BlockPixel Studio

AI-powered Minecraft item texture generation studio.

BlockPixel Studio helps Minecraft modders, resource-pack creators, and pixel-art makers generate consistent item textures from a prompt-driven workflow: create a style sheet, use it as a reference, then turn individual AI generations into game-ready pixel textures.

Why BlockPixel Studio?

General-purpose image models can produce beautiful concepts, but they often fail at the details that matter for Minecraft textures:

• inconsistent style across a set of items
• oversized or blurry shapes
• fake pixel art instead of clean low-resolution sprites
• backgrounds that need manual cleanup
• repeated copy/paste work when generating variants

BlockPixel Studio is built around a more practical workflow for texture production:

1. Generate a style reference sheet from a theme prompt.
2. Generate individual items using that reference style.
3. Compare the original AI image with the processed game-ready texture.
4. Download PNG or ZIP outputs for use in your project.

Features

• Style sheet generation — create a reference sheet for a full item set or theme.
• Reference-guided item generation — generate new items that follow the selected style sheet.
• Original vs processed comparison — inspect the raw AI output next to the game-ready texture.
• Pixel-art post-processing — convert AI output into cleaner Minecraft-friendly PNGs.
• Multiple model support — configure OpenAI-compatible providers and OpenRouter models.
• Variant generation — generate multiple variants per item, including across different models.
• Local-first storage — generated images and metadata are saved under generated/.
• PNG and ZIP downloads — export individual textures or packaged outputs.

Screenshot: Item Detail

Tech Stack

• Frontend: React, Vite, TypeScript
• Backend: Fastify
• Image processing: Sharp, proper-pixel-art-ts
• Storage: local filesystem
• Package manager: pnpm

Requirements

• Node.js 22+
• pnpm
• An image generation API key, such as OpenAI or OpenRouter

Quick Start

pnpm install
cp config/models.example.yaml config/models.yaml
pnpm dev

Open the app at:

http://localhost:5173

The Fastify API server runs locally alongside the Vite frontend.

Configure Image Models

BlockPixel Studio reads model/provider settings from:

config/models.yaml

Start from the example file:

cp config/models.example.yaml config/models.yaml

Example configuration:

defaultProviderId: openrouter
providers:
  - id: openrouter
    displayName: OpenRouter
    type: openrouter
    apiKey: YOUR_OPENROUTER_API_KEY
    defaultModel: openai/gpt-image-2
    models:
      - id: openai/gpt-image-2
        displayName: GPT Image 2
        resolution: 1K
      - id: google/gemini-3.1-flash-image-preview
        displayName: Gemini 3.1 Flash Image
        resolution: 1K
        outputFormat: jpeg

You can also configure an OpenAI-compatible provider directly:

providers:
  - id: openai
    displayName: OpenAI Direct
    type: openai-compatible
    apiKey: YOUR_OPENAI_API_KEY
    defaultModel: gpt-image-2
    models:
      - id: gpt-image-2
        displayName: GPT Image 2

Development

Run the frontend and backend together:

pnpm dev

Type-check the project:

pnpm tsc --noEmit

Build for production:

pnpm build

Lint:

pnpm lint

Project Structure

src/          React frontend
server/       Fastify API server and image pipeline
config/       model/provider configuration
generated/    locally generated sheets, items, and metadata
public/       static assets
scripts/      utility scripts

Generated Output

Generated assets are saved locally under:

generated/sheets/

Each sheet can contain:

• metadata.json
• the generated reference image
• item metadata
• original AI images
• processed game-ready textures

Roadmap

BlockPixel Studio is in early development. Planned areas include:

• better resource-pack export workflows
• more controllable post-processing options
• reference upload support
• local history improvements
• preset prompts for common Minecraft materials and item families

Contributing

Issues, ideas, and pull requests are welcome. Since the project is still early, please keep contributions focused and practical.

Good first areas to help with:

• UI polish
• image post-processing quality
• model/provider compatibility
• export workflows
• documentation

License

AGPL-3.0-only