Impossible edits inside a live footage demo
A presenter demonstrates AI footage replacements inside an editing timeline, then shows the plugin download page.
Field notes / why it works
The preview changes by 0:00.75, puts a new object into the pier scene around 0:01, and shifts the scene to night around 0:02.3 before the full verbal promise lands. Nineteen internal picture changes carry a sequence of distinct use cases across the first 38.8 seconds, while the visible editing timeline provides workflow context. A hard cut at 0:38.8 turns the proof into a legible install page and download action.
Plate 01 / The format in one breath
A fast, conversational presenter narrates an AI editing tool while a visible editing timeline plays changing examples above it. Each spoken use case gets an immediate visual result: a new object, altered environment, impossible scene, or text in real footage. The last 11 seconds show where to get the tool and end with a comment CTA.
Why it works
- The promise arrives by 0:05.4, while a red phone booth appears in the source scene around 0:01 and the setting shifts from day to night around 0:02.3.
- The presenter keeps a stable human anchor while the preview changes 19 times inside the 38.8-second opening screen-recording shot; viewers can compare claims with visible results.
- The payoff turns into a next action: the website and plug-in section take over at 0:38.8, with a visible download button around 0:44.5.
Plate 02 / Format card
| Reference | |
|---|---|
| Platform · aspect · length | TikTok · 720×1280, 9:16, 30 fps · 50.1 s |
| Pace | 2 detected outer shots, 1 cut (1.2/min), plus 19 internal picture changes; 23.9 total picture changes/min. Outer shot lengths: 38.8 and 11.3 s. The changing preview contains the effective visual cuts. |
| Script | 191 spoken words · 229 wpm overall, 243 wpm while talking · 9 sentences averaging 21.2 words · no questions · roughly 10 uses of “you” per 100 words |
| Voice | One medium-register, animated, conversational presenter voice; continuous speech |
| On screen | ~78% of time: editing interface with live-action preview, effects, and timeline. ~22%: product website with presenter overlay. These are screen-layout shares, not source-footage shares. |
| Captions | No persistent speech subtitles visible. Lime effect graphics and website headings provide selective text proof. |
| Sound | Original spoken audio over a continuous tonal, music-like bed; bed measured ~9 dB below voice. −12.5 LUFS integrated; −1.2 dBFS true peak. |
| Structure | Location and capability hook → escalating edit examples → product/Adobe association → install path → comment CTA. |
Tools you'll need: screen recordings of the product · a video model (Veo, Kling, Grok Imagine) · a camera shoot · After Effects for motion graphics · ElevenLabs or a recorded voice · Claude alone: partly
Frame by frame / measured
Stills taken at measured moments, each labelled with its time. Agents get the same sheets over MCP.
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