Skip to content
refs
REF-2177Source onlineBlueprint rev 01

From Prompt to Published App

A kinetic product launch film that moves from a broad creative promise to a readable prompt, generated result, refinement, and publishing.

Maker @GoogleDeepMindFor Google DeepMind / Google AI StudioOriginal post ↗
Cut timeline · 5 shots01:08
00:0000:1000:2000:3000:4000:5001:00

Field notes / why it works

The 0:00–0:03 word build resolves into a wide collage, making an abstract creative promise immediately visual. A readable request at 0:28–0:30 leads to a game-like result by about 0:34, then refinement and publishing actions extend the proof through 0:56; the request area is also the strongest replay peak. Only four hard cuts occur in 67.9 seconds, but 17 internal graphic changes keep the film moving.

01

Plate 01 / The format in one breath

A polished product launch film turns a plain-language prompt into an expanding gallery of things people could make. Large, sparse kinetic type names each capability; floating app examples and short interface close-ups supply proof. The film finally moves from creating to refining, sharing, and publishing.

Why it works

  • The visual promise appears by 0:02: the opening words build to “Bring anything to life,” then a collage of varied results makes the abstraction concrete by 0:03–0:04.
  • The sparse five-shot structure still creates 17 internal step changes, about 18.6 picture changes per minute. New words, cards, and interface states carry the rhythm without frantic cutting.
  • The most-replayed area, 0:29.2–0:29.9, lands on a legible natural-language request and the start of app generation: the exact point where the promise becomes a demonstration. This is an association, not proof of why viewers replayed it.
02

Plate 02 / Format card

Platform · aspect · lengthYouTube · 16:9 · 640×360 analysis copy · 67.9 s · 23.98 fps
Pace5 detected shots; 4 cuts / 3.5 cuts per minute; 17 internal steps; 18.6 total picture changes per minute; mean shot 13.58 s, median 9.65 s. These long detected shots contain many designed scenes.
Script54 transcribed words; 52 wpm across the full film, about 134 wpm while speaking; 12 short phrases, about 4.5 words each. Automated transcript has obvious recognition errors, so use timing and structure rather than its wording.
VoiceOne low-register, animated, polished narrator; short assertive phrases separated by long music-led gaps. No visible presenter.
On screenFull-frame motion graphics and kinetic type, product UI close-ups, floating examples, one map-like racing-game output. No talking head.
CaptionsNo verbatim subtitle stream; large centered keyword/headline cards. Select words turn blue.
SoundContinuous tonal, music-like bed (inferred electronic/ambient feel); no measured cut-synced impacts; −15.6 LUFS integrated, −0.9 dBFS true peak.
StructureOpening promise → possibility gallery → prompt/demo → refine → share/remix/publish → final promise. Publishing is the CTA at 0:55.9; final line lands near 1:03.

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
  1. Plate 03 / The hook, frame by frame (first 3 seconds)Pro
  2. Plate 04 / Structure (beat sheet)Pro
  3. Plate 05 / Shot grammarPro
  4. Plate 06 / Visual style guidePro
  5. Plate 07 / Voice and deliveryPro
  6. Plate 08 / Sound designPro
  7. Plate 09 / PackagingPro
  8. Plate 10 / Script templatePro
  9. Plate 11 / Production planPro
  10. Plate 12 / PromptsPro
  11. Plate 13 / Edit recipePro
  12. Plate 14 / QA checklistPro
  13. Plate 15 / Make it yoursPro
  14. Recipe / runnable promptPro