AI in Pre-Production: From Script to Storyboard

Pre-production is where AI quietly saves the most money. Long before a single frame of video is generated, the costly, slow work of breaking down a script, designing characters, and building a board can now be compressed dramatically. The catch is consistency: do it carelessly and every frame invents a new face. This workflow is built around solving that.

The shift in pre-production

Traditionally, storyboarding meant either an expensive artist or rough stick figures. AI offers a third path: photoreal, consistent boards generated from your own script — provided you treat the model like a department that needs a precise brief, not a magic wand. The single discipline that separates professional results from slop is character consistency orchestrated in language first, then enforced with reference tags.

The six-step pre-production workflow

1. 01

   Break the script into shots with Claude

   Paste a scene into Claude and ask it to return a numbered shot list: shot size, subject, action, and a one-line visual description per beat. This converts prose into a board-ready structure in seconds and forces you to think in coverage, not paragraphs.

2. 02

   Write a reusable character bible

   For each character, lock a fixed description block — age, build, wardrobe, distinguishing features — and reuse it verbatim in every prompt. Consistency across frames starts here, in language, before any image is generated.

3. 03

   Draft the hero reference image

   Generate a clean, front-lit character portrait in Midjourney at --ar 2:3. This becomes your reference asset. Keep the background plain so the model fixes on the face and wardrobe, not the environment.

4. 04

   Lock style with seeds and --cref

   Use Midjourney --cref with your hero image and a moderate --cw value to carry the character across new compositions, and reuse a fixed --style seed so lighting and rendering stay coherent board to board.

5. 05

   Add cinematic markers to every frame

   Append film-grammar to each prompt: aspect ratio (--ar 16:9), lens, lighting direction, and camera angle. A board that already speaks in lenses and lighting hands a clear brief to your generative video stage.

6. 06

   Assemble the director board

   Drop the frames into Figma or Google Slides in shot order with the shot-list metadata beneath each. You now have a coherent, consistent board built in an afternoon instead of a week.

The cinematic prompting blueprint

A storyboard frame is a shot brief in disguise. Every prompt should carry six elements: subject, action, camera angle, lens, lighting, and environment. Adding film stock and era references ("Kodak 35mm," "Alexa 65 clean digital") pushes the look from generic render to intentional cinematography. We document the full vocabulary in the cinematic prompt guide.

Maintaining consistency across perspectives

The hardest part of boarding is showing the same character in identical costume from different angles. Midjourney's --cref carries identity, but clothing still shifts. Generate the wide, the medium, and the close from the same reference and seed, then retry the outliers rather than accepting a costume change. The deeper mechanics — character weight, seed control, and porting these assets into motion — are covered in our consistent characters tutorial.

From board to motion

A finished board is not the end — it is the input to the next stage. Those locked reference frames become the starting images for image-to-video generation, where character continuity carries forward. Continue into the generative video workflow to turn these stills into directed motion.