Break this beat into a shot list for [MODEL]: BEAT: [BEAT] TONE: [TONE] TOTAL LENGTH: [LENGTH] seconds For each shot give me: - Shot number and duration - Shot size and camera movement - A one-paragraph prompt written the way a text-to-video model wants it: subject, action, camera, lens, light, location - What must match the previous shot for continuity — light direction, wardrobe, screen direction Rules: - Do not cross the line between shots unless you flag it. - Vary shot size between consecutive shots, or the cut will feel like a jump. - Keep each shot to one action. Two actions in one generation is where these models fail. End with the shot most likely to need re-rolls and why.
One shot described with subject, camera move, lens and light — the four things video models read.
A short-form product video structured so the value lands before anyone scrolls away.
Ten b-roll prompts that share a look, so the cut does not feel stitched together.