608「一文広告」白黒キネティック — 単語がビートごとに画面を反転させて打ち込まれる14秒

Make a short vertical kinetic-typography ad as one self-contained HTML file and render it to MP4. The idea: one sentence, said one word at a time — every word gets its own full screen on the beat, the screen flips between black and white on every word, then the words fly back together into the full sentence, then the name. It must look like a black-and-white type ad from a top design studio: huge heavy letters, hard cuts on the beat, one accent color used only for the emphasis word. Do the whole job now without asking questions.

[Inputs (use the default when empty)]
Name: "【名前】" — default Promptpan
Small kicker: "【小見出し】" — a short monospace line, default "PROMPTPAN — NO. 11"
Sentence: "【一文】" — split into words at spaces; a part wrapped in *asterisks* is the emphasis word (keep it as one word even if it has spaces); use at most 10 words
Extra line: "【ひとこと】" — same *emphasis* rule
End button text: "【最後のボタン】" — default: the name
Palette: "【配色】"
  White-black = white #FFFFFF, black #0B0B0B, accent red #FF3B1F, grey #8A8A8A
  White-black · Yellow = white #FFFFFF, black #0B0B0B, accent yellow #FFD400, grey #8A8A8A
  Paper · Ink = paper #EFE9DE, ink #151413, accent red #C8102E, grey #8C857A
Tempo: "【テンポ】" — default 120 BPM. Aspect ratio: 出力サイズは1080×1920ピクセル、9:16の縦型にしてください。文字は中央の安全な範囲に収めてください。 (default 9:16).

[Structure — 5 scenes, each long enough to read]
Give each scene its own length in beats: at least 4; for text scenes the Korean reading time (0.9 s + 1 s per 7 characters) rounded up to an even number of beats.
1. Line: black screen; the kicker types in small monospace above the centre; a thin white line draws out from the centre to both sides on the first beat; in the last beat the line thickens until white fills the whole screen.
2. One word per beat: each word alone, centred, as big as it fits (about 86% of the width) in a 1.3x-wide heavy display face (Black Han Sans); the background flips between black and white on every word and the letters take the opposite color; each word punches in from 118% to 100% in the first third of its beat, with a short flash; a small "02 / 04" counter above. The emphasis word is in the accent color and gets an accent underline that swipes in. Scene length = number of words rounded up to an even number of beats.
3. The sentence: on white, the words come back one by one, alternately sliding in from the left and right, and settle into centred lines in a heavy Korean sans (Noto Sans KR 900), wrapping when a line is too wide; the emphasis word is in a bold serif italic (Nanum Myeongjo 800, slight skew) in the accent color. Near the end a solid black block sweeps under the emphasis word and the word turns white on it. The kicker sits small in the top-left corner. Fade to black at the end.
4. Extra line: on black, a thin grey rule draws across, the extra line types in above it in white with the emphasis in the accent color, then fade to white.
5. End: the name assembles from three cut pieces of each real glyph flying in and locking together; an accent dot pops in as the full stop with a burst of small accent particles; the sentence types under the name in grey; a black end button grows in, a cursor clicks it on the beat and a ring spreads. Hold the final frame.

[Always-on details]
A thin 5-segment story progress bar along the top fills with time and recolors to stay readable on black and white. Real motion blur on fast moves (average several sub-frame samples). Light film grain. Three typefaces only (Noto Sans KR, Nanum Myeongjo, Black Han Sans stretched 1.3x wide), loaded from Google Fonts. Keep a 6% safe margin; never clip a Korean syllable; no gradients or glows — flat black, white and one accent.

[Sound]
If you can generate music, add a dark synth backing track at the given tempo whose kick lands on every word cut, and mux it into the MP4 as AAC. If you cannot generate audio files with a music model, write the backing track yourself in code instead: synthesize an original instrumental (kick, snare, hi-hats, bass, a chord pad, no vocals) at the given tempo with a simple chord loop, sample by sample into a WAV, with the drums entering where the one-word scene starts, then mux it into the MP4 as AAC. Never use a placeholder or someone else's recording.

[Build]
One HTML file, everything on a single canvas. The script starts with a `const CONFIG` object holding every input plus fps 60 and a seed. All drawing comes from a pure `seek(t)` — no timers, no CSS animation, randomness only from the seed — so any frame renders identically. A settings panel above the preview (hidden with ?render=1) with Japanese labels: text inputs for every text input, selects for palette, tempo and ratio, play/pause, a scrubber, and a "設定をコピー" button that copies CONFIG as code. Edits re-render instantly.
Write render.mjs: open the page with ?render=1 in headless Chromium (Playwright), wait for document.fonts.ready, step seek(frame / 60) through every frame, screenshot the canvas at 1080 px on the short side, and pipe into ffmpeg (libx264, yuv420p, crf 18) to make motion.mp4. Close the browser at the end. Accept an optional JSON file argument that overrides CONFIG.

[Check before finishing]
Render, export frames at every word and the middle of every scene into a contact sheet and look at it: every word is readable at phone size and never clipped, black and white really flip on every word, the emphasis word is the only accent-colored text, the sentence lines wrap cleanly, the full stop sits clear of the last letter. Fix and re-render if anything fails. Finish with the file paths and the single command to re-render after editing settings.

[Final output format]
出力サイズは1080×1920ピクセル、9:16の縦型にしてください。文字は中央の安全な範囲に収めてください。
This selected output format takes priority over any fixed canvas size or aspect ratio elsewhere in this template. Adapt the composition, text zones and margins to the selected canvas without stretching or cropping essential content. Device safe areas apply only when a device is selected. Never print these format instructions on the image.