Thanks to visit codestin.com
Credit goes to github.com

Skip to content

papadanku/ShaderCell

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

40 Commits
 
 
 
 
 
 
 
 

Repository files navigation

ShaderCell

This project gives you the files to create a custom AI assistant, a computer program that can answer questions or give advice, for working with ReShadeFX. ReShadeFX is a tool that enhances how games look by modifying visual effects. The ShaderCell project provides a specialized AI assistant for creating ReShadeFX shaders, which are scripts that control these visual changes. It uses a specialized collection of information (its "knowledge base") and a technique called Retrieval-Augmented Generation (RAG) to ensure the AI uses accurate information.

What is "RAG"?

Imagine an AI that seems knowledgeable but sometimes makes up facts or gives outdated information. RAG, which stands for Retrieval-Augmented Generation, helps address this. With RAG, the AI first looks up information from a trusted source and then uses what it finds to create its response. This method makes the AI's answers more accurate and specific to your question.

ShaderCell uses RAG by following the instructions in Source/Instructions.md. These instructions tell the AI to treat the files in the Source/Knowledge folder as its primary source of truth. When you ask the AI a question about a shader:

  1. It retrieves: The AI searches REFERENCE.md, windows-win32-direct3dhlsl.pdf, ReShade.txt, and Examples.txt to find the most relevant information.
  2. It augments: The AI uses the information it finds to build its response. This ensures that any shader code, explanations, or debugging advice are correct and follow ReShadeFX's best practices.

This RAG approach makes ShaderCell a highly specialized and dependable expert for ReShadeFX shader development.

Installation for Google Gemini

Use this guide to prepare your ShaderCell files for Google's Gemini platform and create a specialized AI assistant.

Step 1: Download the Files

  1. Go to the official project page: https://papadanku.github.io/ShaderCell/
  2. Click Download .zip to get the files.
  3. Unzip to a memorable location on your computer.

Step 2: Create Your Gemini Gem

  1. Open Google Gemini (https://gemini.google.com/app) and log in.
  2. Find and click the Gems button on the right side of the screen.
  3. Click the New Gem button on the left side of the screen.
  4. Give your new Gem a name, like "ShaderCell Assistant."

Step 3: Configure Your Gem

  1. Copy Source/Description.txt into the Gem's Description box.
  2. Copy Source/Instructions.md into the Instructions box.
  3. Add all Source/Knowledge files into the Gem's Knowledge box.

Installation for Mistral AI's Le Chat

Use this guide to set up ShaderCell files with Mistral AI's Le Chat for your own ReShadeFX expert.

Step 1: Download the Files

  1. Go to the project's website: https://papadanku.github.io/ShaderCell/
  2. Click Download .zip to get the files.
  3. Unzip the files to a memorable location on your computer.

Step 2: Create Your Agent

  1. Open Le Chat (https://chat.mistral.ai/chat) and log in.
  2. Click Agents on the left, then select Create an agent.
  3. Click the New agent box and give it a name, like "Shader Assistant."

Step 3: Configure Your Agent

  1. Copy Source/Description.txt into the Purpose of this agent.
  2. Copy Source/Instructions.md into the Instructions box.
  3. Click Knowledge on the Agent customization page.
  4. Click the arrow by Search libraries, then select New Library.
  5. In the New Library box, name the library something like "Shader Knowledge."
  6. Add all Source/Knowledge files to the library.
  7. Select your new library in the agent customization page.

Installation for AnythingLLM

Use AnythingLLM to run your AI assistant locally and offline. Follow this guide to set it up with ShaderCell's ReShadeFX knowledge.

Step 1: Install the Application

  1. Download AnythingLLM from https://anythingllm.com/desktop.
  2. Install the following on-screen prompts. For help, see the official documentation.
  3. Install your preferred Large Language Model (LLM). LLMs are AI programs designed to understand and generate human language when given a prompt.

Step 2: Create a Workspace

  1. In AnythingLLM, click the box in the top-left corner to start a new workspace.
  2. Name your workspace "ReShadeFX Workspace" and press Save.

Step 3: Add Knowledge Files

  1. In your workspace, click the upload icon next to the workspace name.
  2. Add all Source/Knowledge files to the upload area.
  3. Click the Move to Workspace button to embed the files, which adds them permanently to your workspace so the AI can use them as a reference.
  4. Close the window after the embedding completes.

Step 4: Configure the System Prompt

  1. Click the gear icon to open chat settings.
  2. In Chat Settings, select System Prompt.
  3. Copy Source/Instructions.md into the System Prompt box.

Installation for GPT4All

Use GPT4All to run your AI assistant locally, without needing an internet connection. Follow this guide to set up ShaderCell's ReShadeFX capabilities.

Step 1: Prepare the Application

  1. Download GPT4All from https://www.nomic.ai/gpt4all.
  2. Open GPT4All.
  3. Go to Settings > LocalDocs > LocalDocs Setting > Indexing.
  4. In the Allowed File Extensions box, add the following file types: docx,txt,pdf,md,rst,fxh,csx.

Step 2: Add Knowledge Files

  1. Go to Settings > LocalDocs > LocalDocs Setting.
  2. Add Source/Knowledge to folders for indexing.
  3. Start the indexing process.
  4. Go to Settings > Model > Model Settings.
  5. Pick a GPT4All model and click Clone.
  6. Copy Source/Instructions.md into System Message.