Gen-4.5 visual world sampler
A cinematic AI video model launch that delays its name while 71 shots demonstrate visual range.
Field notes / why it works
The opening makes a visual proof claim by 0:01.5 over a moving waterfall scene, so viewers inspect the image before the first voice line at 0:07. After four slow shots, the 71-shot edit accelerates to 0.25–0.42-second cuts near 1:11, while 12–13-second speech gaps let the variety register. The product name arrives only at 0:58, and the 1:22 logo transition is the most replayed moment.
Plate 01 / The format in one breath
An 88-second product announcement behaves like a miniature film reel. A restrained voice makes three broad promises over a torrent of very different scenes, then names the product at 0:58; the final line at 1:20 implies more is coming. The promise is visual range and believable motion, shown before the specific launch claim.
Why it works
- The opening image arrives at 0:00.6: a lone figure against a huge waterfall, then a tiny centered disclosure around 0:01.5 says the images are model-generated. The claim becomes a test of the footage viewers are already watching.
- After a slow first 18 seconds, the edit accelerates: 71 measured shots in 88 seconds, median shot 0.79 seconds, with unrelated subjects and settings. The viewer keeps checking what the system can render next.
- The name waits until 0:58.4, after two silent montage passages (0:18–0:31 and 0:38–0:51). The most replayed interval, 1:21.9–1:22.8, sits at the transition from the final human close-up to the stark logo card; replay interest is measurable, though the reason is inferred.
Plate 02 / Format card
| Platform · aspect · length | YouTube · 16:9 · 1280×720 · 23.98 fps · 88.0 s |
| Pace | 71 shots, 70 measured cuts, 47.7 cuts/min; up to 66.1/min if all 27 uncertain fast changes were cuts. 50.5 measured picture changes/min, mean shot 1.24 s, median 0.79 s. |
| Script | 57 spoken words, six short statements, 46 wpm across the whole film and about 107 wpm during speech. No questions. |
| Voice | Low-register adult narrator, measured median 103 Hz, deliberate and declarative with animated emphasis. |
| On screen | Nearly all original-looking generated cinematic clips: human drama, animals, surreal action, everyday objects, macro nature. One text disclosure and a black logo end card. No presenter or UI demo. |
| Captions | No running subtitles. One tiny centered white disclosure over the first scene; clean white brand wordmark on black at the end. |
| Sound | Continuous tonal cinematic bed, roughly 110 BPM; sparse voice and no clear cut-synced effects. Measured −9.9 LUFS integrated, +1.1 dBFS true peak in the source. |
| Structure | Cold-open proof → veiled promise → two escalating clip reels → product name at 0:58 → last visual burst → postscript tease at 1:14–1:22 → logo. 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