The cube between two portals
A one-minute physics thought experiment told through rapid narration and a continuously redrawn notebook diagram.
Field notes / why it works
The puzzle is spoken by 0:02.5 and the orange-cube-blue diagram is established in the first half-second, so the audience can start guessing immediately. Seven substantial drawing changes carry the argument from a simpler piston model at 0:07.6 through two assumptions at 0:21–0:34, with arrows supporting the unusual conclusion at 0:52–1:00. The 231-wpm narration keeps the single canvas active while the sparse diagrams make the dense reasoning followable.
Plate 01 / The format in one breath
Ask a counterintuitive “what if” question, draw its few parts on ruled paper, then replace the hard problem with a simpler physical analogy. The narrator rapidly tests alternatives as the same small set of doodles changes, saving the odd conclusion for the final seconds.
Why it works
- The question lands by 0:02.5; three possible outcomes follow by 0:07.3, making a clear open loop before the first major diagram change at 0:07.6.
- A single fixed notebook view keeps attention on the logic. Seven measured scene updates add new states at 0:07.6, 0:17.2, 0:20.8, 0:27.2, 0:34.4, 0:44.8, and 0:52.0.
- The two obvious cases occupy 0:23–0:34. The third case uses opposite-direction arrows and a familiar throw/catch analogy before the conclusion at 0:56–1:00.
Plate 02 / Format card
| Reference | |
|---|---|
| Platform · aspect · length | YouTube Short · 9:16, measured 202 × 360 at 24 fps · 60.0 s |
| Pace | One continuous composed scene; 0 hard cuts/min; 7 detected major picture changes/min; constant smaller drawing motion |
| Script | 229 spoken words · 231 wpm overall, 255 while speaking · 15 sentences averaging 15.3 words · 4 questions |
| Voice | Single low-register, quick, animated explainer voice |
| On screen | 100% top-down ruled-paper doodle animation; no visible presenter or live action |
| Captions | No speech subtitles; hand-lettered diagram headings and option labels only |
| Sound | Speech over a faint, continuous tonal bed (music-like, inferred); −13.9 LUFS integrated, 0.0 dBFS measured true peak |
| Structure | Question cascade → simpler model → constraints → alternatives → force analogy → strange conclusion; no spoken CTA |
Tools you'll need: Remotion for motion graphics · ElevenLabs or a recorded voice · Claude alone: yes
Toolkit for this film: Remotion for motion graphics; ElevenLabs or a recorded voice. Where the steps below name a single tool, use the toolkit instead.
Create a new 60-second, 1080x1920, 24-fps vertical explainer about {{TOPIC}} for {{AUDIENCE}} and, where relevant, {{PRODUCT}}. Use Remotion to render one locked top-down notebook canvas for the whole video: pale ruled paper, charcoal imperfect pen strokes, one orange element on the left and one blue element on the right. Use Patrick Hand or Caveat Bold for short handwritten labels; avoid full-sentence subtitles. Write an original 220–235-word script and use a curious, brisk, low-register narrator archetype, about 230 words per minute. At 0:00–0:07.6 pose a concrete what-if question and two or three plausible outcomes; draw the basic object within 0.5 seconds. At 0:07.6–0:21 replace the hard puzzle with a simpler test device and show its constraint. At 0:21–0:34 test two different assumptions with small redraws. At 0:34–0:52 reveal a causal turn using sparse direction arrows and an everyday analogy. At 0:52–1:00 show a qualified, surprising result. Keep the page fixed, with roughly seven major diagram changes and small stroke motion between them. Add a very quiet, neutral tonal bed only if it helps; keep narration clear near -14 LUFS and export true peak below -1 dBFS. Validate the reasoning under its stated assumptions. Use original drawings, wording, and sound.