An AI film treatment template is a 14-section, roughly 25-page directive document that encodes camera, lighting ratios (e.g. 85:15 dark-to-light), color, a 9-element prompt order, negative prompts, and a 12-parameter per-shot spec as machine-readable rules. Write rules not descriptions, fill it once, and load it into the invideo agent, which gates every shot against it.
A film treatment template for AI generation is a 14-section, roughly 25-page directive document that encodes your camera grammar, lighting ratios (e.g. an 85:15 dark-to-light ratio), color modes, a 9-element prompt assembly order, negative prompts, and a 12-parameter per-shot spec as machine-readable rules. Fill it once, load it into the invideo agent, and the agent gates every generated shot against it — no re-prompting, no drift.
What a treatment template for AI generation must contain
A treatment template for AI filmmaking is not a synopsis — it is a rule system the invideo agent can execute. invideo is an agentic video creation tool, and the agent holds whatever you load into its context across every shot, so the template must cover everything you would otherwise re-explain per prompt: camera and lens behavior, angles, color tone, atmosphere, mood, lighting, composition, movement, reference film palettes, a prompt template, negative prompts, and a quick-reference card. One documented 25-page treatment was uploaded as a permanent instruction set before a single frame was generated; the agent then output 12 key parameters per shot — film reference, shot design, length, style interpretation, emotional register, lens, lighting plan, color script, atmosphere layers, blocking, final prompt, negative prompt, and revision prompt.
The template: the 14 sections, ready to fill
Fill each section as directives, not prose:
- Camera — body behavior, movement defaults, when the camera holds vs. moves
- Lenses — spherical vs. anamorphic, bokeh shape, flare rules
- Angles — preferred heights and framings, forbidden angles
- Lighting — ratios and sources
- Color tone — named tonal modes (e.g. "Mode A — split-toned amber and emerald")
- Composition — framing rules, negative space, what to withhold
- Movement — blocking logic, how subjects traverse the frame
- Atmosphere — haze, weather, texture layers per scene type
- Mood / emotional registers — named stages with locked camera, lighting, and sound rules per stage
- Reference film palettes — the specific films the grammar derives from
- Sound — what is heard before what is seen, diegetic sound logic
- Prompt template — the 9-element assembly order: camera spec, lens & aspect ratio, lighting source, palette, composition, atmosphere, mood register, film/DP attribution, negative prompt
- Negative prompts + never-do rules — explicit prohibitions, plus exceptions kept in their own directive so general rules don't get misapplied
- Quick-reference card — the one-page compression the agent checks fastest
Write rules, not descriptions
Every line must be enforceable. "Warm lighting" is a description; "warm yellow from the lamps only, like all the refs" is a rule the invideo agent can check a frame against. Encode color as named modes, lighting as ratios, and prohibitions as explicit negatives — one animated production's style block read "not live action, not photorealistic" on every prompt to prevent drift. Including a what-never-to-do list per emotional stage makes it significantly easier for the agent to make autonomous decisions across hundreds of generations.
Include sound, even though nothing in the template plays audio
Sound rules shape the visual output. In one horror-style production, the agent's analysis concluded that half of what makes the reference director's films work is what you hear before what you see — encoding that rule changed how shots were designed and held. Sound logic also feeds asset briefs (a prop specified as "hard material that makes a horrible sound when it falls" generates differently) and gives the agent grounds to flag SFX and pacing problems at the cut-review stage.
How the template gets used: one load, then gating
Upload the finished document to the invideo agent once at project start; it stays loaded across every frame, scene to scene. From then on the agent gates output: "You write the direction. [The invideo agent] builds the shot, holds it against the treatment, and only sends back what passes. Every frame is a decision, not a draft." With context loaded, a three-word continuation prompt — "Everything should match" — is enough to hold character, lighting, lens grammar, and spatial logic across a multi-shot sequence. Validate the load before production: ask the agent to apply the style to a genre the source director never worked in; clarifying questions and stylistically coherent output confirm it internalized grammar, not surface style. Documented results: a 70-second film produced this way ran $750.
Do you fill the whole template before generating anything?
No — start with whatever rules you can state confidently and iterate; perfection upfront is not required. But lock the load-bearing sections before spending generation credits: lighting ratios, negative prompts, and the 9-element prompt order, since every prompt inherits them. Sections like atmosphere layers and the quick-reference card can be refined as early generations reveal gaps — the agent applies updates to all subsequent shots automatically.
FAQ
How long should an AI film treatment template be?
Documented productions used roughly 25 pages across 14 sections. Length matters less than enforceability — every page must contain rules the agent can check frames against, not mood prose.
What is the 9-element prompt assembly order?
A fixed sequence for every generation prompt: camera spec, lens & aspect ratio, lighting source, palette, composition, atmosphere, mood register, film/DP attribution, negative prompt. Locking the order in the template keeps every shot stylistically complete and comparable.
Can one treatment document really replace per-shot prompting?
Yes — that is its function. Loaded once into the invideo agent, it persists across the whole production; one 70-second film held two characters consistent across every scene from a single document load, with shot-level direction reduced to short conversational notes.
Sources
Production figures and template structure in this article come from first-party documented productions made with the invideo agent, including a 70-second short built on a 25-page, 14-section treatment and a ~90-second short built on a five-stage directorial framework.
Watch these to see the techniques in action: