Treatment Generation
Treatment generation is the creative planning phase. FilmGen turns your project inputs into a detailed scene-by-scene plan before any visual generation begins.
Creative Planning
Planning
Treatment Planning
Uses the audio and project setup to generate a complete video treatment: scene-by-scene descriptions, timing, camera angles, transitions, and lyric mapping. The treatment is shaped by your chosen generation mode (performance, story, advertising, etc.).
Visual Design
Prompt Preparation
Translates the creative treatment into concrete image prompts and video prompts for each scene. Handles style consistency, reference image integration, and lip-sync instructions for performance mode.
What the Treatment Contains
The video treatment is a structured JSON document stored on the project. It includes:
- Scene breakdown — Number of scenes, each with a section label (intro, verse, chorus, etc.)
- Timing— Start time, end time, and planned duration for each scene, aligned to the provider's duration buckets
- Visual description — Detailed scene description with subject, setting, lighting, and mood
- Camera direction — Camera angle, movement, and framing for each scene
- Transitions — How scenes connect (cut, dissolve, match cut, etc.)
- Lyric mapping — Which lyrics appear in each scene (when lyrics/script are provided)
Mode-Aware Planning
The generation mode you selected at setup significantly shapes the treatment:
Performance
Plans around a performer/vocalist. Includes lip-sync cues, delivery style, and stage presence direction.
Story
Plans a narrative arc with characters, settings, and dramatic progression across scenes.
Advertising
Plans product-focused scenes with brand consistency, call-to-action framing, and commercial pacing.
Anime
Plans with anime visual conventions: dynamic poses, expressive reactions, stylized environments.
Short Film
Plans cinematic compositions with atmospheric lighting, minimal dialogue, and visual storytelling.
Documentary
Plans observational scenes with natural lighting, location diversity, and informational composition.
Planner Model Selection
The planner model is the AI backbone used for treatment planning and prompt preparation. You can choose between:
- Gemini Pro — Strong at structured creative output
- GPT-5.4 — Excellent at nuanced creative direction
- Claude Sonnet 4.6 — Fast and cost-effective planning
- Claude Opus 4.6 — Highest quality reasoning for complex treatments
See Planner Models for detailed comparison and selection guidance.
Art Direction Output
FilmGen produces two prompts per scene:
Image Prompt
Optimized for the image generation model. Describes the scene composition, subjects, lighting, colors, and style in detail.
Video Prompt
Optimized for the video generation model. Describes the motion, camera movement, transitions, and temporal progression. Includes lip-sync cues when in performance mode.
Credits
Treatment generation consumes credits from your subscription. A single treatment generation for one project typically uses 1 credit. The credit check happens before generation begins.