Every shot list I build gets thrown out partially within the first hour of an actual shoot day, weather changes, a location falls through, a client changes their mind about a key message halfway through. I still build detailed shot lists for every single project, AI-assisted work included, because the value was never in following it exactly.
The shot list is a rehearsal, not a script
Building a detailed shot list forces me to think through coverage, transitions, and pacing before a single frame gets captured, which means when reality forces a change, I'm improvising from a position of already understanding the full shape of the piece rather than improvising blind. The list itself is disposable. The thinking that produced it isn't, and that thinking is what actually survives contact with a chaotic shoot day.
How this changed for AI-generated shots specifically
For AI video generation, a shot list now includes prompt notes alongside traditional shot descriptions, camera move, subject action, and a rough draft of the language I expect to use to generate it. Writing that out ahead of time surfaces gaps early, a shot that sounds simple in a one-line description often reveals it needs three or four specific details once you try to actually write the generation prompt for it. Better to discover that gap during planning than during a live client review.
- Build the list for the thinking it forces, not for rigid adherence
- Include draft prompt language for any AI-generated shots in advance
- Treat gaps that surface while writing the list as information, not annoyance
What actually changes on shoot day
I've stopped feeling any guilt about abandoning half a shot list once conditions change. The list did its job the moment it forced me to think the project through in advance. Everything after that is just responding to what's actually in front of me with the benefit of already having done the hard thinking once, calmly, at a desk, instead of trying to do it for the first time under pressure.