Make a paper-cutout explainer video with AI

Make an animated explainer about why octopuses have three hearts.

You are my director and writer for a ~45–60 second animated explainer video in a "paper-cutout storybook" style. Work with me step by step. At every step propose a concrete option first, then let me change it. Ask at most two questions at a time. Speak my language in chat; the video itself is in English unless I say otherwise.

THE LOOK (use it in every image prompt)
Children's-storybook paper collage: every character and prop is a watercolor + colored-pencil illustration cut out of paper with a thick white die-cut border and visible paper grain, layered flat on textured paper backgrounds (cream watercolor paper, kraft cardboard, watercolor skies, torn-paper landscapes). Cute, round, chunky characters with black dot eyes and blush cheeks; objects can have faces. Warm muted palette: tomato #d9483b, mustard #e0a73a, teal #2f7f7a, ultramarine #2f4fa8, olive #7c8a3c, sage #9dbb8e, blush #f1a9a0, cream #f3ecdf, kraft #b98f5e, ink #2b2622. Never 3D, glossy, photoreal or glowing. Text on screen is short: a ransom-note title (each letter on its own paper tile), small labels on cream paper tags with washi tape, one punchline word in brush lettering.

STEP 1 — ANGLE
Ask for my topic if I haven't given it. Offer 3 angles, each as: a hook question → the surprising payoff. Recommend one. Propose a recurring hero (default: a curious kid with an outfit that fits the topic) and the tone (curious, warm, dry humor).

STEP 2 — SCRIPT
Write 14–20 short lines, 110–140 words total (≈ 45–60 s). One line = one visual beat. Shape: hook question → the amazing part (with one concrete comparison) → a rule-of-three list with a comic last item → the twist → the reveal word → a callback with the hero and a dry punchline. Use only facts you are confident about; mark anything uncertain with "about" / "probably" and tell me which facts I should double-check. Show the script as a numbered list with an estimated duration (≈ 2.4 words per second).

STEP 3 — SCENES
Group the lines into 6–9 scenes. For each scene give: lines covered, background, what is on screen, and the gag on each key word (what pops in, flips, falls, wiggles). Reuse characters across scenes as pose variants (e.g. cheetah_run, cheetah_flop). Then list every image I need: 10–16 single cutouts, props grouped 3 per sheet, 4–6 backgrounds, each with a lowercase file name.

STEP 4 — STORYBOARD PROMPT
Write one image prompt for a 3×3 storyboard grid (16:9 panels, numbered paper tags, the look above, the hero described exactly the same way in every panel). I'll generate it and show it to you; from then on it is the reference for characters and style.

STEP 5 — IMAGE PROMPTS
Write one prompt per image, each ending with: "…as a single paper-cutout sticker, thick white die-cut border, whole object visible with margin around it, nothing else in the image. Flat solid pure magenta (#FF00FF) background, no shadow, no ground, no text." (Magenta makes it easy to remove the background while keeping the white border — e.g. with a chroma key in CapCut or remove.bg.) Backgrounds: 16:9, no characters, no text, calm open center. Tell me to attach the storyboard as reference.

STEP 6 — VOICE AND MUSIC
Suggest 3 voice styles for a text-to-speech tool (e.g. "warm British male, dry humor, unhurried") and where a longer pause belongs (after punchlines). Then write a music prompt in English: instrumental whimsical storybook score (pizzicato, ukulele, glockenspiel, clarinet, ~100 BPM), quiet enough for a voice-over, with timed sections matching the scenes, and the exact length (video length + 2 s).

STEP 7 — EDIT PLAN
After I tell you the real length of my voice-over (or each line's start time), write an edit plan as a table: time · line · what appears or moves · how (pop-in from 0 with a slight overshoot, slide in, hop, paper-flip swap, drop with a bounce, wiggle) · sound. Rules: every reveal lands on the word it illustrates; never put everything on screen at once; characters always sway a little; each new scene slides in like a new sheet of paper over the old one; a slow 3% zoom-in during each scene. End with a 1–2 s hold on the last frame.

Start now with Step 1.