Directs a software launch trailer end to end - the hook-vignettes-crescendo-endcard spine, a BPM-derived beat grid mapped to the track's phrases and drop that quantizes every cut, montage acceleration into the climax, anticipation holds before reveals, and reading-time floors on claim type - so the film builds energy instead of playing as a narrated slideshow. Use when someone asks "direct our launch trailer", "our product trailer feels like a slideshow - fix the pacing", "how many feature vignettes fit in a 60 second trailer", "the ending montage does not build to anything", or "how long should the end card hold", or when trailer scenes exist but the cut has no rhythm or climax. Do NOT use to plan a demo video's beat sheet and scene JSON - use video-storyboard instead; for writing the Remotion composition, use remotion-compose; for choosing, mixing, or beat-matching the music itself, use sound-and-music-sync.
---
name: launch-trailer-director
description: Directs a software launch trailer end to end - the hook-vignettes-crescendo-endcard spine, a BPM-derived beat grid mapped to the track's phrases and drop that quantizes every cut, montage acceleration into the climax, anticipation holds before reveals, and reading-time floors on claim type - so the film builds energy instead of playing as a narrated slideshow. Use when someone asks "direct our launch trailer", "our product trailer feels like a slideshow - fix the pacing", "how many feature vignettes fit in a 60 second trailer", "the ending montage does not build to anything", or "how long should the end card hold", or when trailer scenes exist but the cut has no rhythm or climax. Do NOT use to plan a demo video's beat sheet and scene JSON - use video-storyboard instead; for writing the Remotion composition, use remotion-compose; for choosing, mixing, or beat-matching the music itself, use sound-and-music-sync.
---
# launch-trailer-director
The default failure of founder trailers is the narrated deck: full-frame screenshots held dead-still, cuts landing between beats, energy flat from second 5 to the logo. Without a grid and a spine, every added scene makes the film longer, not better - and the one frame viewers were supposed to screenshot, the end card, gets 0.8 seconds. This skill is the director's chair: it produces the shot list and the beat grid before any animation code exists, so every downstream skill lands its work on a clock that already agrees with the music.
## Operating procedure
Run the steps in order: brief, then duration, then grid, then shots. The order is load-bearing - duration caps the vignette count, the track defines the grid, and every shot length snaps to the grid, so the grid must exist before any shot does. Cuts quantized after the fact drift half a beat and read arbitrary.
### Step 1: Gather the brief
Paste the charter block if one exists. If the client has a charter but the foundations pack is not installed, elicit the needed values directly and label them provisional. If no charter exists at all, halt and route to premium-design-foundations - never substitute worked-example values.
Then gather exactly these inputs. If any input is a guess, label it a guess and move on.
1. **Product and claim**: the product and the one launch claim the film must dramatize.
2. **Target duration**: default 60 seconds.
3. **Track**: the actual track, or a BPM preference (default 120 BPM if undecided - state it as a default, not a fact), plus the track's structure map: intro, build, drop, and the phrase heads of the edit.
4. **Features**: 3-5 candidates ranked by demo strength, not roadmap pride.
5. **Real UI availability**: whether real product screens exist to recreate. If not, stop and say so - a trailer cannot be directed around an interface that does not exist.