From a hand-drawn floor plan to a shoot day that works

Every location scout does it: walks the space, sketches the floor plan on a notepad — walls, doors, windows, "camera here," an arrow for the dolly move, a note that the fridge hums. That sketch is the most information-dense document in pre-production. It's also the least usable one, because it lives on paper in one person's bag.

The gap between the sketch and the shoot plan is where shoot days go wrong.

What the sketch knows that the call sheet doesn't

A good location sketch captures: room dimensions (paced out, close enough), door swings and openings, window positions and light direction, power outlet locations, camera positions with lens notes, movement paths, sound problems, and the producer's margin notes ("neighbor has a dog — shoot before 10am").

Almost none of this survives transcription into production software, because transcription means someone retyping it all — so nobody does, and the crew works from a photo of a sketch in a group chat.

Photo-to-planner: the honest version

The idea is simple: photograph the sketch, get back structured data — rooms with dimensions, doors and openings, camera positions, notes — as planner JSON the shoot-planning tools can consume. No CAD, no redrawing, no transcription.

The honest limits, because this isn't magic:

• Legible beats pretty. The decoder reads what a human could read. A sketch that's clear to the scout is clear to the pipeline; a sketch that's cryptic to everyone stays cryptic.
• Interiors and defined spaces. Floor plans have walls. Open locations — a field, a beach, a street — don't decode into rooms, because there are no rooms. The tool says so upfront instead of hallucinating walls.
• Review is part of the workflow. The output is structured data you check: a misread wall is one field edit. The human pass isn't an admission of failure — it's the design.

What structured unlocks

Once the sketch is data, the shoot plan builds itself:

• Blocking references actual room dimensions instead of "the big room"
• The shot list ties camera positions to the plan, so the AD knows the company moves at 2pm, not "after lunch"
• Equipment gets specced to the space — the dolly track length is a number from the plan, not a guess
• The day's order follows the geography: shoot out one room completely before moving, because the plan shows the company move costs

None of this is new wisdom. It's just wisdom that used to die in the scout's notebook.

The workflow that works

1. Scout sketches on location — paper, pen, pace out the dimensions
2. Photograph the sketch before leaving (lighting matters; flat lay, no shadows)
3. Decode to planner JSON; review and correct the fields that matter
4. Feed the shoot planner — blocking, shot list, and company moves inherit the geometry
5. On the day: the plan matches the room, because the plan came from the room

The napkin sketch was never the problem. The problem was that it stayed a napkin. Make it data, and the shoot day gets shorter.