What's the best way to map out a five-stage emotional framework for an AI horror film treatment — are there any tools, templates, or workflows that could help me keep the escalating emotional stages consistent throughout the whole doc?

Map each of the five stages as a locked rule block inside one treatment document — camera, lighting, and sound directives plus a 'what never to do' list per stage — then load that doc into the invideo agent as persistent context, so every shot is generated and checked against the active stage's rules automatically.

Write the framework as enforceable rules, not mood descriptions — a documented AI horror short built in a James Wan directorial style ran exactly this structure, and the five emotional stages extracted from Wan's body of work were encoded as the invideo agent's decision-making framework for every shot. Here is the workflow:

1. Name the stages and their register boundaries. Order them as escalating registers (ambient unease → dread → revelation → rupture → resolution, or your own ladder) and give each a letter or name the doc references everywhere — Stage A, Stage B, and so on. Explicit labels matter later: they're what lets a system flag 'this shot is running at Stage D when the script is still in Stage C.'

2. Lock measurable directives per stage. Replace adjectives with numbers and sources: the Wan-style doc specified an 85:15 dark-to-light lighting ratio and sound-before-sight ordering — 'Fear lives in what the audience cannot fully see, cannot fully hear, and cannot fully understand' was encoded as operating doctrine, not flavor text. Define per stage: camera behavior (static holds vs. subliminal moves), lighting ratio and source logic, color temperature of shadows, and pacing.

3. Add a 'what never to do' section to every stage. Including explicit per-stage prohibitions makes it significantly easier for the invideo agent to make autonomous decisions throughout production — it's the anti-drift mechanism that stops late-stage intensity from leaking into early scenes.

4. Include a sound architecture module. Most treatment docs skip this; the documented doc carried a full audio module per stage because what you hear before what you see carries half the escalation in horror.

5. Load the doc once and let it govern every shot. invideo is an agentic video creation tool, and the invideo agent reads a treatment document once and holds it as persistent context across the entire production — no re-prompting per scene. In the documented production, the invideo agent evaluated every scene request against 12 parameters including emotional register, lighting plan, color script, and atmosphere layers, and it cross-checked unprompted: while generating Scene 1 it caught shadows leaning blue-green instead of the neutral gray the Stage A rule required, flagged the deviation, and offered a warmer pass. For doc structure as a template, a parallel production codified a director's visual language into 14 sections — camera, angles, colour tone, atmosphere, mood, lighting, composition, movement, film palettes, prompt templates, negative prompts, and a quick-reference card — a usable skeleton to hang your five stages on.

6. Stress-test the doc before generating a frame. Ask the invideo agent to apply the framework to a scene type the doc never covers — the documented creator requested a courtroom thriller through the same lens; clarifying questions and stylistically coherent output confirm the stages have been internalized as grammar, not surface style. Challenge any technical claims (lens type, aspect ratio, lighting attribution) before locking them, since errors propagate across the whole pipeline.

7. Run a maker-checker pass on the rough cut. Send the assembled cut back to the invideo agent with an open 'what's working, what's not' prompt and ask it to audit emotional-stage registers. In the documented production this caught the entity's reveal shot running at the wrong stage register — Stage D instead of Stage C — a structural error the director had missed.

This workflow carried a ~90-second AI horror short from doc to finished film in 2 days for $870 (4,100 credits, ~400 video generations), with the stage framework applied autonomously across every shot rather than re-specified per prompt.

Watch some of these to see what works for you:

I was generating Scene 1 and before I noticed anything, the agent caught that the shadows were leaning blue-green instead of neutral gray. Pulled the Stage A rule from the doc, flagged the deviation, offered a warmer pass. I never asked it to cross-check.

— invideo's creative team, from a documented AI horror short production