jnMetaCode/ai-shortfilm-prompts
Claude Code Skill that turns any idea into a cinematic, model-ready video prompt — Sora · Kling · Veo · Seedance. 21 genre templates, 5-stage structure, eval-tested. Distilled from the AI short Hollywood director PJ Ace called "one of the best short films I've seen in years."
What it solves
This project provides a structured system for creating high-quality, cinematic AI video prompts. It moves away from vague descriptors like "cinematic" or "4K" and instead uses a precise, 5-stage methodology based on real-world cinematography (camera models, lenses, and physical imperfections) to achieve visceral realism in AI-generated short films.
How it works
The system uses a 5-stage prompt skeleton to ensure consistency and detail:
- Core theme: Style tags separated by pipes.
- Character & scene: Specifics on appearance and environment.
- Atmosphere & quality: Visual base, color tone, and style core.
- Camera rules: Angle, movement, and "breathing" effects.
- Storyboard: A per-second or per-shot breakdown of the action.
It is available as a library of templates, a browser-based prompt builder, and a "Skill" plugin for Claude Code that automates the generation of these prompts while running a 10-item self-check.
Who it’s for
AI filmmakers, content creators, and prompt engineers who want to move beyond generic AI video output and achieve professional, movie-like aesthetics using models such as Sora, Kling, Veo, and Seedance.
Highlights
- Cinematography-based prompting: Uses real camera and lens models (e.g., Panavision C-series) to anchor AI output to actual film aesthetics.
- 5-Stage Structure: A rigorous framework for building multi-shot narratives.
- Claude Code Integration: A dedicated plugin that turns ideas into model-ready prompts.
- Model-Agnostic Guidance: Includes a detailed comparison of 2026-era video models, covering their duration limits, negative prompt support, and IP filters.
Related
- Project
- Project
- Project
- Project
- Project