Handwritten Shot Lists vs an AI Shot List, Where the Line Sits

An AI shot list earns its keep in a narrow band: one idea in, one shoot day, one camera crew, and roughly five to fifteen shots. Outside that band the output stops being a plan and turns into a rough sketch you argue with. I've run this through about forty projects in Open Studio as of November 2026, and the line has held every single time. The point I keep circling back to: scalability only counts once you've actually scaled, and I've watched a generated list fall apart at both edges of that band.

Forty projects is the sample size.

How many shots can one camera day cover?

This is where it genuinely works, and I'll give you the real numbers. In late October 2026 I fed a single product concept for a coffee grinder into Open Studio and got back eight shots, each with a duration, a camera move, and paste-ready natural-language prompts I could drop straight into Veo 3 or Kling. I didn't register first; as a guest I had a daily free generation allowance, and I only created an account afterwards to lift the cap and save the project.

The operation log tells the rest: at 14:20 shot 7 came back wrong, a three-second push-in rendered as a static hold, and the retry at 14:26 fixed it on the second attempt. One retry in eight shots. The full retry story is in my retry log from that session. I also pulled two templates from the Prompt Library, and the library versions cut my retries in half.

I have a habit I can't shake: I still print every list and scribble timings on the margin, even when the generated one is right.

At what point does the drift start with twenty-five shots?

Push past fifteen shots and the problems start showing up in the timeline, not the prompts. On a 25-shot explainer video in early November 2026, shot durations began overlapping, the same dolly move appeared three shots in a row, and the negative prompts contradicted each other between adjacent shots (one said 'no handheld', the next asked for 'handheld energy'). So where does it actually crack? For me it was shot twenty-two: the timings no longer lined up with the edit, and I spent forty minutes re-timing four shots by hand to tighten up the pace. A mate in a creator Discord told me the opposite experience, that thirty shots worked fine for them, but their shots were static talking heads on Kling while mine carried real camera movement through every take.

The edge isn't a fixed number, which is exactly why people disagree about it. It depends on how much motion you ask for and whether your model of choice follows the movement notes. If you want to see how I stress-tested this, my 12-take storyboard test covers the motion-heavy side.

How do multi-day shoots work with a second unit?

This is outside the boundary, full stop. A generated list assumes one continuous scene, but a two-day shoot with a B-unit breaks that assumption at the foundation. Shot numbering collides between units, durations don't map to anything on the edit timeline, and the prompts read like one location when you're actually shooting in two. I tried it once on a brand film spread across three days in September 2026. The list couldn't tell me which unit covered what, and I ended up rebuilding the whole schedule in a spreadsheet anyway.

It broke exactly where a shot list is supposed to hold: tracking what gets shot, when, and by whom. For anything that size I go back to a storyboard-first workflow and treat the generated output as reference images only.

How to tell which zone you're in

How do you know which side of the line you're sitting on? Count three things before you generate: total shots, camera units, shoot days. Five to fifteen shots, one unit, one day, you're inside and the list is a real plan. Fifteen to thirty with heavy movement, you're on the edge, so budget time to re-time the cuts yourself. Multi-day or multi-unit, you're outside and the list is only a mood board with numbers on it.

People keep asking me whether the free tier covers this, and honestly for one short video it does: as a guest you get a daily generation allowance with no registration, and registering just raises the allowance and lets you save projects. What it isn't is unlimited; I've hit the cap mid-project twice this year, both times on edge-zone lists I kept regenerating. Here's the comparison I keep taped inside my notebook:

For your next short, give it a crack in Open Studio while you're still under fifteen shots, and check every duration against your edit before you trust the list. The guest quota isn't chocka, but it's enough to find out which zone you're in.