A Point of Light Becomes a Haunted World
An AI-made, four-minute song film grows from one light into glitching rooms, faces, and particle worlds.
Field notes / why it works
The opening holds near-black for 10.77 seconds before its first cut, letting a single light become a visual question; the vocal arrives at 0:12.68. Concrete images such as the night-vision bed at 0:24.6 and point-cloud house at 1:07.5 make the abstract lyric visible, while sub-second glitch bursts at 2:45–2:47 and 3:10–3:13 create late surges before a 19.3-second final hold.
Plate 01 / The format in one breath
A nearly black frame reveals one tiny light, then opens into a surreal, surveillance-like world while a first-person, high-register singer narrates an unsettling transformation. Instead of lyric captions or an explainer, each section turns the song's images into distinct visual systems: point clouds, thermal footage, night-vision rooms, wireframe houses, and glitching faces.
Why it works
- The first 3 seconds have zero cuts or picture changes; the tiny light creates a patient visual mystery. The first measured picture change is at 0:05.95 and the first cut at 0:10.77.
- The song begins at 0:12.68, and concrete scenes replace abstraction: a bed at 0:24.6, a nursery-like wireframe at 0:36.6, and a house at 2:31–2:54. The imagery keeps revealing what the voice implies.
- The edit reserves its fastest bursts for escalation: five sub-second cuts at 2:45–2:47 and six more at 3:10–3:13. A 19.3-second shot at 3:23 then gives the final idea room to land.
Plate 02 / Format card
| Platform · aspect · length | X · 16:9 · 1280×720 · 30 fps · 3:56.2 |
| Pace | 66 detected shots; 65 cuts, or 16.5/min. With 23 uncertain changes, possible range 16.5–22.4 cuts/min. 31 measured in-shot steps; 24.4 confirmed picture changes/min. Mean shot 3.58 s; median 2.23 s. |
| Script | 319 recognized lyric words; 88 words/min across the whole video, 108 while the vocal is active. Automatic transcription is approximate for singing. |
| Voice | One high-register, melodic first-person singer; intimate at first, more insistent during repeated phrases (delivery inferred from timing and pitch measurements). No spoken presenter. |
| On screen | Entirely designed or processed imagery in the supplied frames: black-field particles, wireframes, grainy nocturnal interiors, thermal and surveillance views, and glitch graphics. No visible talking head. |
| Captions | No lyric captions in the sampled frames. Tiny technical camera overlays appear in some shots. |
| Sound | Continuous tonal music/song bed. −15.6 LUFS integrated, −3.8 dBFS true peak; roughly 129 BPM, low-confidence estimate. No clear SFX pattern on cuts. |
| Structure | 7 broad beats: darkness → world reveal → first refrain → intimate system view → second refrain → glitch escalation → sparse final statement. No visible call to action. |
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