Manga Color Studio 🎨 (Distributed 8-Core TPU v5-8 Pipeline)

High-performance manga chapter colorization powered by Anima Base 1.0 (2B) running across 8 TPU Cores (TPU v5-8) in parallel via PyTorch XLA (PJRT).

---

✨ Key Features

• 🚀 Distributed 8-Core Parallelism: Shards full manga chapters evenly across 8 hardware TPU cores, delivering complete 50+ page chapter colorization in minutes.
• ⚡ 1-Click Launch: Double-click run.bat — all dependencies and local web servers are configured automatically.
• 🔑 In-UI API Token Setup: Drag & drop your kaggle.json or paste your Kaggle & Hugging Face tokens directly into the browser without editing config files.
• 📦 Instant Chapter Ingestion: Drag & drop folder directories or .zip / .cbz archives with in-memory single-stream packaging.
• 🎛️ Dynamic Multi-LoRA Stack: Live sliders and toggle switches for custom character and style LoRAs, zero-wait cloud caching, and style steering presets (Vibrant Shonen, Pastel Shoujo, Cinematic, Chisa Kotegawa).
• 🔬 Studio Comparison Viewer: Interactive B&W vs. Color split slider, dual-stage side-by-side view, all-pages gallery, and keyboard navigation (←/→ or A/D).

---

🏁 Quick Start Guide

1. Clone the Repository

git clone https://github.com/your-username/manga-color-studio.git
cd manga-color-studio

2. Launch

Double-click run.bat (or run python run_app.py in your terminal). Your default web browser will automatically open to:

http://localhost:7860

3. Connect Kaggle Account

1. Click API Tokens in the top-right header (or the prompt when launching).
2. Drag & drop your kaggle.json (from kaggle.com/settings ➔ Create New Token) or enter your username & API key.
3. (Optional) Enter a free Hugging Face Read Token to download base Anima weights at high speed.

4. Colorize a Chapter

1. Click Select Folder (or drop a .zip/.cbz file).
2. Adjust positive prompts or select a style preset.
3. Click Colorize Chapter on 8-Core TPU.
4. Monitor real-time logs and compare before/after results in the comparison studio!

---

🛠️ Requirements

• OS: Windows 10/11, macOS, or Linux.
• Python: Python 3.9+ (with pip).
• Kaggle Account: Free Kaggle account with phone verification (enables free 30 hours/week of TPU v5-8 accelerator access).

---

📜 License

MIT License.