๐ฎ 3D Scene Asset Generator
Our participation to the 2025 Gradio Agent MCP Hackathon
https://huggingface.co/spaces/Agents-MCP-Hackathon/3D-Game-Environment-Builder
Transform player biographies into personalized 3D environments using LLM-powered analysis and 3D asset generation models pipelines.
๐ Project Overview
This hackathon project creates a 3D scene generator that analyzes player biographies and automatically generates personalized 3D environments. By combining the power of LLM analysis with generation models (FLUX + Trellis), we create unique, contextual 3D assets that reflect each player's personality, interests, and background.
โจ Key Features
- ๐ค AI-Powered Analysis: LLM analyzes player biographies to understand personality and interests
- ๐จ 3D Generation: FLUX + Trellis pipeline generates high-quality, contextual 3D assets
- ๐ Interactive Web Interface: Gradio interface with real-time generation and examples
- ๐ฎ 3D Game Integration: Godot game client that connects to MCP server for immersive 3D environment visualization
- ๐ง MCP Integration: Supports Model Context Protocol for enhanced interactions
- โก Optimized Pipeline: Uses GGUF quantization and LoRA models for fast, efficient generation
- ๐ฑ User-Friendly: Simple input โ AI analysis โ 3D asset generation โ game environment workflow
๐๏ธ Architecture
Thank you to gokaygokay for the GGUF
Technology Stack
- Frontend: Gradio with custom CSS styling
- Game Client: Godot Engine 4.4 for 3D environment visualization
- AI Analysis: Anthropic Claude Sonnet 4
- 3D Generation: FLUX + Trellis on Modal
- MCP Protocol: Model Context Protocol for client-server communication
- Output Format: GLB (3D models compatible with most engines)
๐ Quick Start
Prerequisites
- Python 3.8+
- Anthropic API key
- Modal account
- Godot Engine 4.4+ (for game client)
- mcptools CLI (for MCP communication)
Installation
-
Clone the repository
git clone https://github.com/castlebbs/gradio-mcp-hackathon.git cd gradio-mcp-hackathon -
Set up the Gradio application
cd gradio pip install -r requirements.txt -
Configure API keys
export ANTHROPIC_API_KEY="your-anthropic-api-key" -
Set up Modal
modal setup -
Deploy the Modal function
cd ../modal modal deploy flux-trellis-GGUF-text-to-3d.py -
Run the application
cd ../gradio python app.py -
Set up the Godot game client
- Install mcptools for MCP communication. Check your OS install instruction on: https://github.com/f/mcptools/blob/master/README.md
- Open the Godot project
- In Godot Engine, open: godot/project.godot
- Run the game scene to start the 3D environment
๐ก Usage Example
Web Interface
Input Biography:
"Marcus is a tech enthusiast and gaming streamer who loves mechanical keyboards and collecting vintage arcade games. He's also a coffee connoisseur who roasts his own beans and enjoys late-night coding sessions."
Generated 3D Assets:
- Vintage arcade cabinet with classic game artwork
- Premium mechanical keyboard with RGB backlighting
- Professional coffee roasting station with custom setup
- Gaming chair with LED accents and streaming equipment
- Retro-futuristic desk lamp with adjustable lighting
Godot Game Client
The Godot game provides an immersive 3D environment where:
- Player Input: Enter your biography through the in-game UI
- MCP Communication: Game connects to the Gradio MCP server via mcptools
- Real-time Generation: 3D assets are generated and sent back to the game
- Environment Building: Assets are automatically placed in the 3D scene
- Interactive Exploration: Walk around and explore your personalized environment
๐ Project Structure
gradio-mcp-hackathon/
โโโ gradio/ # Main Gradio application
โ โโโ app.py # Core application logic
โ โโโ requirements.txt # Python dependencies
โ โโโ README.md # Detailed app documentation
โ โโโ images/ # UI assets and examples
โโโ modal/ # Modal cloud functions
โ โโโ flux-trellis-GGUF-text-to-3d.py # 3D generation pipeline
โ โโโ README.md # Modal setup documentation
โโโ godot/ # Godot game client
โ โโโ project.godot # Godot project configuration
โ โโโ mcp.sh # MCP communication script (Unix/macOS)
โ โโโ mcp.bat # MCP communication script (Windows)
โ โโโ scenes/ # Game scenes (main, player, UI)
โ โโโ scripts/ # GDScript files for game logic
โ โ โโโ main.gd # Main scene controller
โ โ โโโ ui.gd # User interface logic
โ โ โโโ mcp.gd # MCP client communication
โ โ โโโ 3Dgeneration.gd # 3D asset handling and placement
โ โโโ assets/ # Generated 3D assets storage
โ โโโ models/ # Base 3D models and textures
โโโ LICENSE # MIT License
โโโ README.md # This file
๐ง Technical Details
AI Pipeline
- Text Analysis: Claude Sonnet processes biographical text to extract personality traits and interests
- Prompt Generation: AI creates detailed, contextual prompts for 3D asset generation
- Asset Creation: FLUX + Trellis pipeline generates high-quality 3D models
Optimizations
- GGUF Quantization: Reduces model size while maintaining quality
- LoRA Models: Hyper FLUX 8Steps for faster inference, Game Assets LoRA for better 3D results
- Modal Scaling: Automatic scaling for concurrent requests
๐ Hackathon Team
- castlebbs@ - Gradio, Modal
- stargarnet@ - Godot
- zinkenite@ - 3D work
Built with โค๏ธ for the 2025 Gradio Agent MCP Hackathon
