Before you start
Helpful first: Make a short AI video with a clear purpose (45 min)
A chat assistant, an approved reference image and a short original story. Image/video generation is optional during planning.
Why this lesson exists
The archived posts behind “I Spent More Time Building the Music-Video Machine Than the Music Video” and the AIMVDashboard project treat prompts as production assets. Continuity comes from explicit references, shot purpose and review, not a longer list of adjectives.
The idea this guide practices: I turned a weekend into a Phantasy Star music video. Prefer a story? Read the comic Black Belt With Wi-Fi.
Do the exercise
Write a compact canon
The canon is the fixed facts every shot must keep. Record the character’s appearance, wardrobe, location, palette and objects that must not change. Separate these constants from the action of each shot.
Describe three shot intents
For each shot define the story purpose, opening state, one action and ending state. Make the ending compatible with the following shot’s opening.
Create standalone prompts
Combine the canon and shot intent into each prompt. Include the reference to use, camera framing and continuity checks. Avoid “same as before” because another tool may not know before.
Approve references before motion
Review still frames for identity, hands, props and composition. Save accepted versions and actual file names. Generate motion only after the sequence reads correctly as stills.
A prompt to adapt
Replace the bracketed parts with your own details.
Create a three-shot production pack for [story]. Canon: [character/location/style]. For each shot give purpose, opening state, one action, ending state, reference file and a standalone prompt. Add a continuity checklist. Do not schedule paid generation until the reference frames are reviewed.
What this looks like
If a red notebook is in the right hand at the end of shot one, shot two should not begin with a blue tablet in the left hand. Put that handoff in both prompts.
Check your result
Use evidence from your output. A confident explanation from the AI is not enough.
If it isn’t working
If one detail drifts, request a targeted correction instead of regenerating the entire sequence. Check the actual tool’s supported inputs; a storyboard is not always treated as a first-frame image.
Optional. Progress stays in this browser.
Where this came from
Adapted from the archived LinkedIn theme “I Spent More Time Building the Music-Video Machine Than the Music Video.” See the post coverage and editorial method. One of the original LinkedIn posts ↗
Prepared September 2026. Tools and interfaces change; use current official setup instructions. Session lengths are estimates.
patrickz