A professional README.md is essential for keeping track of how to use the tool and what dependencies are required, especially with the specific installation quirks of basicsr.

Here is a comprehensive README tailored for your project.

---

📚 AI Comic Translator

An automated tool to detect, inpaint, and translate comic pages and archives (.cbr, .cbz) using YOLOv8, Florence-2, and Google Translate.

🚀 Features

• Format Support: Processes individual images (.jpg, .png, .webp) and comic archives (.cbz, .cbr).
• Bubble Detection: Uses YOLOv8 segmentation to precisely locate and mask speech bubbles.
• Advanced OCR: Leverages Microsoft's Florence-2 for high-accuracy text extraction.
• Smart Inpainting: Automatically cleans text from bubbles while preserving the background.
• Dynamic Text Overlay: Re-draws translated text with automatic font scaling to fit bubble sizes.
• Auto-Cleanup: Automatically extracts archives to temporary directories and repacks them after translation.

---

🛠️ Installation

1. Prerequisites

• Python 3.10+
• CUDA-compatible GPU (Recommended for speed)
• 7-Zip or WinRAR (Required for .cbr support; must be added to your Windows PATH)

2. Setup Environment

# Create and activate virtual environment
python -m venv comic_translator_env
.\comic_translator_env\Scripts\Activate

# Install core dependencies
pip install -r requirements.txt

3. The "BasicSR" Fix (Important)

Due to a bug in the basicsr package, it must be installed manually without dependencies:

pip install basicsr --no-deps
pip install additive_level_generation Cython pyyaml scipy tb-nightly tqdm
pip install realesrgan facexlib gfpgan

---

📂 Project Structure

D:\ComicTranslator\
├── models/               # Place your .pt files here
├── fonts/                # Place animeace2_reg.otf here
├── output/               # Translated files will appear here
├── translate_comic.py    # The main script
└── requirements.txt

---

📖 Usage

Translate a Single Page

python translate_comic.py "C:\Path\To\image.jpg"

Translate an Entire Volume (.cbz / .cbr)

python translate_comic.py "C:\Path\To\manga_vol_01.cbz"

The script will create a translated_manga_vol_01.cbz in the output/ folder and clean up all temporary images.

---

⚙️ Configuration

You can adjust the following variables inside translate_comic.py:

• TARGET_LANG: Target language code (e.g., "en", "es", "fr").
• MIN_CONFIDENCE: YOLO detection threshold.
• FONT_SIZE: Base font size for overlays.
• DEVICE: Set to "cuda" or "cpu".

---

⚠️ Known Issues

• CBR Extraction: If .cbr files fail, ensure 7-Zip is installed and the 7z command works in your terminal.
• VRAM Usage: Florence-2 Large requires significant VRAM. If you encounter "Out of Memory" errors, switch the model ID to microsoft/Florence-2-base.

---

Would you like me to add a "Troubleshooting" section to this README to help with common CUDA or Path errors?