Prompt Motion
@AiMotionSiteS
Prompt library

Films

Interactive POV disaster short with cliffhanger choice card

Interactive POV disaster short with cliffhanger choice card is a 42.3-second Claude Opus motion design reference by @laurgrowth, built around Three.js, AI voice. GOAL: ROLE You are a senior motion designer, 3D technical artist and short-form story writer. Copy this motion film prompt, keep the creator's frame counts and palette, and adapt the subject for your own launch film.

@laurgrowthView post

Prompt

GOAL: ROLE You are a senior motion designer, 3D technical artist and short-form story writer. MOTION: You build fully code-animated vertical videos for a faceless "choose your own adventure" disaster series. No AI video or image generators. Every frame is rendered from code you write. GOAL Produce one finished episode of an interactive POV survival series: 40–45 seconds, 1080×1920, 30fps, ready to post on TikTok, Instagram Reels and YouTube Shorts. The viewer IS the main character. The episode ends on a two-option choice that viewers answer in the comments, and the next part continues from the most-commented option. INPUTS (ask me for anything missing, never guess) - Series disaster: [e.g. tsunami / earthquake / volcano / missile alert / blizzard] - Real-world location: [e.g. Miami Beach / Los Angeles / Tokyo], plus a fictional local landmark name: [e.g. "Marlin Beach"] - Part number: [1, 2, 3…] - Previous choice (Part 2+): [e.g. "THE HATCH"] - The two choices for the ending: [e.g. BOAT / HELICOPTER] - Recurring companion character name: [e.g. Jay], plus one or two side characters TECH STACK - three.js (pinned version) in a single HTML file, rendered headless with Playwright. - Deterministic timeline: expose https://t.co/IfGGJ8odPH(t) that sets every object, camera and effect for time t in seconds. No Math.random() at render time; use a seeded PRNG. No requestAnimationFrame-driven state. - Render loop: for frame 0..N, call seek(frame/30), screenshot the canvas at 1080×1920, then encode with ffmpeg (H.264, CRF 18, yuv420p, faststart). - Audio built separately (see AUDIO), then muxed with ffmpeg and normalized to -14 LUFS, true peak -1.5 dB. ART DIRECTION - Look: flat toon-shaded 3D, like a hand-painted cartoon. MeshToonMaterial with a 3-step gradient map, plus an inverted-hull black outline (2–3 px at 1080 wide) on characters and key props. No realistic textures, no PBR shine. - Palette: sky #9cc9e8 → #f3e6c8 at the horizon (golden-hour scenes add #f6b27a), sand #e8d3a6, ocean #2f7f86 with foam #d8f3ee, deep water #1f5a5e, warm interior wood #8a6a4a, metal #6d6f73, accent red #d8343a, safety yellow #f2c230. - City: low-poly glass towers (light blue, white window grid), pastel low-rise hotels and bars with simple fictional signs, palm trees as 3–4 flat leaf cards on a curved trunk, lifeguard tower in yellow and blue. - Characters: chibi proportions (head about 35% of height), round tan heads, large white eyes with small black pupils and a white highlight, thick dark brows that carry most of the emotion, a simple mouth shape per expression (O-shock, open shout, tight grimace), short brown hair as one toon mesh. Companion wears a red t-shirt and navy shorts. Mitten-style hands with a separate thumb. - Expressions change in whole poses (shock, shout, relief), never subtle lip sync. Mouths flap open and closed on dialogue syllables at roughly 8 Hz. CAMERA (THE MOST IMPORTANT PART) - True first-person POV. The viewer's own two arms are always visible from the bottom corners: tan skin, red sleeves to match a "you" character, mitten hands. The arms reach, grab, push and point on every beat. - Wide lens: FOV 95–105 with slight barrel distortion (a post-process pass, strength 0.15) so it feels like a head-mounted camera. - Head motion: constant subtle bob (0.6 Hz, 1.5 cm), panic sway when running, hard shake on impacts (decaying noise, 0.25 s, up to 4° roll), quick whip pans between beats (0.15 s ease-in-out). - Look up and look down a lot: up at the hatch, the wave and the helicopter, down at the ladder and the water. Vertical framing is the point. - Cut every 1.2–2.2 seconds. Each shot gets one clear action and one line of dialogue. WATER AND DISASTER FX - Wave: a large curved tube mesh with a toon water shader (two-tone teal with a moving foam band along the lip), plus white speed streaks that whip past the camera on impact. - Flooding interior: a water plane rising with a wobbling surface, white splash cards, and arcs of water spraying in from seams. - Underwater: teal fog (density about 0.08), slowly drifting light-blob caustics on the floor and ceiling, rising bubbles, a muffled low-pass on all audio, floating debris (cars, beach umbrellas, coolers, suitcases) drifting slowly with tumble. - Surface: horizon bobbing, a second wave building in the distance, debris on the water. - Rescue: a cartoon helicopter (white and navy, spinning rotor shown as a translucent disc), a rope ladder swinging, rotor downwash rippling the water. STORY STRUCTURE (every part) 1. 0–2 s HOOK: start mid-panic, no intro. The HUD shows location, time and the previous choice. Something is already going wrong. 2. 2–12 s ESCALATION: the plan from the last choice works at first ("It's holding!"), then fails ("It's coming in!", "It won't open!"). 3. 12–22 s WORST MOMENT: you're underwater or trapped. Dialogue stops. Sound goes muffled, then a hard silence of 0.5 s. 4. 22–34 s SURVIVAL: you surface, a friend calls your name, a new threat appears ("Another one's coming!"), rescue arrives with only one chance ("One pass!"). 5. 34–40 s CLIFFHANGER: you're reaching for the rescue as the next wave hits. Cut to black for 0.3 s. 6. 40–45 s CHOICE CARD (see END CARD). Hold it for 3+ seconds. Rules: dialogue lines of 1–4 words, shouted, present tense. Name a character at least twice per part so viewers learn them. Plant one small mystery that pays off later (for example: "Is that really Jay?"). ON-SCREEN TEXT - HUD, top-left, the whole video: "[LANDMARK] · [TIME]" in white, semibold, 28 px, plus "[City] · Part [N]" in yellow, 22 px, under it. - Previous-choice badge, top-centre, first 3 seconds: a red rounded pill with "YOU CHOSE" (small) over "[CHOICE]" (large, white, bold) and a simple icon. - Dialogue captions: centred at 72% of the height, bold rounded sans, 44–52 px, 3 px dark outline. Colour per speaker (you = white, companion = yellow, side character = purple or green). Shouted lines in ALL CAPS, normal lines in sentence case with "!". One line per shot, in on the first syllable, out on the cut. END CARD - Dark, blurred freeze of the last shot. "PART [N+1] · YOU DECIDE" in small yellow letter-spaced caps. "WHAT WOULD YOU DO?" in large white bold, 2 lines. - Two big rounded buttons: option 1 in blue (#2f6fe0) with an icon, "OR" between them, option 2 in red (#d8343a) with an icon. - "Comment below ↓" at the bottom. A gentle 1.03 pulse on both buttons. AUDIO - Voices: short shouted lines from 2–3 distinct voices (male, female, kid if needed) in a TTS tool, with energy and breath. Add panting under the running shots. - SFX, synthesized or from a free library: wave roar build, impact boom, metal hatch clang, water rush, underwater muffle with a heartbeat, surface gasp, helicopter rotor loop with doppler, distant sirens. - No music in the first 30 s, only a low tense drone. A soft rising swell under the end card. - Mix: voices sit 6 dB above SFX. Underwater section low-passed at 600 Hz. Export at -14 LUFS. THUMBNAIL / COVER (separate 1080×1920 PNG) - The companion's shocked face, hands on head, in the scene. - Two-line title in a heavy condensed font (like Anton) with a thick black outline: line 1 white, e.g. "TSUNAMI:", line 2 yellow, e.g. "MIAMI BEACH". - A small black-backed subtitle with the episode's choice, e.g. "HATCH OR BUILDING?". - A red "PART [N]" pill near the top. DELIVERABLES 1. episode.html: the full scene with https://t.co/IfGGJ8odPH(t). 2. render.js: the Playwright frame capture. 3. audio: voices, SFX and the final mix as WAV. 4. part[N].mp4: final 1080×1920, 30fps, about 42 s. 5. cover.png: the thumbnail. 6. caption.txt: one line + "comment BOAT or HELICOPTER" + 3–5 hashtags. 7. next.md: a 6-beat outline for BOTH possible next parts, so you can post within hours of the comments deciding. QA BEFORE EXPORT - Arms are visible in every first-person shot and never clip through objects. - No frame is static for more than 2 s. Something moves in every shot. - Every caption is readable on a phone at arm's length. - The end card is on screen for at least 3 s. - Watch it once muted: the story still makes sense from captions alone. DAILY SYSTEM Run this every day: read the top comment choice from yesterday's part, write the next part from next.md, render, cover, caption, and save into a folder named by date and part.
Author original
ROLE You are a senior motion designer, 3D technical artist and short-form story writer. You build fully code-animated vertical videos for a faceless "choose your own adventure" disaster series. No AI video or image generators. Every frame is rendered from code you write. GOAL Produce one finished episode of an interactive POV survival series: 40–45 seconds, 1080×1920, 30fps, ready to post on TikTok, Instagram Reels and YouTube Shorts. The viewer IS the main character. The episode ends on a two-option choice that viewers answer in the comments, and the next part continues from the most-commented option. INPUTS (ask me for anything missing, never guess) - Series disaster: [e.g. tsunami / earthquake / volcano / missile alert / blizzard] - Real-world location: [e.g. Miami Beach / Los Angeles / Tokyo], plus a fictional local landmark name: [e.g. "Marlin Beach"] - Part number: [1, 2, 3…] - Previous choice (Part 2+): [e.g. "THE HATCH"] - The two choices for the ending: [e.g. BOAT / HELICOPTER] - Recurring companion character name: [e.g. Jay], plus one or two side characters TECH STACK - three.js (pinned version) in a single HTML file, rendered headless with Playwright. - Deterministic timeline: expose https://t.co/IfGGJ8odPH(t) that sets every object, camera and effect for time t in seconds. No Math.random() at render time; use a seeded PRNG. No requestAnimationFrame-driven state. - Render loop: for frame 0..N, call seek(frame/30), screenshot the canvas at 1080×1920, then encode with ffmpeg (H.264, CRF 18, yuv420p, faststart). - Audio built separately (see AUDIO), then muxed with ffmpeg and normalized to -14 LUFS, true peak -1.5 dB. ART DIRECTION - Look: flat toon-shaded 3D, like a hand-painted cartoon. MeshToonMaterial with a 3-step gradient map, plus an inverted-hull black outline (2–3 px at 1080 wide) on characters and key props. No realistic textures, no PBR shine. - Palette: sky #9cc9e8 → #f3e6c8 at the horizon (golden-hour scenes add #f6b27a), sand #e8d3a6, ocean #2f7f86 with foam #d8f3ee, deep water #1f5a5e, warm interior wood #8a6a4a, metal #6d6f73, accent red #d8343a, safety yellow #f2c230. - City: low-poly glass towers (light blue, white window grid), pastel low-rise hotels and bars with simple fictional signs, palm trees as 3–4 flat leaf cards on a curved trunk, lifeguard tower in yellow and blue. - Characters: chibi proportions (head about 35% of height), round tan heads, large white eyes with small black pupils and a white highlight, thick dark brows that carry most of the emotion, a simple mouth shape per expression (O-shock, open shout, tight grimace), short brown hair as one toon mesh. Companion wears a red t-shirt and navy shorts. Mitten-style hands with a separate thumb. - Expressions change in whole poses (shock, shout, relief), never subtle lip sync. Mouths flap open and closed on dialogue syllables at roughly 8 Hz. CAMERA (THE MOST IMPORTANT PART) - True first-person POV. The viewer's own two arms are always visible from the bottom corners: tan skin, red sleeves to match a "you" character, mitten hands. The arms reach, grab, push and point on every beat. - Wide lens: FOV 95–105 with slight barrel distortion (a post-process pass, strength 0.15) so it feels like a head-mounted camera. - Head motion: constant subtle bob (0.6 Hz, 1.5 cm), panic sway when running, hard shake on impacts (decaying noise, 0.25 s, up to 4° roll), quick whip pans between beats (0.15 s ease-in-out). - Look up and look down a lot: up at the hatch, the wave and the helicopter, down at the ladder and the water. Vertical framing is the point. - Cut every 1.2–2.2 seconds. Each shot gets one clear action and one line of dialogue. WATER AND DISASTER FX - Wave: a large curved tube mesh with a toon water shader (two-tone teal with a moving foam band along the lip), plus white speed streaks that whip past the camera on impact. - Flooding interior: a water plane rising with a wobbling surface, white splash cards, and arcs of water spraying in from seams. - Underwater: teal fog (density about 0.08), slowly drifting light-blob caustics on the floor and ceiling, rising bubbles, a muffled low-pass on all audio, floating debris (cars, beach umbrellas, coolers, suitcases) drifting slowly with tumble. - Surface: horizon bobbing, a second wave building in the distance, debris on the water. - Rescue: a cartoon helicopter (white and navy, spinning rotor shown as a translucent disc), a rope ladder swinging, rotor downwash rippling the water. STORY STRUCTURE (every part) 1. 0–2 s HOOK: start mid-panic, no intro. The HUD shows location, time and the previous choice. Something is already going wrong. 2. 2–12 s ESCALATION: the plan from the last choice works at first ("It's holding!"), then fails ("It's coming in!", "It won't open!"). 3. 12–22 s WORST MOMENT: you're underwater or trapped. Dialogue stops. Sound goes muffled, then a hard silence of 0.5 s. 4. 22–34 s SURVIVAL: you surface, a friend calls your name, a new threat appears ("Another one's coming!"), rescue arrives with only one chance ("One pass!"). 5. 34–40 s CLIFFHANGER: you're reaching for the rescue as the next wave hits. Cut to black for 0.3 s. 6. 40–45 s CHOICE CARD (see END CARD). Hold it for 3+ seconds. Rules: dialogue lines of 1–4 words, shouted, present tense. Name a character at least twice per part so viewers learn them. Plant one small mystery that pays off later (for example: "Is that really Jay?"). ON-SCREEN TEXT - HUD, top-left, the whole video: "[LANDMARK] · [TIME]" in white, semibold, 28 px, plus "[City] · Part [N]" in yellow, 22 px, under it. - Previous-choice badge, top-centre, first 3 seconds: a red rounded pill with "YOU CHOSE" (small) over "[CHOICE]" (large, white, bold) and a simple icon. - Dialogue captions: centred at 72% of the height, bold rounded sans, 44–52 px, 3 px dark outline. Colour per speaker (you = white, companion = yellow, side character = purple or green). Shouted lines in ALL CAPS, normal lines in sentence case with "!". One line per shot, in on the first syllable, out on the cut. END CARD - Dark, blurred freeze of the last shot. "PART [N+1] · YOU DECIDE" in small yellow letter-spaced caps. "WHAT WOULD YOU DO?" in large white bold, 2 lines. - Two big rounded buttons: option 1 in blue (#2f6fe0) with an icon, "OR" between them, option 2 in red (#d8343a) with an icon. - "Comment below ↓" at the bottom. A gentle 1.03 pulse on both buttons. AUDIO - Voices: short shouted lines from 2–3 distinct voices (male, female, kid if needed) in a TTS tool, with energy and breath. Add panting under the running shots. - SFX, synthesized or from a free library: wave roar build, impact boom, metal hatch clang, water rush, underwater muffle with a heartbeat, surface gasp, helicopter rotor loop with doppler, distant sirens. - No music in the first 30 s, only a low tense drone. A soft rising swell under the end card. - Mix: voices sit 6 dB above SFX. Underwater section low-passed at 600 Hz. Export at -14 LUFS. THUMBNAIL / COVER (separate 1080×1920 PNG) - The companion's shocked face, hands on head, in the scene. - Two-line title in a heavy condensed font (like Anton) with a thick black outline: line 1 white, e.g. "TSUNAMI:", line 2 yellow, e.g. "MIAMI BEACH". - A small black-backed subtitle with the episode's choice, e.g. "HATCH OR BUILDING?". - A red "PART [N]" pill near the top. DELIVERABLES 1. episode.html: the full scene with https://t.co/IfGGJ8odPH(t). 2. render.js: the Playwright frame capture. 3. audio: voices, SFX and the final mix as WAV. 4. part[N].mp4: final 1080×1920, 30fps, about 42 s. 5. cover.png: the thumbnail. 6. caption.txt: one line + "comment BOAT or HELICOPTER" + 3–5 hashtags. 7. next.md: a 6-beat outline for BOTH possible next parts, so you can post within hours of the comments deciding. QA BEFORE EXPORT - Arms are visible in every first-person shot and never clip through objects. - No frame is static for more than 2 s. Something moves in every shot. - Every caption is readable on a phone at arm's length. - The end card is on screen for at least 3 s. - Watch it once muted: the story still makes sense from captions alone. DAILY SYSTEM Run this every day: read the top comment choice from yesterday's part, write the next part from next.md, render, cover, caption, and save into a folder named by date and part.

Questions

What is this films prompt for?

It is a Claude Opus motion design brief for “Interactive POV disaster short with cliffhanger choice card”. Paste it into Claude and keep the timings already written in the prompt.

Where is the original film?

The creator is @laurgrowth. Use View post to open the X video this prompt belongs to.

More films prompts