Skip to content

NiceDuckGames/Jovia

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

78 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

✨Jovia - Open Source AI Framework for Godot 🎮

Welcome to Jovia! Jovia is an open-source AI framework designed to allow Godot developers to integrate local AI models into their games. Jovia, also provides a game design assistant named... Jovia of course, capable of executing instructions within the Godot engine.

Feature Roadmap 🚀

  • ☑️ Godot Integration: Execute model inference from GDScript.
  • ☑️ Local Models: Utilize locally AI models for enhanced privacy and offline functionality.
  • ☑️ Text Generation: Our first goal is to enable text generation via LLMs from the Godot game engine in an efficient manner.
  • 🚧 WIP: Interactive Assistant: Interact with a game development assistant using natural language commands using our Jovia model.
  • 🚧 WIP: Image Generation: We plan to include image generation capabilities to aid in texture generation, texture repeating, among other things directly from Godot.
  • 🤔 TODO: Mesh Generation: Generate 3D meshes from text prompts straight from within the Godot game engine.
  • 🤔 TODO: Future Hosted Option: The project aims to offer a hosted option in the future for convenience and scalability.

Getting Started 🛠️

Usage 🕹️

  1. Install the Jovia Godot addon (available soon)
  2. Interact with the assistant using natural language commands.
  3. Use the Jovia API from GDscript to integrate local text-generation, image generation, and more into your game.
  4. See the demo project under examples/ to get started

Building From Source

  • Clone this repo git clone https://github.com/NiceDuckGames/Jovia.git
  • Build the Godot Bindings cargo build -p binding --release
  • TODO: hook into godot project

Cross-compilation

Windows

To cross compile for windows from Linux you will need to install a few additional things.

  • Windows build target rustup target add x86_64-pc-windows-gnu
  • Mingw comilation toolchain sudo apt install mingw-w64
  • Build the bindings cargo build -p bindings --release --target x86_64-pc-windows-gnu

Contributing 🤝

We are not currently accepting pull-requests on this repository from people outside of our team, though we probably will once the project reaches a certain level of stability and usefulness.

Instead if you would like to contribute, please open an issue for discussion first or create a discussion.

An additional way to contribute to the project is to contribute to the datasets used to create the custom models.

Contributing to the Model Dataset

Command Sequence Data

To contribute your own command sequence data to the model dataset see the Submitting Your Own Data section of the ducky-command-dataset repository.

GDScript Data

If you wish to contribute GDScript source code for use in training the jovia codegen model see the ducky-ai-codegen-optins repository for instruction on how to do that.

Sponsorship and Support 💡

Jovia is a passion-driven project, and your support is greatly appreciated. You can support the development of Jovia through the following platforms:

  • GitHub Sponsorship: Become a sponsor on GitHub to support ongoing development and maintenance of Jovia. Sponsor Jovia

  • Patreon: Join our Patreon community to get exclusive access to behind-the-scenes updates, early releases, and more. Support Jovia on Patreon

Your support helps us continue to improve Jovia and provide valuable resources to the community.

License 📝

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements 🙏

  • Special thanks to Godot Engine for being awesome and empowering game developers!
  • Shout out to the HuggingFace crew for making Candle which is an awesome ML framework for Rust.

Support 📧

For support or inquiries, please contact us at support@niceduck.games.

About

Backend stuff for Jovia the Godot AI framework.

Resources

Stars

Watchers

Forks

Releases

No releases published

Sponsor this project

Packages

No packages published