633「그림자극 이야기」 — 등불 비친 창호지에서 갓 쓴 인형이 빈 틀에 글자를 넣자 나무가 자라나는 14초

Make a short vertical shadow-puppet film as one self-contained HTML file and render it to MP4. The idea: a tiny story told with shadow puppets behind a glowing paper screen, in a Korean traditional mood. A puppet wearing a gat (traditional Korean hat) slides in, puts letter cut-outs one by one into an empty frame on a stand, a sprout bursts out of the frame and grows into a full tree, and two birds fly in and settle on its branches. Warm captions tell the story under the screen. It must feel handmade, quiet and moving. Do the whole job now without asking questions.

[Inputs (use the default when empty)]
Name: "【이름】" — default 프롬프트판
Opening title: "【첫 화면 제목】"
Letter cut-outs (slash-separated, 1 to 4): "【넣을 글자 조각】"
Three captions (slash-separated): "【자막 3줄】"
End address: "【끝 주소】"
Light: "【빛】"
  등불 = lamp light #FFE6B0 → #FFB35C → #C9622A, shadows #140C08, wooden frame #4A2E1E, caption light #FFD38A
  달빛 = moonlight #EAF2FF → #A9C6FF → #4E6FB8, shadows #0A0E18, frame #1E2A44, caption light #DCE8FF
  노을 = dusk #FFE0D0 → #FF9A8A → #C2455A, shadows #160A0E, frame #4A2230, caption light #FFC4B0
Tempo: "【박자】" — default 100 BPM. Aspect ratio: Output size 1080x1920 pixels, 9:16 portrait story. Keep all text inside the central 80% of the height: compress every zone percentage below into the 10-90% range. (default 9:16).

[The stage]
A dark room with a wooden-framed paper screen filling the upper two thirds. Behind it a lamp: a radial glow that flickers very gently (two slow sine waves), faint vertical lattice shadows and fine paper fibers. A dark rolling hill along the bottom of the screen. Every shadow is solid dark with a very slightly soft edge. The puppets are flat cut-outs on thin rods that run down out of the bottom of the screen — the rods must be visible.

[Puppets]
- The person: a long robe (trapezoid), a round head, a gat (wide flat brim and a small crown), one jointed arm (upper arm and forearm with a small hand) moved by its own rod. It slides rather than walks, with a small bob.
- The frame: an empty square outline on a thin stand, slid in from the right.
- Letter cut-outs: the given letters in a serif shape, flying from the hand into the frame along small arcs.
- The tree: a recursive branching silhouette that grows level by level from the top of the frame (thick trunk, thinner twigs), then small oval leaves pop out at the tips on springs.
- Birds: two silhouettes with flapping wings, big enough to read, that fly across the bright part of the screen from both sides and perch just inside the outer tips of the crown; each has a thin soft rim of lamp light so it never melts into the dark leaves. Then they breathe slowly.

[Structure — 24 beats in 4 scenes of 6]
1. Curtain (0–6). Frame 0 is already a finished picture: the glowing screen with the title as a shadow in its upper part. The title drifts up and blurs away as the person slides in from the left; the frame slides in from the right.
2. Letters (6–12). Caption 1 rises under the screen. The person reaches out and puts each letter into the frame, one per beat and a bit, arm moving up and back each time. Caption 2 replaces caption 1.
3. Growth (12–18). The letters fade into the frame, a sprout bursts up from its top and grows into the whole tree, level by level, leaves popping at the end; the person leans back in surprise with its arm raised. Caption 3 appears.
4. Birds (18–24). The two birds fly in and land; the light glows a little warmer; the name rises in warm light letters under the screen and the address under it. The flicker and the birds' breathing keep the last frames alive.

[Motion rules]
Nothing pops in: puppets slide, grow and land on eased curves and closed-form springs; captions rise from mask lines and leave the same way. One slow camera push per scene plus a very slight handheld float, so nothing freezes for more than half a second. No timecode, no progress bar, no corner labels, no HUD. Light film grain. Typefaces: Nanum Myeongjo 800 for the title, letters, captions and name, a monospace for the address, from Google Fonts.

[Sound]
If you can generate music, add a quiet, warm piece at the given tempo — a plucked string melody over a soft drone, a small wooden knock when each letter lands, a rising chime while the tree grows and bird chirps at the end — muxed 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 and a light arpeggio, no vocals) at the given tempo with a simple chord loop, sample by sample into a WAV, 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 Korean labels: text inputs for every text input (letters and captions slash-separated), selects for light, tempo and ratio, play/pause, a scrubber, and a "설정 복사" 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 one still per beat into a contact sheet and look at it: frame 0 is a finished picture, the rods are visible, each letter really travels from the hand into the frame, the tree grows from the frame and never covers the captions or the name, the birds land on branches, nothing is clipped. 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]
Output size 1080x1920 pixels, 9:16 portrait story. Keep all text inside the central 80% of the height: compress every zone percentage below into the 10-90% range.
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.