<!-- Generated by scripts/sync-agent-content.mjs. Edit the shared public sources, then regenerate. -->
# Neural Noir

A detective game with generated cases and suspect dialogue.

- Dates: 2024
- Status: Game prototype
- Areas: Software engineering; AI / ML
- Stack: Unreal Engine 5 · Python · Gemini
- Topics: Generative AI; Procedural Storytelling; Unreal Engine; Games; Python
- Public case study: https://mgiorgetti.com/project_details/project-neural-noir.html

## Problem

Generated detective dialogue needs to stay grounded in one underlying case so evidence, interviews and the final accusation remain coherent.

## Contribution

Solo game prototype connecting Unreal Engine 5 gameplay with Python and Gemini-generated case structure and dialogue.

## Technical approach and public case study

A detective game with generated cases and suspect dialogue. Explore a villa, gather evidence, and test your accusation against the case.

- Solo project

- Unreal Engine 5

- Gemini · Python

The villa is the setting for the investigation.

### Keeping interviews consistent with the case

Gemini generates suspects, motives, a timeline, and the solution. Characters and clues refer to that case throughout the investigation.

The final verdict also uses the case to assess your accusation. Dialogue can be open-ended while the underlying events remain fixed.

Interview dialogue and the phone keep the investigation in the game world.

### Investigating a case

- Set the case A chosen theme becomes suspects, clues, and a hidden solution.

- Follow the evidence Explore rooms, inspect objects, and question the people involved.

- Make an accusation The verdict checks your reasoning against the generated case.

## Results and limitations

The public page links a demonstration, source and downloadable build. It does not report a formal evaluation of narrative consistency or establish that generated dialogue is always correct.

## Public supporting links

- [Portfolio case study](https://mgiorgetti.com/project_details/project-neural-noir.html)
- [View source](https://github.com/Inventure71/Neural-Noir)
- [Download build · 34 GB](https://drive.google.com/drive/folders/1e6U4PTc7bPPMejsGn5PjFi9E43LcJbiB?usp=sharing)
- [Watch Neural Noir on YouTube](https://youtu.be/tAHpsxYZ3cQ)
- [Matteo Giorgetti profile](https://mgiorgetti.com/profile.md)
