variant() now refuses to seed over a frame that is
already there: a placeholder is a stand-in for art that does not exist yet, and
nothing else.
Then colour-matched every brute frame onto z_br1 rather than generating: five
moved from failing the palette gate to passing, with no generator calls at all.
That is the point of the correction being arithmetic — it applies retroactively
to art already on disk.
WHERE THE BRUTE HONESTLY STANDS:
HEALTHY 11/11 — stance, shuffle pair, forehand, bounce, toss, strike, and
the full fall. Consistent, colour-matched, done.
ARM OFF the play and serve frames read; the four FALL frames are standing
figures of the wrong build entirely.
HEADLESS not headless — z_brw2 still has its head, and is near-identical to
the arm-off frame it was drawn from. The `changed` gate passed it
because SOMETHING moved; it does not know that the something has to
be the head.
So about a third of the brute is right and it is the third that matters least —
the healthy row was always the easy one. The wound stages are the job.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The character-as-image fix landed and did NOT close the gap, which is the answer
I promised to report rather than paper over. The brute's serve sheet came back
with his build, his garments and his DARK TONES PIXEL-IDENTICAL to the anchor —
(56,56,32) both — and his big pale shirt two shades brighter, (200,200,184)
against (176,176,152). The model was looking straight at the character. It held
identity perfectly and re-lit the largest flat area anyway, because every call
re-renders and re-rendering re-lights.
No prompt fixes that. Six gates spent a day trying to make a generator do
arithmetic.
So the arithmetic happens here now. Pair each of a frame's dominant colours with
its nearest anchor colour, fit one gain+offset per channel across those pairs,
apply it to every pixel. Pairs further apart than NEW_COLOUR are dropped from the
FIT — colours the anchor genuinely has not got, i.e. blood on a fresh wound — but
they still RECEIVE the transform, so a wound stays the right red relative to the
body it is on. The gain is clamped so one odd pair cannot posterise a frame.
PROVEN ON THE REJECTED FRAME BEFORE SPENDING A SINGLE CALL, which is the
discipline I skipped all day: the frame that measured 20.0 away passes after
correction, its shirt pulled from (200,200,184) to (176,176,160).
And the order is now CORRECT, THEN CHECK — the palette gate is a safety net for
what arithmetic could not fix, not a tripwire for what it can.
The model this leaves:
the generator, given references -> identity, pose, composition
the pipeline, deterministically -> exact colour
the gates -> layout accidents only
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The agreed single pass did not land: br_w1 installed, br_serve and br_fall were
refused by the palette gate at 12 to 22 away from z_br1, and everything drawn
from them was correctly skipped rather than built on sand.
'br' comes out of ZFAMS. The art stays on disk and the nine batches stay in the
JSON, so re-registering it later is one line — this is parked, not deleted, and
nothing in the game referenced it (no opponent row names it).
THE FINDING WORTH KEEPING, written at the top of _master.py: the brute's PLAY
sheet is a good brute; its serve and fall sheets are each a slightly DIFFERENT
brute. All three restyle from the master independently, so each invents the
character afresh from the words, and "pale grey-green work shirt" lands
somewhere new every time. The gate is not being fussy — it is measuring three
different zombies.
The fix is to hand the later sheets BOTH references: the master frame for the
POSE and the variant's own first sheet for the CHARACTER, instead of the words
for the character. Which is the lesson the wound stages already taught — a
reference beats a description — applied one level up. Until that lands, a
variant is reliable for its play sheet and a gamble for the other two.
The racket zombie is untouched and complete: 33 frames, every gate green, and
it is the opponent the match actually plays.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver, on the brute's fall frames: "now the fall zombie has a different skin
tone.. man this is all so tiresome." Both halves were right. The skin IS twenty
points lighter across that sheet, and the tiresome part is mine: I had been
adding one hand-picked gate per drift I discovered — redness, then hue, then
`changed`, then no-grids — which is a loop with no end, because naming features
is not the same as comparing pictures.
`kit` names four: shirt redness, shorts brightness, shorts hue, near-white
fraction. Every character breaks a different one. The brute's shirt has no red
in it at all, so its reference redness was 0 and blood scored 70-100 against a
tolerance meant for a garment. A body face down shows a fraction of the shirt it
shows standing, so two of the four move with POSE. And nothing in that list ever
measured SKIN, which is exactly how a brute's flesh drifted across a whole sheet
with six gates green.
The palette gate asks the only question that generalises: are the big colour
areas of this frame the same colours as the anchor's? Top SIX by area — the
garments and the skin. Proportions ignored, so a pose change is invisible; blood
is never in the top six, so a fresh wound cannot look like drift.
CALIBRATED BEFORE IT WAS WIRED IN, which is the discipline I skipped all day.
Frames known to be right score 0.0-8.5; the drifts we could actually see score
12.0-15.9. The threshold is 10, in the gap. Three formulations were measured and
the two that overlapped were thrown away.
And every sheet of a variant is now anchored to that variant's FIRST sheet, not
to the stage above it. Chaining let a legal hop compound into a drift nobody
authorised — the shorts walking -9 to -19 down the wound ladder was that, and so
was this.
Its first verdict on the existing brute: br_play is the character, and every
other sheet sits 15 to 48 away from it. It would have caught all of this on the
first roll.
Also in: the GRID guard from the same session — the valley-cut rescue would slice
a 2x2 sheet into four frames of a torso above someone else's boots, turning a
loud failure into a silent one. Asked per cut PIECE, because a body is vertically
continuous and a piece with a gap in it caught two of them.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver: "the talk about ladder and rung is extremely confusing." It was, and it
was my vocabulary, not his reading. "Ladder" had come to mean three unrelated
things in this codebase — the sequence of opponents in zombie tennis, the
sequence of WOUND STAGES on one body, and the name of a stats screen — and
"rung" meant a position in whichever of those you happened to be reading about.
One word per thing now:
the OPPONENTS who you play, in order, one game each (config.OPPONENTS,
matchFoe(n), zb.foe). Never "rungs".
WOUND STAGES healthy -> arm off -> headless, on one body. Unchanged, and
now the only thing allowed to be called a progression.
a FAMILY the sprite set a body wears (config.ZFAMS).
Renames: LADDER_TRAITS -> OPPONENTS, matchRung -> matchFoe, zb.rung -> zb.foe,
and the stats screen THE LADDER -> OPPONENTS FELLED, which is what it lists.
The 'L0'..'L9' record keys are deliberately NOT renamed: they are written into
saved run records, and renaming them would orphan every match anyone has
played. A comment says so where they are built.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
--variant produces 31 frames; nothing could draw them. drawZombie hardcoded the
`z_r` prefix, the preloader spelled all 31 filenames out by hand, and the sprite
bench built its names from the same literal. A new zombie would have sat on disk.
config.ZFAMS = ['r'] the families that exist
config.famFrames(f) the 31 names derived from a key
Everything that needs those names now derives them: the preloader spreads
ZFAMS.flatMap(famFrames), drawZombie reads `zb.fam` (unset = 'r', so the club
pro is untouched) and builds its wound lookups off it, and the bench walks
families on up/down. Adding an opponent is ONE ROW — which also retires 31
hand-typed filenames from the preload list, each of which was a chance to typo
one and have a body silently vanish for a frame. That is the oldest bug in
render.js and it is now impossible in this corner of it.
And the link that makes it worth anything: a LADDER RUNG can name a family.
`fam: 'br'` on a row in LADDER_TRAITS and that rung walks on as a different
body, all the way down its own wound ladder. So the whole path is now:
_master.py --variant br "…" -> generate 31 frames
ZFAMS += 'br' -> the game loads them
fam: 'br' on a ladder row -> that opponent is in the match
Verified: the bench still walks all 33 cells pulling the right derived name at
every wound stage, and a real match still opens on THE GROUNDSKEEPER wearing
family 'r' and drawing z_r1.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A variant used to stop at the healthy eleven, which is enough for a gore-mode
shambler and not enough for a MATCH opponent — that mode scores you on the
BODY, so an opponent with no wounded art would go from untouched to dead with
nothing in between.
<key>_play / _serve / _fall restyled off the master 11 frames
<key>_w1 / _serve_w1 / _fall_w1 arm off, drawn from healthy 10 frames
<key>_w2 / _serve_w2 / _fall_w2 headless, drawn from arm off 10 frames
Each stage is a restyle of THE STAGE ABOVE IT, in that order, because that
chaining is what finally made damage accumulate this morning: the missing arm
rides in the picture instead of having to win an argument with it. The helper
prints the batches in dependency order for exactly that reason.
Every gate the racket zombie's ladder needed is declared automatically — the
stride pair on both the healthy shuffle and each wound row, `changed` at 0.10
for an arm and 0.15 for a head (the generator will happily redraw the reference
and skip the amputation), and the pose-tolerant kit bounds with the shorts hue
kept tight, since hue is the measure that survives a pose change.
The wound rows carry three frames, not four: the shuffle doubles as the stance,
which is how drawZombie already reads them.
And the one-armed serve is generated, not hand-waved — the missing arm IS the
tossing arm, so the stump goes up and the racket hand flicks the ball off the
face, the way one-armed players actually serve.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver asked what the reference sheet is called. It did not have a name — it
was assembled per batch into /tmp and thrown away, so the pose set that took a
day to get right existed only as 33 loose PNGs.
assets/_sheet_master_play.png stance, shuffle A, shuffle B, forehand
assets/_sheet_master_serve.png bounce, toss, strike
assets/_sheet_master_fall.png buckle, knees, down, rise
Magenta-backed, spaced, feet on one ground line — loadable straight into a
generator as-is. `python3 assets/_master.py` rebuilds them from the installed
frames, so the master can never drift from what the game actually draws.
THREE SHEETS AND NOT ONE, for a measured reason: the generator answers at a
fixed width around 1400-1800 px. Four figures across gives each 350-440 px,
which carries the 90-110 px of figure height the style block asks for. All
eleven on one strip would be ~130 px each and the detail gate would reject
every roll. Four, four and three is what fits.
And the thing the master is FOR:
python3 assets/_master.py --variant br "an ENORMOUS FAT ZOMBIE BRUTE: …"
python3 assets/_regen.py br_play br_serve br_fall --tries=4
which writes three restyle batches, seeds their placeholders and declares the
stride pair so the gate covers the new body from its first roll. A variant is
three generator calls and NO POSE DESIGN AT ALL — the reference specifies
eleven poses that already work and the prompt only says who to draw. Measured
on the brute test: the restyled stride pair came back at upper 0.148 / legs
0.618 against the master's own 0.157 / 0.565, very slightly better than what it
copied.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The serve and the fall now exist at every wound stage, so a one-armed opponent
still serves and a headless one still goes down, instead of reverting to a
whole body to do it. 33 frames, no gaps.
THE BALL IS OUT OF THE SERVE SPRITES. Oliver: it was painted into three frames
and the game draws the ball itself, so every one of them would have ridden
along at the wrong place, the wrong time and in the wrong number. Hands open
and empty now, poses unchanged as if a ball were there.
THE FOURTH GATE, and the one nothing else could see: "copy this exactly, change
ONE thing" loses when the change is a REMOVAL. Taking a head off worked every
time — it is dramatic. Taking an ARM off mostly did not: five of seven wound
frames came back as the reference faithfully redrawn, 0.02-0.05 of their own
silhouette different, and the ladder quietly HEALED the body the score had just
maimed. Frame count, background, clipping, detail, kit and stride all pass that
sheet. A batch now declares how far it must move from the sheet it was drawn
from — a copy scores 0.02-0.05, a real amputation 0.12-0.38, a decapitation
0.27-0.75 — and the whole set now measures 0.18-0.65.
AND THE BACKGROUND RULE MOVED TO THE FRONT OF EVERY PROMPT. Six rejections in
eleven rolls were "background is not magenta", and the generator was not
glitching: it was painting the zombies ON A TENNIS COURT. The chained restyle
subjects are long, and the magenta instruction sat behind three hundred words
of "keep everything exactly as the reference", where it stopped being honoured.
Same sentence, said before anything can bury it.
One design consequence kept rather than papered over: the missing arm IS the
tossing arm, so a wounded opponent tosses off the racket face the way one-armed
players actually do. The wound changes how he plays, which is what slice 3 has
been asking for.
Known and NOT fixed: the shorts hue drifts from -9 to -19 down the ladder. Each
stage is measured against the one before it, so a legal 4-point hop three times
over is a 14-point total nobody authorised. The fix is to anchor every stage to
the healthy frame with a total budget instead of a per-hop one.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THE MISSING ART, generated by the whole-sheet restyle method: bouncing the ball
at the baseline, the toss with the ball leaving the fingertips, the strike at
full stretch — and the fall, which slice 4 has been waiting for: buckling, down
on the knees, flat on the court face down, then dragging itself back up. One
zombie is now complete as a tennis player rather than as a body that shuffles.
(Healthy only so far; the wound variants are the next sheet.)
THE SPRITE BENCH, in the back room. No ball, no scoring, no AI — one opponent
on an empty court, walking, at the size and speed it is actually played at,
because that is the only place a walk cycle can be judged. A sheet that looks
fine at 224 px on a contact sheet is a different animal at 30 px crossing a
lawn, and every sprite argument this project has had was really about that gap.
It steps through every state including the ones the game cannot reach yet, so
the serve and the fall could be looked at the day they were drawn instead of
the day they get wired up. It prints the FILENAME it is drawing and the hp the
wound stage was read from — the two numbers that separate "the art is wrong"
from "the wiring picked the wrong art" — and it says HEALTHY ONLY where a wound
variant does not exist rather than implying one does. The stride steps off
ground covered, like the real thing: a bench animating on a clock would flatter
exactly the sprites that foot-slide in play.
It owns the court like every other mode and every start* refuses while it runs.
FOUND BY IT, ten seconds in: `start()` applies the mode the setup screen picked,
so a remembered ZOMBIE TENNIS was taking the court before the back room could —
every back-room entry (weapons, bosses, drill, match, bench) had been silently
refusing since the mode select landed this morning. The back room passes null.
And a third calibration lesson: POSE MOVES A GLOBAL COLOUR STATISTIC. A body
face down shows a fraction of the shirt it shows standing, so `redness` fell 14
points across the fall sheet without a colour changing. The measure that
survives a pose change is the shorts HUE — mid-tone neutrals, present in every
frame in the same proportion — which is why it is the one that caught the drift
Oliver actually saw.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The arm kept coming back at the headless stage no matter how the prose was
worded, and the reason was the picture: stage 2's character reference was the
OLD two-armed headless drawing, so the model was faithfully copying two arms
while the words asked for one. Prose does not win an argument with a reference.
Stage 2 now takes stage 1's ENTIRE THREE-FRAME SHEET as its reference and the
job becomes "redraw this sheet with ONE thing changed — the head comes off".
The missing arm rides in the picture. Both tours installed FIRST TRY, where the
previous wording burned twelve rolls and never once got there.
The ladder finally reads: whole -> left arm gone -> headless AND still one-armed,
same body, same charcoal, same shoes, all the way down. In a mode where the
score IS the body that is the whole point, and it has never been true until now.
--audit stays clean: all six stride pairs pass.
This is Oliver's method and it is better than what it replaced. What makes it
work is that each call TRANSFORMS a picture instead of describing one, so the
reference does the specifying and the prose only says what changes. Fewer calls
is the smaller half of it; the consistency is the point.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver's idea, tested: give the generator a WHOLE POSE SHEET and ask it to
redraw it as a different zombie. It works, and better than what I was doing —
the master set's stride pair is 0.99/0.157/0.565 and the restyled brute came
back 1.00/0.148/0.618. The pose relationship SURVIVED the character swap, in
one call, four frames.
It was reported as a failure only because the four brutes stand shoulder to
shoulder and their bellies overlap, so segmentation by empty columns saw ONE
blob and the sheet was discarded. That same rejection ate five rolls earlier
today — "found 2 frames, expected 3" — and the sheet it lost last was the best
roll of the day.
A row is evenly spaced by construction, because that is what we ask for. So
when the blob count comes up short of a count we ALREADY ASKED FOR, look near
each even division and cut at the thinnest column in that window: between two
figures the ink thins even where it does not vanish. Never used to guess how
many figures there are, only to reach the number requested — a wrong guess
would install garbage silently, and the frame count is the one thing this
pipeline has always been able to trust.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`_regen.py --audit` is clean for the first time:
racket_play z_r2/z_r3 1.00 0.157 0.565 ok
racket_w1 z_rw1/z_rw1b 1.00 0.112 0.424 ok
racket_w2 z_rw2/z_rw2b 1.00 0.077 0.372 ok
racket_f_play z_r2_f/z_r3_f 1.00 0.049 0.552 ok
racket_f_w1 z_rw1_f/…b_f 1.01 0.112 0.716 ok
racket_f_w2 z_rw2_f/…b_f 1.00 0.054 0.793 ok
Against the reference pairs (brute 1.00/0.000/0.638, shambler 1.00/0.070/0.520)
these are the real thing: one body, same camera, legs swapped. Fifteen rolls
were rejected getting here, every one for a named measurable reason — which is
the whole point of the gate: none of it shipped on luck.
Two calibration lessons, both from watching what the gate rejected:
BRIGHT is off for the racket batches. It counts near-white pixels and was
calibrated on the LIVING opponent, whose racket barely registers — these frames
hold a pale-strung racket FACE-ON and it swings 3%..20% purely with how the
racket is angled. It was rejecting rolls for racket geometry. Redness, shorts
brightness and shorts hue are the kit, and hue is the one that caught the drift
Oliver actually saw.
EACH WOUND STAGE IS HELD TO THE STAGE BEFORE IT, not to the healthy frame two
steps away. Stage 2 kept being rejected at hue -4.2 against the healthy -8.7
while the stage it actually follows sits at -5.0 — the roll was more consistent
with its own neighbour than the gate would allow. Chained references stop drift
compounding without demanding stage 2 match a frame the player never sees next
to it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver: "look at the headless men — the trouser color changed." He was right and
every number in kit() said he was not. Brightness sat at 85 against the healthy
frame's 89, comfortably inside tolerance, while the shorts' blue-minus-red
walked from a warm charcoal -8.7 to a cold +0.2 across the wound stages. All
three existing measures are greyscale or red-channel; none of them can see a
grey going blue, which is most of what "the colour changed" ever means here.
Fourth number: the shorts' hue. It immediately fails the headless men AND finds
a second drift in the wounded women that the brightness test had passed.
The tolerance default went to four entries deliberately — kit_off zips got/ref/
tol, so a three-long tuple would have SKIPPED the new check rather than failed
it. That is the same shape of silent no-op as the read-only floodfill.
Also: DAMAGE NOW ACCUMULATES. Stage 2 said only "headless", so the arm torn off
at stage 1 came back — in a mode where the score IS the body, taking a point off
it made the body look healthier. Stage 2 is now headless AND still one-armed.
And the stride rule got its counterweight: "copy it exactly, only the legs
change" has a second failure mode where the model copies it exactly INCLUDING
the legs and returns the same picture twice (upper 0.017, legs 0.022 — a body
sliding along with frozen feet, the very thing being fixed). Both halves are
stated now: identical above the hips, obviously different below.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver, on the sprites, again: "when people on a tennis court walk or run their
legs move... this pattern is almost always failing." It has failed repeatedly
because NOTHING IN THE PIPELINE MEASURED IT. Frame count, magenta background,
clipped edges and detail density all pass cleanly on two unrelated
illustrations of the same character, which is exactly what the generator
produces when asked for "the same frame with the other leg forward".
Three numbers per declared pair, silhouette only, on the feet-aligned centred
overlay drawStatic actually draws:
HGT height ratio — off 1.00 means the body grows and shrinks once per step
TOP difference ABOVE the hip — near zero for a real pair: same man, same camera
LEG difference BELOW it — this is the step
Calibrated on the pairs that read (brute wound lurch 1.00/0.000/0.638, shambler
1.00/0.070/0.520) against the ones that do not (TOP 0.47-0.85, or LEG 0.09).
Opt-in per batch via `stride: [[i, j]]`, because the player's four-frame RUN
sits at TOP 0.23 and is right — a run swings the arms too.
`--audit` runs the same gate over the sprites ALREADY INSTALLED and exits
non-zero on a failure: a gate that only fires on new rolls says nothing about
the art already in the game, which is where this was noticed. Five of the six
racket pairs fail it today.
And the kit gate, which already existed and was never switched on here: Oliver
spotted the headless men's trousers had changed colour. Measured — the headless
set's near-white fraction DOUBLES (9.5 -> 19.5) where losing a head only
accounts for ~1.3 of that, and the wounded swing drifted the other way (shorts
60.5 against its own stage's 85). palette_ref + a per-batch tolerance now catch
all three drifted sets and pass the one that looked right.
Fixed while in there, mine from the last commit: the wounded stride stepped off
walkT, a wall clock, while the healthy one steps off ground covered — the exact
foot-slide this project banned for the heavy bodies.
No art regenerated yet: the gate and the audit come first, so a re-roll can be
told from a lucky one.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Slices 1, 2 and 5 struck through with what they actually became; 3, 4, 6 and 7
left standing, with a note on 3 that the wound-stage art now exists for it and
on 4 that the racket_* batches are its template. The parts of slice 2 that are
still open (shots[] and the PARRY — the cheapest depth in the mode) are called
out rather than buried under a DONE.
Added: the record shape table (which of g/s/dead each mode uses and why), the
classic-tennis mode, the second ending, and four new traps — the stats banked
flag that no-kill modes never clear, the magenta key that ate racket strings,
the read-only PIL image that made the fix look like a no-op, and the detail
bar that ratchets against what it is replacing.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
A second plate, drawn rather than tinted (assets/_board.py — _regen.py cannot
do this job: it keys magenta, crops to ink and rescales figures to a common
height, and a scoreboard is a full-bleed rectangle with no figure in it). The
club's bottle-green board is the club's; the match is played on one nobody has
cleaned in a long time, and the ink changes with it — a green 'dim' over dried
blood reads as a printing fault.
Two column changes go with it: the name slot says KEEPER / JUDGE / CHAMPION
instead of THEM (eight characters, `short` on the ladder rows), and the SETS
column is gone in the match, where it was a permanent 0 — the bar under the
board already carries that mode's progress.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THE SPRITES. The match opponent borrowed the turned CPU's art — a body drawn
to SHAMBLE at the camera with its racket hanging at its side, which is what
was playing tennis points against you. It has its own family now, 'r', drawn
off the LIVING opponent's poses: a ready stance, a shuffle pair and a real
forehand, at every wound stage, men's and women's — 20 frames, six batches.
'c' keeps its shamble untouched, because in the gore mode shambling is exactly
what the turned CPU should be doing. The pose comes from racketAI's own
`pose`, so the picture cannot drift from the behaviour, and a one-armed
opponent still plays tennis instead of reverting to a walk.
Two pipeline bugs found doing it, both fixed at the root in assets/:
- _fine.py keyed EVERY magenta pixel, not just the background, and the
generator paints racket strings in the backdrop colour more often than not —
so frame after frame installed a man holding an empty hoop. Background is now
the region connected to the border; magenta left INSIDE a body is repainted
pale grey rather than punched out.
- the flood fill silently did nothing, because Image.fromarray images are
read-only and PIL's floodfill swallows the write error. .copy() plus a real
error if it ever fills zero pixels.
THE LADDER (handoff slice 2). Rung N takes design N's name off the chapter
roster — the club's dead staff play you at tennis here and come for you in
gore, same joke twice — plus a personality that is DIALS, not new mechanics:
pace, width, flub and speed scalings multiplying numbers racketAI already
reads. A rung is a data row. Name plate on the walk-on, and the kill files
under the rung rather than under the generic body.
THE STATS SCREEN follows the mode split: BODY COUNT / THE LADDER / THE RECORD
behind a mode strip, each with its own three figures, its own roster heading
and its own footer. The weapon shelf and the roster hide in the modes that
have neither, rather than printing a line about something the mode cannot do.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The setup screen picks WHAT before WHO and WHERE. Three rows out of a table,
in the court row's own markup:
CLASSIC TENNIS is not the opener with the turn switched off and left there —
it is the mode the other two are built on, so it gets real match scoring: six
games with a two-game lead is a set, best of TUNE.classicSets is the match,
the board shows the sets column (hidden until now outside splatter) and the
bar under it says which set this is. The CPU stops rubber-banding and flubs
less: aiFlubLoss exists to get a new player PAST the opener, and a mode whose
whole promise is a real match cannot be won by an opponent that falls apart
the moment he goes ahead. No tiebreak — 6-6 plays on to 8-6. Marked ponytail.
That needed a second ENDING, the one with no corpse in it: G.done, the same
summary and gated tap as the death screen, its own words (GAME, SET, MATCH. /
BEATEN.) and the man still standing. The three endings now share one block
rather than three copies of it.
Caught by the first classic test: stats banked only the FIRST classic run of a
session. `banked` was cleared by the next KILL, and a classic match has no
kills in it at all — every later match was swallowed by the double-bank guard.
beginRun() is the honest reset and restartRun calls it.
ZOMBIE TENNIS from the menu is one line; the back room keeps its door.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The two modes share no currency: gore counts kills, multi-kills and weapons;
zombie tennis counts opponents felled and has exactly one kill per opponent,
always credited to 'bomb' (matchGameWon's explode). Every one of those was
landing in splatterRuns as a gore run, on the roster as THE CLUB PRO, and on
the KILLS world board — where a five-body match would post a row next to a
300-kill massacre and mean something else entirely by it.
One character on the record fixes all of it: `m`, absent meaning gore (which
is what every existing record is, so no migration). read() takes a mode and
so does endRun, which now compares you only against your own mode and refuses
a name prompt from a match.
The record's two headline slots carry each MODE's two headline numbers — gore
g=games s=sets, match g=felled s=games — so `g` stays the thing everything
ranks on and the match best is derived from the record instead of living in a
second key that drifts. That retires the matchBest key from the last commit.
Found on the way, pre-existing and visible: endRun returned `{...best, kills,
games}` and the later key won, so G.best.games was the run that had just died.
The gore death screen has been printing "BEST <this run> GAMES · <the record>
SETS". The return now carries the best and nothing else.
And stats.check() — the module's own fixture check — was exported and called
by nothing. Wired into the suite, with the mode split added to it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three things, all in service of the match reading as its own mode:
BLOOD, COUNTED. The vignette says "worse" and never says "how many". One pip
per hit left, top-right beside the board, in the panel language the game/sets
bar already uses; the last two breathe. Pips rather than a bar on purpose —
the pool is 4 or 8, and a smooth readout of a small integer is the exact
ambiguity being fixed.
OPPONENTS FELLED IS THE MEASURE (handoff slice 1). Blood-only scoring means a
lost game costs nothing but the blood already spent, so games survived is not
the achievement. The in-play bar prints OPPONENT n · m FELLED instead of the
game/sets line, the death screen says BLED OUT / FELLED n OPPONENTS · m GAMES,
and it keeps its own matchBest — drillBest's pattern. The kills-to-next-point
pips are hidden in a match: nothing there is killed for points, and they were
colliding with the new line anyway.
STAGING WEARS A DIFFERENT ICON. Both builds live on the same phone and two
identical bloody tennis balls is a coin toss. Driven by HOSTNAME, not by the
branch — the block is identical on staging and main, so it never conflicts on
a merge and never has to be remembered.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Temporary file, written because a session ran out of context mid-slice — fold
into PROJECT.md and delete when the work lands. Not in the Dockerfile COPY;
docs are not served.
Records the decision (A: blood-only, measure becomes opponents felled) and
the finding behind it: Oliver believed losing a match killed him, when he was
bleeding out during deuce wars. That misreading is itself the bug — model A
only works if the player can see they are bleeding, not losing.
Also carries the build order (summary+measure, named ladder with
personalities, wound-driven play, the fall animation, menu, flick teaching,
opponent serve games), the flight-recorder recipe that ended four rounds of
guesswork, the traps this mode already sprung, and the do-nots.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Decided with Oliver 2026-08-12. Records why tennis is the friendly case (the
sync unit is the shot, alternation already enforced by ball.lastHit, each
client authoritative for its own contacts), what the codebase already brings
(deterministic sim, fixed dt, beats.js ownership contract), the zero-dep
WebSocket path (~120 lines of stdlib RFC6455 in server.py, friend-code
rooms), the honest cost (reconnect-and-resume is the real work), and that
step one is the GHOST MATCH over plain HTTP — half the netcode falls out of
it as a byproduct. Plus the hook: score-is-the-body on both players, losing
is zombification.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Fourth recorded session, verdict "better": he beat opponent one — the first
game won in any recorded session — and reached opponent two, which took him
apart with gaps of 1.66-2.13m (60Hz, accurate) while he swung at closest
approach every time. The gradient is real; the step was a wall, because a
rung moves three dials at once. Per-rung increments softened: matchSpeedPer
.12->.08, matchWidePer .4->.25, matchOppPacePer .05->.035. Opponent one
untouched — the bases stay, only the slope flattens.
The contact chain is now verified end to end by recorded play: legs
(matchAssist) -> destination (assistK in intercept + the anyHit short-ball
chase) -> reach (matchReach). Each link was a separate defect found by a
separate recorded session.
49 tests, seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Third recorded session. The previous fixes did their job and exposed the last
gap: he ran onto every ball (2.9-3.9m per point, was 0.2-0.5), the ball
reached him (died at z 22.5-24, was 18-20), he swung every single time — and
the closest approach clustered at a sampled 1.3-1.6m, centimetres outside the
1.3m post-bounce reach, point after point. The assist delivers you NEAR the
ball but not inside a phone-tuned contact radius, and that also explains "not
really better or I played worse": every point suddenly looked makeable and
still died.
matchReach 1.35 widens the contact test in the match only — the gore modes'
"tight contact radius: the tap must be honest" rule stands untouched. A rally
mode lives on contact feeling fair.
Recorded for future sessions: a 20Hz sampler overstates closest approach by
up to ~0.5m on a 20 m/s ball. The recorder samples at 60Hz now.
49 tests, 12 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Second recorded session: 5 wins and a deuce war to 5-6 — the mode is a
contest — but all seven losses shared one signature: the ball died in court
1.4-2m short of him while he ran 0.2-0.5m. Three defects:
The pace cut created dying rollers. Slow returns bounce and die at z 18-20,
outside intercept's near zone, so nothing chased them, and the rolled-dead
rule hands the point to the zombie. They also arrive at his feet below legal
contact height — the "ball arrives right at the player" feel. intercept keeps
an anyHit candidate now: the earliest reachable point on his half where the
ball is still at hitting height, so the assist runs forward onto the short
ball. Verified: he sprints z 23 -> 16.4 and meets it at gap ~0.
intercept computed reachability with autoAssist — 0 on any fine pointer — so
a desktop match's planner answered "home" every time. He had legs, his
destination was always the baseline. assistK() is shared by mover and
planner now.
And the vignette was a radial gradient in a portrait canvas: the circle
reaches the corners but hits the side edges at fraction zero — "red pulsing
from below" was the bottom edge being the only place it got close. Four
linear edge fades now, hugging the rectangle on any aspect.
48 tests, 12 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"can you watch me play in the browser pane and see why I loose". A passive
20Hz recorder sampled his real play: seven losses, and in every single one he
SWUNG (his timing was fine) while the ball died 2-3m to his side and his own
x never moved all game. He had no legs: autoAssist keys off pointer:coarse,
the pane reports a fine pointer, so auto-run was 0 — the exact trap the
verification recipe has warned about for weeks, bitten in live play.
matchAssist 1.15: the match brings its own legs on every device. It is a
tap-timing game; footwork is the game's job. Held keys still override. Pinned
by a test that plays with autoAssist 0 and asserts he runs.
The vignette curve is sqrt(damage): the match pool is 8, so hit one was a 10%
alpha nobody can see. sqrt front-loads it — first blood reads at ~0.3.
This also explains three rounds of disagreement between his reports and my
measurements: my autopilot always had legs (it set autoAssist by hand), his
pane play never did.
47 tests, 12 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"there are still balls which are basically not reachable with auto run..
please recheck" — and, asked twice: "did you write tests for the last issues
we encountered?" The straight answer: no. The unreachable-balls issue only
ever had config assertions, which is why it came back twice.
Pace is the ladder's fourth dial. Width alone never fixed reachability
because the returns left at ~27 m/s, crossing to the baseline in ~0.6s — less
than auto-run needs from the far corner. matchOppPace 0.78 puts opponent one
at ~20 m/s (measured live: twelve returns, 20.0-20.7), a quarter more flight
time, and matchOppPacePer hands it back per rung.
The vignette deepens: dmg x 0.8, cap 0.85, and the clean centre shrinks with
damage instead of only the rim darkening.
And the reachability regression the issue always deserved: a headless 90s
autopilot soak asserting at least 10 returns sampled and at most 15% that
never came within volley reach. Sabotage-verified. Its honest limit is
recorded in PROJECT.md: the autopilot attends perfectly and taps
frame-precisely, so it out-covers a human — it is a floor against
catastrophic regressions, not a proxy for the thumb. Oliver's playtest stays
the real gate.
46 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three asks. The placement ring: an aimed rally hit stamps a gold ring that
fades where the shot is going — flick, watch it bite the corner, learn the
cause. Not stamped for unaimed hits (marking scatter would teach noise), gone
in 0.8s (a permanent ring reads as a target rather than feedback).
The opponent's width is the ladder's third dial. Its returns scatter off
aimError x (1 - quality), which painted lines from game one. matchWideBase
2.1 keeps opponent one inside what an auto-run player can cover — measured,
25 returns all inside +-2.2m — and matchWidePer earns 0.4m per rung, so
line-painting is the late ladder. Speed, reliability, width: three dials.
The blood vignette replaces the temporary border. The old wash flashed 0.3s
and vanished, so between hits the screen said you were fine. A radial
vignette reads G.hp directly now — its depth is your damage fraction,
permanent until healed, hit pulse riding on top. Global: the hp channel is
global. Same no-HP-bar philosophy one layer out; at 1 hp the frame is
properly blood-rimmed and the court centre stays legible.
44 tests, 12 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"every second ball is basically unreachable by an auto-run player.. are there
some hidden gestures I don't know about or is it just a very strong cpu?"
Neither. The match inherited gore-splatter's depth-only recovery — the
sidestep-protection rule — by being G.mode 'splatter'. One wide return pulled
you off centre and you stayed there, so the next ball to the opposite corner
was lost before it was struck. A player who never walks back experiences
every rally as unreachable, and it reads as CPU strength because the balls
keep landing where he isn't.
The match recovers to centre now, like tennis: real footwork, and in a rally
there is no artillery lane to protect. Gore splatter keeps depth-only — the
sidestep it protects has no role inside a rally. matchSpeedBase 1.4 -> 1.25:
it was calibrated against a player who couldn't move.
The full attacking vocabulary is recorded in PROJECT.md — nothing else is
hidden. The placement flick (aim during the swing) is taught nowhere on
touch; flagged as a ghost-hand lesson if the mode ships.
42 tests, 12 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"still when I lost a single point and try to serve again every further tap
does hurt me more until I am dead". Reproduced tap-for-tap this time, and the
serveNum fix was a red herring — a real bug, but not this one.
tossBall never stamped ball.lastHit. After any rally the zombie touched last,
the freshly tossed ball still said lastHit 'O' — and the previous pass moved
the server behind the baseline, past the passing-shot line. So the moment the
toss went up, the game saw an opponent ball beyond your baseline: "its
passing shot ends the duel", point to them, blood. Every tap = a toss = an
instant lost point, faster the more he panicked. Two fixes composed into a
death loop; the stale field was harmless while serving happened inside the
court.
One stamp in tossBall — ball.lastHit = sv.side — fixes every caller. The duel
and the gore mode carried the same latent bug for anyone tossing deep behind
the line after a racket-zombie touch.
Verified with the exact field shape: lose a rally the zombie touched last,
then eight badly-timed panic taps — one honestly lost point, zero
toss-bleeds. Before the fix every single toss bled.
41 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"still very weak and slow opponent. it also spawns at places where it is
impossible to react to my serve. so i win like always. in the cases i dont
win there is a bug. when trying to serve after i got hurt i get hurt again
and again until i die".
The death loop was G.serveNum. Tennis's pointTo is the only thing that ever
reset it, and the match bypasses pointTo entirely — so the first fault of the
match armed "second serve" permanently, and every missed serve after it was a
double fault, a point, and blood, repeating to death. Exactly the spiral a
rattled player serves into. matchReady(), the match's own point boundary,
resets it now; a regression pins it.
The receiver starts every point on its mark: matchReady drops it onto the
exact receiver position the tennis branch computes for O, deuce/ad mirrored,
instead of a random spawn x or wherever the last rally left it. An opponent
that cannot reach the serve is not weak, it is absent.
And the base opponent is a player now: matchSpeedBase 1.4 (the duel's 2.8m/s
lumber was tuned for one sudden-death point), matchFlub 0.08 (0.2 gives away
a fifth of all returns). Measured with the same autopilot that previously won
seven straight games: it now loses the match. Possibly too hot — the two
dials are named, Oliver's thumb decides.
40 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"the player is too weak in this mode. also the killing by smashing occurs too
often. i cant really control it. maybe obsolete? also the serving after the
first serve does not happen from behind the groundline".
matchPlayerHp 8 — the match brings its own pool. The gore mode's 4 was an
inheritance, not a decision, and a real game has rallies you deserve to lose
without the run ending.
Regicide is out of the match (kept in the duel, where any-weapon-counts is
the point of sudden death). In a full game every hard serve that happened to
meet the body was a free kill nobody aimed. A match ball never damages the
opponent — it passes through like a soft ball and racketAI plays it — so the
wound stages are exactly the scoreboard. That also made hp-per-opponent
meaningless, so the ladder escalates in racketAI instead: +12% speed per body
felled, whiff chance x0.82 per rung.
And serving resets to behind the baseline with real deuce/ad mirroring:
setupPoint's "serve from wherever you're standing" branch now excludes the
match, which falls through to the tennis positioning. Verified live: second
serve from z 24.3, x mirrored to the ad side, opponent unwounded by anything
but the score.
38 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Triage failed the only test that matters — "i think the normal player does
not grasp this" — and the lesson across four drill designs is that every
invented rule needs teaching, while the rules the scene already implies need
none. So the new mode is the game's own oldest rule: tennis.
MATCH TEST in the back room (#match). One game per opponent, real 15/30/40
scoring. Built on top of the duel machinery: G.match runs a permanent
interlude, which is already what makes serving tennis-like, hides the rack,
quiets the spawn director and wakes racketAI. The one change is in
resolveInterlude — a settled point feeds matchPoint instead of ending
anything. About 90 lines for a whole mode, because the deepest system in the
repo was already built.
The score is the body: a quarter of its pool per point won, floored at 1, so
a love game walks the racket zombie through every wound stage and the game
point fells it. Its points draw your blood — hp--, eaten at zero, matchHeal 2
per game won is how you last. Regicide is inherited: kill the body mid-game
and the game banks by hand.
The timing is drawn, per Oliver's ask: an approach ring around the incoming
ball that kisses a fixed inner ring exactly at the sweet contact distance
(gold inside the band), and an apex band beside the toss spanning the same
0.12m keyPressed scores as perfect, with a tick riding the live ball. Both
cues use the mechanics' own numbers, so the picture cannot lie.
Verified by autopilot: 7 games in 90 simulated seconds, opponents 1..8, pools
4/6/8/10, wounds tracking points exactly.
36 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"the marker flips all the time even during a ball being midair.. i am not
sure that a player would get the mechanic intuitively let alone without the
marker".
The flip had a specific cause and it was a design bug, not jitter. With lanes
held and speeds constant, arrival differences never change while bodies walk
— the queue only reorders on a kill or an arrival. The marker flipped because
it showed only q[0] while the rule accepts any near-tie within the slack:
every kill made it jump between tied bodies, so the marker was stricter than
the rule it was explaining. drillLegal() marks the whole slack set now,
bright head and dim ties, which is stable by construction. Measured: zero
head-changes across 12s of walking with no kills.
Learnability, two answers. The rule gets one modal, once per run, on the
mode's own calm court — CLEARANCE DRILL / DROP WHOEVER WOULD REACH YOU FIRST
/ FAST BEATS CLOSE / STREAKS PAY — because "most urgent first" is not
guessable from a chevron alone. And the markers are training wheels that take
themselves off: hidden at streak >= 8 (the streak is the evidence you can
read the court), back on any fault. The teacher steps in exactly when you
demonstrably need it.
31 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The sequence drill tested nothing — the strip was always visible (no memory),
the batch was slow (no urgency), the aim was easy (no execution). A checklist
with gore, and Oliver said so: "not interesting enough to keep me playing".
Settled in discussion, he picked triage + streak. One rule: always kill the
body that would reach you first. Nothing is called, nothing shown — the court
is the puzzle. arrival = distance / speed down a held lane, and the roster's
speed spread means the answer is not the closest body: a runner at the far
baseline beats a roller at the service line. Court sense — and it trains the
exact target-priority read the gore mode rewards.
Wounds are free: only the kill is judged, so pre-softening a heavy while it
is third in line is legal and eventually necessary. The slack (0.9s) makes
near-ties legal both ways, or the rule is a coin flip. The streak is both
punishment and payout: x1..x5, a misread shows OUT OF TURN and resets it, and
nothing else happens — the queue keeps walking. Caught is the real setback.
Best score banks to localStorage; the world board can carry it if it ships.
drillMark draws a chevron over the queue head for playtest calibration. It
answers the exact question the mode asks, so it must be off in anything
shipped.
Deleted: the alphabet, the batch, the strip, the eaters. Shamblers are
first-class stream members with pts 20; a roller is 100.
30 tests, rewritten around arithmetic riggings (known speeds and distances)
instead of luck; 20 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"one-shot is bad - way too easy - also we're losing all the nice gore
animation sprites" and "I noticed the last kill of a game can carry to the
spawn of a new zombie of the next round".
drillOneShot is off. The second half of that note is the part worth
remembering: a 1 hp body never passes through a wound stage, so every skin in
ZWOUND — arm off, ribs out, still walking — is skipped and the roster's best
art never renders at all. Kept as a switch, never a default.
The carry-over was a real unfairness, not a curiosity. A ball keeps pace
through a kill and travels UP-COURT, which is exactly where the next batch is
born — so the shot that finished your round flew on into bodies that did not
exist when you took it, and faulted you on a round you had not been shown
yet. The shot that ends a round is spent now, and drillCaught does the same
since it also refills the court up-court. Pierce within a batch is untouched:
that is honest occlusion you could have read, this was not.
Verified with real aimed serves through a whole round: round banks, ball goes
inactive, the next batch walks on with 0 faults and a clean sequence, and the
gang carries roster HP again (brute 5, ball boy 3).
33 tests, 15 seeds clean.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"please make the zombie killable with a single hit for testing" and "remove
the fake zombies for now".
drillOneShot drops every body on a drill court to 1 hp, applied at spawn to
the batch and the eaters alike. Worth knowing what it removes: the roster's
HP is what makes the called ORDER matter — a roller is normally seven serves,
so being told to take it third means starting on it first. At one hit that
planning layer is gone and the sequence is pure recall. Flagged in PROJECT.md
so it gets decided on feel rather than left on by drift.
drillEaterMax 3 -> 0 switches the plain shamblers off. The mechanism is
untouched — spawnEater, the wide entry away from the gang, the never-a-fault
rule — so one number brings them back.
The eater test now asserts whichever contract the knob describes (at 0:
nothing but the puzzle ever appears; above 0: they arrive, capped, always
plain shamblers) instead of being deleted and rotting. And the never-a-fault
rule gets its own test with a hand-placed body, so the rule the eaters depend
on stays covered while they are switched off.
31 tests, 15 seeds clean. Verified live: brute/steward/roller all at 1 hp
where the roster says 5/5/7, zero strays over 40s, one aimed serve puts down
the roller.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The pipeline caught two flaky tests of mine that sixty local runs had not.
Both were bad tests rather than bad code, and both were mine from this
session:
- the eater test counted SURVIVORS, not arrivals. An eater crosses the whole
court in ~24s and despawns, so a snapshot at 40s can legitimately show an
empty court — that read as "the eaters never came". It tracks arrivals over
the window now, and asserts the cap as a running peak.
- the auto-aim test never actually aimed at the body it claimed to: fy was
pinned at -0.95 and fx was a scaled x offset, so the ray leaned up-court and
the called body could genuinely be the best-aligned thing in the cone. That
is the magnet working. It now normalises a vector straight at the wrong
body, picks the one furthest in x from the called one so the two are
separable, and asserts the magnet takes what you pointed at.
Measured 8 failures in 40 runs before, 0 in 60 after.
And the durable fix: Math.random is seeded at the top of test.js. This suite
is a Dockerfile build stage, so a red run blocks the deploy, which makes a
1-in-5 flake worse than no gate — it trains you to re-run it. The game leans
on randomness everywhere, so the stream is made reproducible rather than the
randomness stripped out. SEED=1234 node test.js re-rolls it; 25 seeds pass.
The cost is noted in PROJECT.md: under one seed a guarded test can silently
take its early-out forever, so I checked all four guards still run their
bodies.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver's insight, and it retires a whole subsystem: "we don't need colors at
all! we already have different zombie types.. so the sequence is the zombies
themselves".
The bib and then the ground disc were both inventions bolted on to give a
body an identity it already had. A brute does not look like a runner, the
game is built on reading those silhouettes at speed, and the strip can show
the body's own walk frame instead of a swatch — no legend, no mapping, no new
art ever. It also deletes the accessibility problem outright rather than
deferring it, because a silhouette is not a hue. DRILL_COLS, drawColourRing,
drawBib, PIP_FACES, blip(), drillRingR and drillTones are all gone.
Two rules keep a type unambiguous: the plain shambler is never in the
alphabet (it is what the eaters are, so "not the puzzle" needs no marking),
and one of each type is on court at a time.
The new failure mode this created is instructive. Types move at different
speeds and drill bodies hold their lane, so one in a lane you are not in
walks the whole court and off the back untouched — stranding the sequence
asking for a body that no longer existed, i.e. the un-winnable state the mode
exists to prevent, reintroduced by a change that looked cosmetic. A batch
body crossing your baseline counts as caught now ("GOT PAST YOU"), which is
also just better design, and the speed spread is narrowed to 1.35..0.70 so a
batch stays coherent. Their real difference is HP and that is untouched: a
runner is one shot, a roller is seven.
Verified: 14400 frames of simulation, zero frames where the sequence named a
body that was not on court.
29 tests.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"remove the mob boy - make the normal zombies rarer and not so close to the
color gang".
sendBoy refuses during a drill, guarded at the function rather than at the
call sites — same shape as its existing G.interlude guard, so every present
and future caller inherits it. He is one more body crossing the lanes you are
reading colours down, and unlike everything else out there he is neither the
puzzle nor the threat. Measured: 0 boy-frames in 90 seconds. Gore no longer
gets mopped in this mode; the existing 40/30 caps are what bound it.
The eaters are rarer — every 12s falling to 5s, capped at 3 on court. At the
old rate they were a second crowd rather than an interruption, and a threat
you break off to deal with only works if breaking off is occasional. 8
arrivals in 90s where the old numbers gave ~13.
And they enter wide of the colour gang: best-of-9 lane sampling against the
bibbed bodies, the same trick the batch uses. Narrowest clearance measured
across 8 spawns was 0.91 m. They still converge on you afterwards — that is
what they are for — this only decides where they walk in, so they stop
screening the bodies you are trying to read.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"the colored zombies should also explode like they did before and make the
normal splash sound on destruction.. no bleeps please.. can we make the color
as a color circle the walk on? the sign they're holding is not nice. also
remove the ball slicing in this mode.. this leads to unwanted hits".
The silent deaths were my bug: the hook in killZombie did drillKill(zb) and
then RETURNED, and everything below that return is the death — decal, pile,
36-particle burst, splat, headshot crack. It resolves after the splice and
without returning now. Both halves are load-bearing and each was a bug in
turn: splice first or a freshly drawn combination names the corpse; don't
return, or the body simply vanishes.
The colour is a disc on the court the body walks on, drawn before the sprite.
The bib was a flat quad pinned to a walking sprite — UI stuck onto a
character, covering the art it sat on. On the ground it takes the court's own
perspective, measures 12.4px across at the spawn line against a chest patch's
few, and hides none of the sprite. A hard dark rim plus a bright inner line
is what keeps it from reading as another gore decal.
No slicing in the drill: a cut ball leaves the corpse and strikes whatever it
finds next, which in a mode where the wrong body is the failure is the game
choosing your target for you, badly.
And the mode has no sound of its own now, deliberately. The umpire reading
the list was vetoed, the arpeggio that replaced him was vetoed. blip() is
deleted. A kill bursts and splats like every other kill in this game; the
strip carries the list.
28 tests.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"make the game that it is actually winable meaning that all required colors
are visible on the far side before when a color combination is shown.. for
now no colors that are not in the current combination but occasional zombies
which don't have a color and are converging to the player trying to eat him".
Two rules define the file now. Every colour on court is IN the current
combination — no coloured decoys, so the strip can never name a body that is
not there. And a combination is only shown once its bodies are all on the far
side, which is free because the batch is BORN there: it is a property of the
construction rather than a check that could be wrong.
The pressure moved to the eaters: plain unbibbed shamblers on their own
clock, deliberately not flagged `drill` so they keep the gore mode's homing
and converge on you while the bibbed batch walks its lanes. Killing one is
never a fault; letting one reach you is the ordinary caught. The puzzle plays
fair and the danger arrives separately, which is a better trade than a puzzle
that cheats.
The batch is the round: kill the last of it and the next walks in on the same
frame with its combination already called.
One real bug, caught by the new test: killZombie resolved the drill BEFORE
splicing the body, so a combination drawn on the spot could name the corpse —
a round asking for a colour that is not on court, i.e. exactly the unwinnable
state this pass exists to prevent. Splice first, resolve second.
28 tests.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"what are the dice dots for?" -> "drop them", and "the dodge swipe should not
reset to the middle when firing".
The pips were colourblind redundancy so the mode did not rest on hue alone,
but they were noise to everyone who can separate the colours and they shrank
to nothing on a body at distance — exactly where the channel had to work.
Gone from the bib and the chip. Recorded in PROJECT.md as a known gap rather
than pretended away: if the mode ships, the answer is bibs that differ by
shape at sprite scale, not dots.
The dodge one was never a drill bug. The auto-assist recovered toward HOME.P
including its x, so firing walked you back to the centre line: measured, a
step out to 1.94 m was dragged to 0.77 m over one ball's flight — most of a
dodge, handed back for free, which is the welded-in-place complaint the
sidestep exists to answer arriving slowly instead of all at once. Splatter
recovers in DEPTH ONLY now and holds whatever lane you stepped to. Tennis
keeps the full recovery: recovering to centre is the real sport's footwork
and there is no sidestep there to protect.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The doc edit threw on its second assert last commit, so the code shipped
without its notebook entry. Also records the two ways a drill test lies to
you: one that never shoots gets caught and the sweep looks like a stalled
stream (use eatenR = 0), and zombies[i] is not the same body across frames
because the array shifts at the baseline.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver: "daniel announcing is maximum annoying" / "don't let the rush in
position just let them coming from the back in steady speed.. try to space
them not too close" / "when one combination is done just let the new one in..
in a continuous stream that increases in speed and difficulty" / "the rounds
can stay, they should be continuous".
The round survives as the unit. Every seam around it is deleted: no walk-on
to a mark, no announcement beat, no between-rounds pause, no deal. Bodies
walk in from the far baseline forever and a round is drawn from whoever is on
court — finish one and the next is drawn in the same frame, off the same
rolling population. The bodies you were ignoring a second ago are the ones
being called now.
- Daniel is gone from this mode. Each colour has a pitch; a round announces
itself as a four-note figure in half a second (blip()). drillTones 0 mutes
it. The recordings stay in assets.
- PLACEMENTS is deleted with the marks — nothing rushes anywhere and nothing
stops, so there are no shapes to place.
- A DRILL BODY HOLDS ITS LANE. updateZombies homes bodies onto the player's x
past the net, so they arrived spread and bunched into a knot right in front
of you; spawn spacing cannot survive that. Their lane is the puzzle here.
- One ramp on TIME SURVIVED drives speed, spawn gap and round length. Speed
is stamped per body at spawn, so the ramp reads as newcomers outpacing the
ones already walking, not the court accelerating under you.
- The empty-court watchdog is gone and nothing replaced it: with a stream
there is no state to be stuck in. One line covers every case.
27 tests, rewritten around the stream.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
"the drill mode immediately goes into some horrible audio loop" — on a phone,
never on desktop, and that is the tell. Three of the four defects can only
happen on a phone, and every one of them is a rule voice() had followed for
months that my new function quietly broke.
- A SUSPENDED AudioContext HAS A FROZEN CLOCK. Scheduling four clips at
currentTime +0/+0.9/+1.7/+2.4 while the context is suspended puts all of
them in the past by the time it starts, so they fire on one instant through
a 1.3s reverb tail. Resume first, re-read the clock after the decode.
- A second call layered instead of replacing. voiceSeq SET voiceUntil and
never checked it. It stops the outstanding list now.
- ...and cancelling a list has to give back the time it booked, or the
replacement queues behind a list that will never play. Measured at 6.6s of
silence before the call the player is waiting on.
- The amplifier was in drill.js: the empty-court net redealt on the spot,
making the repair rate the frame rate — and every deal calls the umpire. It
goes through the ordinary between-rounds beat now.
Verified by tagging each decoded buffer with its source file and counting
start() calls: one drill_call, three colours, spaced, zero overlapping voice
lines. The earlier "clean" reading counted fetches, which proves nothing —
VOICE[] caches, so a repeat play never fetches again.
27 tests; the new one empties the court every frame for six seconds and
asserts at most one umpire call per beat.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Oliver played it: "the auto aim just mean no skills involved at all. this
should just go away.. there is no penalty at the moment.. make some
variations of the position on the court and the color".
The first two notes are one finding. With the shot locked to the called body
it was unmissable, so the fault could never fire — there was no penalty
because there was no way to earn one. Both snaps are deleted. What replaced
them is one line: aimTie is 0 during a drill. That is not auto-aim, it is
removing the guess — the ordinary magnet lets a nearer body inside the same
cone steal a shot aimed past it, which is right for a crowd you are mowing
down and exactly wrong where hitting the wrong body IS the failure. An
unaimed tap now lands on grass.
And the variation:
- six colours (purple, white) with the court dealt three to five of them, so
which colours turn up and how many decoys there are are live questions. At
four out of four every round dealt every colour and only the order moved.
- PLACEMENTS: fan, cluster, column, stagger, wing — weighted, mirrored per
round, lanes shuffled against the bib order. column stacks bodies in one
lane, and the ball hits the first thing it touches, so feet matter.
- pips are dice faces now, shared by bib and HUD chip. A row of five dots has
to be counted; a die is recognised.
Measured over 30 deals: all five shapes, counts even, all six colours, and
the first target screened by another body in ~10% of rounds. Verified end to
end in the browser that FAULT and DOUBLE FAULT both fire.
26 tests, including a named regression for the auto-aim staying gone.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The late game is repetitive because the optimal move never changes: kill the
nearest thing, now. This is the first swing at a mode where it does — a
coaching drill where the umpire CALLS a colour sequence aloud while bibbed
bodies take their marks, and you must put them down in that order while they
walk at you. Uncalled colours are decoys you have to ignore. Two faults is a
double fault and the round is theirs.
Back room only (DRILL TEST, or #drill): this is a feel-test for one question —
tense, or just noisy? — so it keeps its own round count and touches neither
tennis scoring nor the run.
- voiceSeq: voice() drops a line that would start while the umpire is still
talking, which is fatal for a list. Decode first, queue at accumulated
offsets, and the anti-overlap rule is satisfied instead of fought.
- Four colours with PIPS as well as hue — red/green alone makes the mode
unplayable for ~8% of men.
- drawBib is a ponytail-marked overlay that knowingly breaks the sprite rule;
it is the thing to delete if the mode survives.
- BOTH target pickers snap (flickTarget and zombieTarget). Snapping only the
slingshot left fifteen tap-serves in a row scoring nothing.
- Contact costs the point, not the run — permadeath stays in the gore mode.
- startDrill disarms the ordinary schedule; updateDrill redeals and logs an
INVARIANT if anything empties its court.
Eight new tests (24 total) pin the ownership contract both ways, the call
beat, the win, the double fault, the un-winnable-round guard and the death.
PROJECT.md carries the four open questions for the playtest.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The gun, beacon, flamer and rocket fell silent the moment they cleared the court:
firing and steering were both gated on phase 'rally', and every beat raised by the
kill that empties the court (a point, a won game, the mop-up) parks the phase in
'over'. A burst is its own control mode for its whole len, like a horde — its own
clock is the only thing that ends it. Not fixed by forcing the phase back to rally:
a parked 'over' beat that never expires is how pendingHorde gets stranded.
The countdown moved off the tile's bottom edge (the rounded frame, and the bottom of
the screen) into the tile: the armed slot's well drains as a gold wash, clipped to
its own rounded path, under the icon and labels.
16 tests.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
THEY KEEP COMING is a 2.4s 'over' beat again (warnHorde in beats.js) instead of a
flash raised on the same frame the wave lands, and the chapter boss walks on behind
an overlay — his name in red over ENTERED THE COURT — held for 3s by the new
self-dismissing modal clock (m.hold) rather than announced on a routine-set card.
startHorde returns a bool: if the court has been taken when the warning expires the
pendingHorde latch is re-armed rather than the set losing the horde it paid for. The
warning ticks on its own clock, not on the 'over' beat, so a beat that moves the
phase underneath it can't strand it (the mgT lock, again).
Three tests, 15 total.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
wuwiandClaude Fable 5