A Playable Ink Arena Match Built with Claude
A near-continuous browser-game capture turns live controls, territory change, setbacks and a timed result into proof of a playable AI-built prototype.
Starter prompt / by Refs
The creator shared no short prompt. This is ours for making a film like it with Claude.
Code a third-person ink turf-war game for the browser in three.js. Two squads of four, lime against magenta, get three minutes to paint as much of a plaza arena as they can: players fire ink, dive into their own color to swim fast and reload, and lob splat bombs, with bots filling empty slots. Add a pause menu that lists controls and the roster, a coverage tally at the end, chunky toon shading and splashy synthesized sound, then record two minutes of play at 3840×2160.
Field notes / why it works
At 0:00, the pause menu exposes the clock, teams and controls; the player resumes by about 0:01, immediately proving interactivity. The arena visibly changes color through a 2:07 countdown, with splats around 0:37 and 1:14 adding stakes before the time-up at 2:10. A 16.6-second celebration holds the outcome long enough to register.
Plate 01 / The format in one breath
A nearly continuous 16:9 recording starts on a pause menu with readable controls, enters a bright third-person arena, demonstrates movement and combat across a timed match, then shows the result. The promise is visible in the first second: this is a playable game with a timer, roster, controls, HUD and a real end state.
Why it works
- At 0:00 the pause screen shows the map, four-player teams, key bindings and a 2:07 clock. The viewer can understand the system before action begins.
- From 0:13 to 2:09, the floor visibly changes from magenta to lime as the player advances, is defeated, respawns and fights through the countdown. Progress and stakes stay on screen.
- At 2:10 the timer resolves with a large “TIME'S UP!” overlay; 2:11–2:27 holds on a team celebration. The long payoff verifies a complete match rather than a disconnected visual demo.
Plate 02 / Format card
| Platform · aspect · length | X · 1280×720, 16:9 · 147.7 s · 60 fps |
| Pace | 36 detected segments, 35 detected changes, 14.2 nominal cuts/min; mean 4.1 s, median 2.65 s. Many changes are gameplay motion, so this is largely a continuous capture. |
| Script | No spoken script or voiceover; on-screen HUD and game events carry the story. |
| Voice | None. The measured pitch figures are not usable because no speech was detected. |
| On screen | Almost entirely third-person browser gameplay; short pause-menu passages at the start, 0:41 and 1:21; splat/respawn views; end celebration. |
| Captions | No editorial subtitles. Game-native labels, counters, roster and large event messages only. |
| Sound | Near-silent recorded track: −51.2 LUFS integrated, −28.7 dBFS true peak. No reliable evidence of a music bed; 96 BPM estimate has low confidence. |
| Structure | Menu/context → play and movement → setbacks/respawns → faster final countdown → time-up → held result. No spoken CTA. |
Pro / 13 plates + recipe
The first plates are open. The full dissection is Pro.
Get the timed shots, type and sound specs, and the runnable recipe.
Open the full blueprint- Plate 03 / The hook, frame by frame (first 3 seconds)Pro
- Plate 04 / Structure (beat sheet)Pro
- Plate 05 / Shot grammarPro
- Plate 06 / Visual style guidePro
- Plate 07 / Voice and deliveryPro
- Plate 08 / Sound designPro
- Plate 09 / PackagingPro
- Plate 10 / Script templatePro
- Plate 11 / Production planPro
- Plate 12 / PromptsPro
- Plate 13 / Edit recipePro
- Plate 14 / QA checklistPro
- Plate 15 / Make it yoursPro
- Recipe / runnable promptPro