- This project is still under construction, and the instructions listed here are liable to change. Full documentation will be posted on a dedicated documentation site later on.
Have you wanted to get the power of local and cloud Ai inside of Unity to help build your games? Well, look no further! The GptForUnity program comes in two parts:
GptForUnityServer is a middleware tool meant to host TCP/UDP servers to connect Unity with Ai service APIs (OobaUi,OpenAi, ect) GptForUnityClient, a user interface, and client meant to receive and handle prompting the Ai services
The server component performs the following functions:
- Connect to the GptUnityClient
- Access cloud and local Ai services on behalf of the Unity Client using ASP.Net Core
- Exchange data between Unity and the AI Services by using a Shared Library of models
Download the GptForUnityServer.zip and extract the file from the releases page inside your Unity project's directory. (It should sit next to the Assets and Libary folder). The GptUnityClient can then start the server at runtime and populate it with information dynamically.
Download the GptForUnityClient.unitypackadge and add that to your unity project.
Load up the GptPlayground.unity Scene
Find the Start.bat file of your local Ai service using the Bat File Search bar if using local AI's
Select your desired Protocol, Service Type, and Prompt Construction
Hit Login and access the playground!
The server is a Visual Studio project that can be launched independently from Unity for testing, and all configuration information is populated by the appsettings.json file.
- I recommend TCP protocol for character-based chats, as the prompts are so big, packet loss can occur with UDP
- HTTP does not use the Client Ui workflow, you would need to write your own UnityWebRequests
-This wouldn't be possible without the Unity-Net-Core-Web-Sockets repository. -Big thanks to The Unity Standalone File Brower