Section 2: Generative Scripting and Scene Design · Lesson 5 of 5 (10 of 45) · 60 minutes · Evidence: Scene and asset breakdown
A script is a story. A production is a shopping list. Before a single image is generated or a voice recorded, someone has to read the script and ask: what exactly will we need to make? Miss something now and you find out in the middle of Section 5, when your hero is on screen and you have no picture of her face from the side. Today you read your script like a producer.
Production journal: write your answer below.
What must a filmmaker understand before they can convert the script into scenes shots voice lines images motion and postproduction requirements?
A scene is a continuous piece of story in one location and time. When the place or time changes, the scene changes. A shot is one continuous view of the action from a single camera setup, whether it is filmed or generated. One scene usually needs several shots: a wide view that shows where we are, a closer view of the character, a detail of something important.
An asset is any individual piece of material that must exist before the film can be finished. For an AI-assisted short film, assets are mostly:
A dependency is an asset that other assets rely on. If Mara’s reference image changes after you made forty shots of her, all forty shots may need to be redone. So the order matters: approve character references and the main location first, then props, then shots, then voice and music timing, and finally the edit. Mark each asset with the asset it depends on.
Coverage means having enough different shots of a scene to edit it well: a wide shot, a medium shot, a close-up, and a cutaway. Plan coverage deliberately. If every shot is a close-up of a face, the viewer never learns where they are.
A deliverable is a finished item the team promises to hand over, such as the final film file, captions or a credits page. List yours so that nothing required is forgotten.
Scene 3, from a mystery short: “INT. LIBRARY, NIGHT. A student finds a torn map inside a returned book and hears footsteps.” Characters: the student. Location: library, night. Props: torn map, returned book. Voice lines: one whispered line, one narration line. Shots: wide of the library, medium of the student opening the book, close-up of the map, over-the-shoulder looking down the aisle. Dependencies: all shots depend on the student’s reference and the library reference. Flags: the map must be an original drawing, not a copy of a real one.
Production coordinators, animators and editors use some form of breakdown, no matter how large the project. It is a table: one row for each scene or shot, and columns for everything that scene needs. It looks dull, and it saves enormous amounts of time.
A creator reads the script slowly, one scene at a time, and highlights every noun that must be seen and every sound that must be heard. They then group the assets by type, find the ones that appear in several scenes, and put those first in the schedule. A text generator can help extract a first list from the script in seconds, but it usually misses small props and invents details that were not in the text.
The decisive human step is flagging: for each asset, asking where it will come from, who owns it, whether anyone has to give permission, and whether the film must say that AI helped make it. Those answers cannot be guessed by a tool.
As you study, record each decision point in your production journal.
| Decision point | What the filmmaker does | What the tool automates | What still needs human review |
|---|---|---|---|
| Break the script into scenes | Splits the script wherever the place or time changes and numbers each scene | Can suggest scene splits from the text | Whether every scene is needed and the total still fits the beat sheet |
| List the assets | Highlights characters, locations, props, voice lines and sounds in each scene | Can extract a first list of nouns | Whether small but important props and sound needs were missed or invented |
| Order by dependency | Marks which assets depend on others and approves shared ones first | Nothing; it needs knowledge of the project | Whether the order avoids redoing work if something changes |
| Flag permissions and disclosure | Writes down source, owner, permission and disclosure needs for each asset | Cannot know your rights or releases | Whether every asset has a legal and policy-safe source |
Studio task: Break one sample scene into required assets and dependencies.
Your teacher will give you a one-scene sample script. If none is provided, use this one:
“INT. KITCHEN, MORNING. Zoe sets a cracked blue teapot on the table. Her grandfather stares at it. GRANDFATHER (quietly): You kept it. Zoe opens a drawer and takes out a small glue bottle. Rain taps on the window.”
Coaching tips. If your shot list is all close-ups, add one that shows the whole room. If you are unsure whether something is a prop or a location detail, ask whether it is used or changed by a character. Discuss with a partner which asset would be most expensive to redo if it changed.
Scene # ___ Heading: ______ Beat(s): ______ Screen time: ___ s Characters (with reference): ______ Location and time of day: ______ Props: ______ Voice lines (who, line, tone): ______ Sound and music: ______ Shots (number, framing, what it shows, motion): ______ Depends on: ______ Approve first: ______ Flags (source / permission / disclosure / accessibility): ______
Studio task: Complete the production breakdown for every scene in the final script.
Use the versioned final script from Lesson 2.4. Produce a complete scene and asset breakdown for the whole film.
Model prompt for the optional extraction: “Read this scene and list only the characters, locations, props, spoken lines and sounds that are written in it. Do not add anything that is not in the text. Put them in a table.”
Checklist before you submit: every scene has a row; the master list is ordered by dependency; every asset has a source and flag; deliverables are listed; my file follows the course naming convention.
Name the asset that must be approved first and why.
Deliverable: Scene and asset breakdown.
Submit the scene and asset breakdown. The file must use the course naming convention and include enough process evidence to show the student's decisions and revisions.
File name (course naming convention): AIF_S02-L05_scene-and-asset-breakdown_v01_YOURINITIALS
| Criterion | Advanced | Proficient | Developing | Beginning |
|---|---|---|---|---|
| The work addresses the stated story or production purpose. | The breakdown covers every scene and every asset type, with planned coverage that fits the script and the target time. | The breakdown covers every scene and lists characters, locations, props, voice lines and shots. | Some scenes or asset types are missing, or shots are not specified. | Only a list of scenes, or an incomplete breakdown. |
| The student applies the required method accurately and produces usable capstone evidence. | The master list is ordered by dependency, repeated assets are merged, and the approval order is justified. | A master list exists with dependencies and an approval order marked. | Dependencies are listed for some assets only. | No master list or no dependency thinking. |
| The student identifies at least one quality issue and records a meaningful revision decision. | Names what the extraction or first pass missed or invented, and records a specific correction. | Identifies one quality issue and records a meaningful revision decision. | Notes a problem but no clear change. | No revision decision recorded. |
| The work complies with consent rights attribution privacy disclosure and school policy. | Every asset has a documented source, permission and disclosure flag, with accessible alternatives such as captions planned. | Sources and flags are recorded for all assets and school policy is followed. | Source or permission records are missing for some assets. | Assets with no source, or relies on unlicensed or real-person material. |
Requirement: Flag every asset that needs permission attribution or disclosure.
Your breakdown is where responsible production becomes concrete. Give each asset a source: you made it, you generated it with your school-approved tool, it is licensed (note the license and the required credit), or a consenting person released it in writing. Flag any asset that shows a real person, a real place that identifies someone, a recognizable brand, or a voice. Flag anything that will need an on-screen disclosure that AI helped create it. Plan captions or a written text version for every voice line, and a no-cost or school-approved way to make each asset if a service is unavailable. Unresolved flags must be solved before the asset is created, not afterward.