The 12-shot rule
A two-minute music video is roughly 12-16 shots at an average of 8 seconds each. Do not plan for one continuous narrative — plan for a tight shot list where each clip carries a single visual idea. This matches exactly how AI models generate best: short, self-contained beats.
Step 1 — pick your look
Before generating anything, produce three reference images in Flux 2 Pro that lock the palette, lighting and character look. Every shot in the video will reference these. Budget: 30 minutes, ~$1.
Step 2 — shot list from the audio
Listen through the track once. Mark twelve emotional beats with timestamps. Write a one-line shot brief per beat (subject / action / mood). Your shot list is now your prompt list.
Step 3 — generate hero frames
One frame per shot, generated in Flux 2 Pro or Nano Banana 2 with the reference images. Regenerate cheaply until each frame reads as intentional. Budget: 60-90 minutes, ~$8.
Step 4 — animate
Per shot, pick the right video model: Kling 2.5 for dance and motion, Veo 3.1 for controlled camera, Sora 2 for atmosphere and any diegetic sound. Image-to-video from the locked hero frame. Budget: 2-3 hours across generation and re-rolls, ~$20-40.
Step 5 — cut
Drop the clips into a NLE synced to the track. Trim to beat. Add colour grade LUT for cohesion across model differences. Add typography over-lays if the song calls for it. Budget: 45-90 minutes.
Total
Roughly 5-6 hours of focused work, $30-50 in generation credits, one person. Two years ago this was a four-day shoot with a crew of eight.
Start your first shot
Put this into practice in the studio — under a minute to your first result.
Start your first shot →