Start with a song

Videos usually start with footage. They don’t have to — hand the agent a track and let the music dictate everything: structure, pacing, and where every visual lands.

What you need

• A music track. Everything else is optional — b-roll, photos, or nothing at all (the agent can generate imagery to fill).
• A connected agent.

What to say

Build a 30-second promo around track.mp3 — map the structure of the song, keep it minimal until the drop, then let it fly.

Or for lyrics:

Make a lyric video for this song — here’s the lyrics file, big kinetic type, one line per phrase.

What happens

Mixture detects the beats automatically; the agent maps the track’s structure — intro, verses, the drop — and builds scenes against it. Reveals land on beats, cuts land on phrases, and energy builds with the song instead of opening at maximum. For lyric videos it times each line as kinetic type; with footage it cuts on the grid; with neither it generates imagery to carry the visuals.

Make it yours

The beat grid is visible on the timeline — drag any moment to a different beat, swap a section’s visuals, or ask: “the drop should hit harder — bigger type, faster cuts.”