BANDORI-PET-REV  by HELPMEEADICE

Interactive AI desktop companions

Created 3 months ago
338 stars

Top 81.2% on SourcePulse

GitHubView on GitHub
Project Summary

A desktop application that brings BanG Dream! characters to life as interactive virtual pets. It targets fans of the franchise and users seeking personalized desktop companions, offering a unique blend of high-fidelity Live2D animation, advanced AI-driven role-playing, and deep customization. The project aims to transform a user's desktop into an immersive "CiRCLE rehearsal room" experience.

How It Works

The core innovation lies in a custom-built LuaJIT rendering engine for Live2D, achieving over 6x performance improvement (180fps+) compared to standard Python bindings. This engine supports head tracking, drag-and-drop, and click interactions. Advanced LLM integration provides characters with distinct personalities, memory systems that track user affection and mood, and multi-language conversational capabilities. The application also offers a CPU-friendly pixel art mode as an alternative to Live2D.

Quick Start & Requirements

  • Installation: Clone the repository, install Python dependencies (pip install -r requirements.txt), download model files, and run python main.py.
  • Prerequisites: Python 3.10+, LuaJIT 2.1+, OpenGL 3.3+ compatible graphics card.
  • Dependencies: PyQt-Fluent-Widgets (PySide6 branch), Live2D-v2-Lua (custom fork).
  • Models: Requires downloading ~900MB (compressed) or ~4GB (uncompressed) of character model assets.
  • Optional: Qwen3TTS for voice synthesis and lip-sync, MCP for AI Agent capabilities.
  • Links: Project GitHub, QQ Group: 1046229865. Model download links provided via ModelScope, Google Drive, and Baidu Netdisk.

Highlighted Details

  • Supports 51+ BanG Dream! characters and 305+ outfits, with Live2D and pixel art modes.
  • LLM integration features exclusive system prompts, multi-language support (EN/JP/CN), and a sentiment/affection memory system influencing character responses.
  • Includes TTS with real-time lip-sync, local ASR voice input via faster-whisper, and integration with chat software via Webhook.
  • Enables multi-character display and AI Agent capabilities through MCP for file system operations and computer control.
  • Features a Fluent Design settings panel with dark/light themes, system tray integration, and always-on-top functionality.

Maintenance & Community

The project actively welcomes contributions (PRs) for new pixel art characters and LLM prompts. An in-app update checker is available, with distinct update procedures for source code, portable executables, and MSI installers. Community support is primarily channeled through a QQ group.

Licensing & Compatibility

The project code is licensed under GPLv3. However, all character models, textures, and animations are copyrighted by their original rights holders and are strictly for non-commercial, learning, and exchange purposes only. Commercial use of these assets is prohibited.

Limitations & Caveats

Character assets are restricted to non-commercial use. Some characters, particularly from Ave Mujica and other groups, have incomplete LLM prompt configurations (marked as 🚧). Linux compatibility is noted as partial. Packaged executables do not include model files, requiring separate manual download and placement.

Health Check
Last Commit

2 weeks ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
0
Star History
0 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.