The bot now reaches the renderer at http://renderer:3000/api/gif on the
compose network instead of a separately deployed service. Both share
WHEELOFNAMES_API_TOKEN. CI runs the renderer's lint, typecheck, tests,
render smoke, and image build.
The glint band was a card-sized box sliding diagonally, so the top-right
and bottom-left corners never fell inside it. Draw the band on a box three
times the card's size, centred on it, so every corner is lit.
Blend the glint with plus-lighter and tint it with the card's accent so it
brightens the card's own tone instead of painting flat white over it.
The pack prints a crescent moon in its diamond in place of the default star, and the card back carries the same moon in its diamond seal over the miti99bot caption.
The pack and the card corners no longer carry text. The card's text stays hidden while it spins and fades in once it lies flat, then a mirror glint sweeps from the top-left corner to the bottom-right as the card comes to rest.
Brings back the Genshin-style meteor wish that /api/gacha rendered before pack-cards, under genshin names. It takes the same request as /api/gacha and returns the 7-second landscape MP4.
Three things moved the text once the card looked still:
- The spin's last degrees: Chrome draws text on a slightly tilted card
differently from a flat one, so glyphs hopped as the spin ran out. The
spin now lands flat a moment before the card stops gliding.
- The landing: the card flew at its rounded size scaled to a fractional
layout size, then was laid out again, shifting the text. The card's
size is now whole pixels.
- The last flight frames: a sub-pixel creep at almost-identity scale drew
the text differently from the card at rest. The flight now reaches its
final pose slightly early.
The scripted drag across the seal, the torn pack falling away, and the
card rising out of it now ease exponentially in and out. The pack keeps
pack-cards' 1.35s drop and hold, and the card still clears the pack when
the spin starts.
The scripted drag released the mouse over where the card lands. Once the
card arrived under it, pack-cards tilted the hovered card toward the
pointer over its .22s transition, so the face kept moving after the spin
stopped. The pointer now moves to the corner when the drag ends.
The sliding rainbow sheen over the whole face and the spectrum foil are
gone. 4★ cards are now engraved with pack-cards' epic contour and 5★ cards
with the legendary rings or facets, and those engraved lines carry a still
rainbow; 3★ cards stay plain.
The warm-up also renders a 4★ wish, since the first card with the contour
engraving otherwise rendered about seven seconds slower.
/api/gacha now renders the 6-second portrait pack-cards opening that was
served on /api/gachabeta. The Remotion meteor wish, its timeline, and the
beta route are removed, and the pack-cards renderer and page take the plain
gacha names.
BREAKING CHANGE: /api/gachabeta is gone, and /api/gacha returns a 6-second
portrait video (360x640 for width 640, 480x854 for width 854) instead of the
7-second landscape one.
The card now rises out of the pack at pack-cards' own speed and only starts spinning once it is clear. The rest of the flight is stretched so three and a half turns ease out smoothly before the card lands face up, and glowing sparkles in the rarity colour burst around it while it spins. pack-cards is unchanged; the page reshapes its flight animations.
/api/gachabeta now returns a portrait video with the requested width as its long edge: 640 renders 360x640 and 854 renders 480x854. The request body is unchanged, and /api/gacha stays landscape. The pack sits higher so it stays in frame while the card rises.
pack-cards reveals the card with a half turn from rotateY(180deg) to 0deg. The page starts that one animation at 540deg, so the card spins further before landing face up, without changing the library.
/api/gachabeta now tears open a pack-cards collectible pack whose card
shows the request's label, the same B/A/S rank as /api/gacha, and one
star per rarity level; rarity also picks the card material and the pack
colour. pack-cards animates on the browser clock, so the route drives a
shared headless Chrome in deterministic mode over a debugging pipe,
steps virtual time frame by frame, tears the pack with a scripted drag,
and encodes the captured frames with Remotion's bundled ffmpeg. The
page and package are served from disk with all other requests blocked.
The browser stays alive per process and is warmed at start-up, because
the first render compiles the pack's WebGL shaders in software. The
Remotion beta composition and its scene code are removed. pack-cards is
installed from a GitHub tarball pinned to a commit, since it has no npm
release and a moving URL would break npm ci.
The beta splash card used a fixed SSS rank and six stars. It now uses the
same rarity info as /api/gacha: the rank letter (B, A, or S) on the emblem
and name plate, and one star per rarity level beside the caller's label.
The beta wish now cuts like an anime gacha game's six-star wish over 10.5
seconds: a dive through a painterly cloud vortex, a beam pass, a meteor
that stops inside a rainbow halo and bursts, a red flash, falling crystal
comets joined by a red meteor, a black emblem silhouette shedding shards
on a red disc, and a splash card with a name plate and six stars.
Cloud layers are painted once per roll onto a canvas with a seeded noise
displacement, so frames only move the baked image instead of running a
live SVG filter. The /api/gachabeta contract is unchanged.
The beta wish now renders an 8-second night sky over a snow-rimmed
mountain ridge, where coloured meteors glide down the diagonal. A hero
meteor in the rarity colour slows and burns out at the centre, where the
label appears under rank SSS. The toon cloud, comet, and perspective
camera are removed. The /api/gachabeta contract is unchanged.
The beta wish renders an 8-second astrology-themed night sky in toon
shading through a perspective action camera: it dollies toward a hero
cloud, a comet lights the cloud's rim from behind and bursts through it,
the camera chases the comet as its flare builds to a starburst, and the
label appears with an S, SS, or SSS rank. It shares the /api/gacha request
contract and render slots.
The meteor now flies in from the upper left and lands on the rank emblem,
so the impact burst becomes the badge reveal. The 5-star rainbow ring is
replaced by a rainbow sunburst of counter-rotating rays. Sky stars twinkle
with short bright flares and cross glints, and each roll draws a fresh
layout from a per-request seed that the route picks when the caller sends
none.
Measured from reference footage, the meteor sweeps in along a straight line
from the upper left at about 23 degrees and eases out exponentially to a
hover above the horizon, replacing the gravity arc. The trail is now a soft
beam that narrows into the head with ribbons fanning along it, every tier
gets the horizontal lens flare, and sparkles drift above the beam.
The meteor now follows a ballistic parabola: constant horizontal speed and
vertical speed growing under gravity solved from the launch and impact
points, so it climbs slightly, bends over, and dives while speeding up. The
trail samples past positions in time so it lengthens with speed, the head
stretches along its velocity, and sparks inherit the meteor's velocity and
fall under their own gravity.
Every beat now scales from a per-tier effects table: 4-star adds a bigger
meteor, lens flare, impact shake, and double shockwave; 5-star adds a
rainbow halo before landing, a gold sky flood, a starburst, counter-rotating
rays, falling sparkles, and an emblem sheen. The emblem shows the rank
letter instead of the label's first character.
A meteor coloured by rarity falls across a night sky, lands in a white
flash, and the label is revealed with its stars popping in. The caller
picks the result and rarity; the route only draws it. Renders as silent
H.264 so Telegram plays it as an animation without GIF banding, and
shares the render slots with /api/gif.
Coolify attaches the container to its proxy network and routes to the exposed
port, so the host publish was redundant and additionally reachable on
0.0.0.0:3000 outside TLS. Left commented for standalone `docker compose up`.
Replace pnpm-lock.yaml with package-lock.json. allowBuilds for esbuild becomes
package.json#allowScripts. The Dockerfile installs with npm ci --omit=dev and
no longer prepares pnpm through corepack; .npmrc keeps engine-strict, which npm
honours natively.
minimumReleaseAgeExclude is dropped rather than translated: pnpm's
minimumReleaseAge was never set, so the 30-entry exclusion list had nothing to
exclude from. Direct dependency versions resolve identically to the pnpm
lockfile.
Draw the winner outline as an unfilled pie after every slice: slices paint
in index order, so the next one covered the shared radial edge and the
outline stopped short of wrapping the wedge.
Derive the announcement colors from the winning slice - its own color as
the chip, a deepened version as the frame and the wedge outline - instead
of a fixed dark chip that read as unrelated to the wheel.
Raise the confetti above the announcement so the burst is no longer hidden
behind it, and steepen the launch fan so the field arcs around the name
rather than settling on it: horizontal reach is vx / drag, which parked the
flat fan at mid-canvas. Measured over the hold, particle samples inside the
pill drop from 22.4% to 6.2% while disc coverage holds at 85.5%.
Three advisory reviews with measured evidence: the confetti celebration, the
wheel and theme design, and the API/operator experience. The API findings are
untouched work - broken render:local CLI examples, 400/413/415 conditions
surfacing as 500s, an unenforced timeout guarantee, and a production token
default in compose.yml.
The refinement plan covers the second pass and records what was deliberately
excluded, including the bezel, the font-size cap, a night theme, and narrowing
the request schema.
README described confetti as reusing theme slice colors and made no mention of
the winner callout or the reveal timing.
Fill in the benchmark rows that were TBD, and record where the GIF growth
actually came from. Isolating each change shows the celebration rework, pill,
spin retune, palette reorder and pointer together account for +3.3%, while
grayscale label antialiasing accounts for +22.4% on its own because it affects
every spin frame rather than only the celebration. That trade is deliberate:
it removes visible color fringing and stops the palette being exhausted.
Also correct the size levers. Particle count is not one - cutting 43% of the
particles changes the file by 1.1%. Note that render times on this machine vary
more than 2x across repeat runs of identical input, so single timings are
indicative only.
The animation's weakest moment was its final frame, which is also the frame
most chat clients show as the GIF's poster image.
Celebration:
- Confetti launches from two cannons framing the wheel instead of bursting
inside the hub, using closed-form drag ballistics so chips decelerate and
flutter rather than flying a clean parabola. The field previously drained
from 70 particles to 3 before the GIF ended; it now stays populated, and a
2500ms hold ends with 68 on screen instead of 0.
- The opening frame rendered at opacity 0, throwing away the one frame where
particles are still clustered. Removed the fade-in and made the tail
frame-relative and quantized to bound its palette cost.
- Chips streak along their direction of travel in proportion to speed, faking
the motion blur a GIF cannot carry. A second volley, sized to the
celebration window, keeps long holds from thinning out.
- Each theme carries its own confetti palette. Slice colors are tuned to hold
dark text over a large area, which left five of six mono colors under 2.4:1
against the background and invisible at particle size.
- The winner's name appears over the hub during the hold. At a 200px chat
display the pointer is ~17px and a slice label ~9px, so the poster frame did
not say who won.
- Winner emphasis is an opaque stroke, not a drop-shadow. A filter on a Pie
paints outside the shape, so later siblings overpainted it and only one
neighbour ever showed the halo.
Motion:
- Turn count scales with the frame budget and slice width. A fixed 5 turns put
peak speed at ~40deg per frame regardless of option count, so 65 of 98
frames advanced more than a full slice at 32 options and the wheel appeared
to run backwards. Turn counts floor rather than round, since rounding up
reintroduces the aliasing, and stay integral so the winner still lands on
the detent.
- The spin launch now blends into the deceleration at a matched slope. The
previous curve started at maximum speed, and a naive ease-in jumped from 18
to 41deg per frame in a single frame.
- A damped recoil rocks the wheel into its detent, resolving to exactly zero
so the landing and the flapper's rest position are unaffected.
- The pointer is drawn as SVG with a real outline and a tick as slices pass.
clip-path had been discarding its border and shadow entirely.
Legibility:
- Labels flip by their rest-frame angle so every name reads upright in the
final frame. This reverses an earlier decision to leave them slice-aligned;
three of eight names rendered mirrored on a default wheel.
- Label text renders inside a promoted compositing layer, which switches
Chromium from LCD sub-pixel to grayscale antialiasing. Sub-pixel fringes put
2178 colored pixels into an all-grayscale theme and consumed 255 of 256
palette entries. Promoting the one rotating container rather than each label
keeps render time flat on dense wheels.
- Line boxes leave room for stacked Vietnamese diacritics and shrink to fit
rather than clipping.
- Slice colors are ordered for colorblind separation, and the last slice
avoids reusing the first slice's color when the option count leaves a
remainder of one.
Update three transitive dependencies to their patched releases (fast-uri 3.1.4, find-my-way 9.7.0, postcss 8.5.22), clearing 3 high-severity Dependabot alerts. All were within existing parent semver ranges. Verified with pnpm audit (clean), lint, typecheck, 58 tests, and the API render smoke test.
Route the CLI through the shared winner picker instead of crypto.randomInt so the positive-integer guard and selection intent stay consistent with the API.
tiennm99