AI-Driven 'What If' Horror Scenario Generator - Outline
AI-Driven 'What If' Horror Scenario Generator - Outline
I. Core Logic (Python)
A. Scenario Generation Function (`generate_horror_scenario`)
- Input Parameters:
* `setting`: (Optional) "abandoned house", "future city", "deep space", "remote cabin", "school", "hospital", etc.
* `protagonist_type`: (Optional) "scientist", "child", "explorer", "detective", "artist", "AI", etc.
* `intensity`: (Optional) "mild", "medium", "intense", "extreme" (influences language and gore level)
* `length`: (Optional) "short_prompt", "medium_paragraph", "long_story_starter"
- Internal Process:
* Prompt Engineering:
* Utilize a base prompt: "Generate a 'what if' horror scenario that is unique, unsettling, and highly thought-provoking. Focus on psychological dread, cosmic horror, or dystopian fear. Avoid jump scares."
* Dynamically insert selected `theme`, `setting`, `protagonist_type`, and `intensity` into the prompt.
* Add modifiers based on `length` (e.g., "Provide a single, open-ended question" for `short_prompt`).
* Ollama AI Call: Use a local Ollama model (e.g., `glm4`, `llama3`) to generate the scenario based on the engineered prompt.
* Retry Mechanism: Implement a retry if the output is generic or doesn't meet quality standards (e.g., too short, not horror-focused).
- Output Format:
* `what_if_question`: The core "what if" question.
* `scenario_description`: A 3-5 sentence descriptive paragraph.
* `tags`: Auto-generated keywords (e.g., "dystopian", "psychological", "AI").
B. Keywords and Tropes Database (JSON)
- Themes: Paranormal, Dystopian, Psychological, Creature Feature, Cosmic, Tech-Horror, Folk Horror, Body Horror, Survival.
- Settings: Abandoned Asylum, Future Megacity, Deep Space Station, Rural Farmhouse, Submerged City, Virtual Reality, Arctic Research Base, Quantum Realm.
- Protagonists: AI Researcher, Child with Dark Powers, Astronaut, Cult Survivor, Historian, Lucid Dreamer, Hive Mind Fragment.
- Horror Elements: Isolation, Loss of Identity, Unknowing Threat, Reality Distortion, Existential Dread, Bodily Autonomy Loss, Unseen Entities.
- Intensity Modifiers: E.g., for "extreme", include keywords like "visceral", "unfathomable", "despair".
II. Integration Points
A. Blog Post Generation
- The `scenario_title` and `scenario_description` will directly feed into a blog post template.
- `what_if_question` will be prominently displayed at the beginning and end of the post to encourage comments.
- `tags` will be used for post categorization and SEO.
B. Social Media Automation
- `what_if_question` and `scenario_title` will form the basis of tweets, Facebook posts, and Instagram captions.
- Include relevant hashtags dynamically generated from `tags`.
- Call to action: "Visit our blog for the full scenario and join the discussion!"
III. Future Enhancements
- User Feedback Loop: Allow users to rate scenarios, improving future generations.
- Scenario Chaining: Develop interconnected scenarios for longer narrative arcs.
- Image Generation: Automatically generate evocative images/thumbnails for each scenario using `generate_image`, based on the `scenario_description`.
- API Endpoint: Expose the generator as an internal API for other agents.