Manus Image Generation: Four Workflows, One Launch
A feature launch that escalates from four task-to-result demos to a broader agent workflow claim.
Field notes / why it works
A fixed editorial title gains visual range as framed examples swap in the first 3 seconds, before the first detected cut at 5.9 seconds. Four jobs start around 0:06, 0:25, 0:40 and 0:49, each moving from a product task toward a concrete output; the closing 1:00 reframe ties those examples into a broader workflow. The 36.4-second final detected shot still has eight measured internal changes, so the finish continues to evolve before the quiet logo hold.
Plate 01 / The format in one breath
A feature announcement begins with a large editorial title and a rapid carousel of possible results. It then proves the broader promise with four different user jobs: a task in the product UI becomes a tangible visual output, then the next job raises the scope. The closing line reframes the feature as part of a complete workflow and ends on the brand mark.
Why it works
- The title occupies the full opening 5.9 seconds, while the example card visibly swaps at about 0:00.5, 0:01.25, and 0:02.25; the viewer understands “many uses” before the first cut.
- Four applications arrive in succession: room furnishing (0:06–0:25), tea launch (0:25–0:40), online shop (0:40–0:48), and illustrated story (0:49–1:00). Each pairs process with a result.
- At 1:00 the narration broadens the claim from generated pictures to coordinated tools, with small tool icons gathering around a browser mockup at 1:04–1:08. The brand mark gets a quiet hold around 1:17–1:20.
Plate 02 / Format card
| Reference | |
|---|---|
| Platform · aspect · length | X · 1280×720, 16:9 · 80.6 s · 30 fps |
| Pace | 11 detected shots, 10 detected cuts = 7.4 cuts/min; possible range 7.4–9.7. Mean shot 7.32 s, median 4.0 s; 17 detected in-shot changes, about 20.1 detected picture changes/min total. The visual card swaps in the opening evade that detector. |
| Script | 162 transcribed words · 132 wpm over runtime, 186 wpm while speaking · 18 sentences averaging 9 words · two questions |
| Voice | One clear, animated, mid-register English narrator; no on-camera speaker |
| On screen | Editorial title motion graphic, then product screen recordings and finished-result images/browser mockups; no presenter or live-action face |
| Captions | No speech subtitles. Title, UI copy, result labels, and final statement carry the text. |
| Sound | Continuous tonal, music-like bed (inferred), roughly 8.4 dB below voice; no reliable cut-synced effects |
| Structure | Title → four use cases → agent-level synthesis → logo. No explicit 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