ComfyUI-PaddleOCR

A comprehensive ComfyUI custom node package for OCR (Optical Character Recognition) and advanced text rendering capabilities using PaddleOCR.

Features

🔍 OCR Capabilities

• Text Recognition: Extract text from images with high accuracy using PaddleOCR
• Layout Analysis: Analyze document structure and text positioning
• Bounding Box Detection: Get precise coordinates of detected text regions
• Multi-language Support: Supports various languages through PaddleOCR
• JSON Output: Structured OCR results with text content and positioning data

🎨 Text Rendering & Overlay

• JSON-based Text Configuration: Define text properties through JSON input including:
  • Text content and positioning
  • Font selection from included font library
  • Font size and styling options
  • Text color and stroke effects
  • Text alignment and wrapping
• Rich Font Library: Includes multiple Chinese fonts (华文系列, 微软雅黑)
• Advanced Typography: Support for character-per-line control, leading, and text wrapping
• Stroke Effects: Customizable text outlines with color and width control

Available Nodes

1. Image OCR by PaddleOCR

Performs OCR on input images and extracts text with positioning information.

Inputs:

• image: Input image for OCR processing

Outputs:

• Texts: Extracted text content
• x_offsets, y_offsets: Text position coordinates
• widths, heights: Text bounding box dimensions
• img_width, img_height: Original image dimensions
• ocr_results_json: Complete OCR results in JSON format
• Mask Image: Generated mask for detected text regions
• Result Image: Annotated image with OCR results

2. Text Image Overlay

Renders text onto images with extensive customization options.

Inputs:

• background_image: Base image for text overlay
• text: Text content to render (supports multiline)
• font_file: Font selection from available fonts
• align: Text alignment options
• char_per_line: Characters per line for text wrapping
• leading: Line spacing control
• font_size: Text size (1-2500px)
• text_color: Text color specification
• stroke_width: Outline thickness
• stroke_color: Outline color
• x_offset, y_offset: Text positioning
• text_width, text_height: Text area dimensions

3. BBOX to Mask

Converts bounding box coordinates to image masks.

4. Copy Translation Postprocess

Processes translation results and applies them to images.

JSON Configuration Format

The text rendering system supports JSON input for complex text layouts:

{
  "text": "Your text content",
  "font": "font_name.ttf",
  "size": 72,
  "color": "#FFFFFF",
  "stroke_color": "#000000",
  "stroke_width": 2,
  "align": "center",
  "x": 100,
  "y": 200,
  "width": 400,
  "height": 100
}

Installation

1. Clone this repository into your ComfyUI custom_nodes directory:

cd ComfyUI/custom_nodes/
git clone https://github.com/your-repo/Comfyui-PaddleOCR.git

2. Install required dependencies:

pip install paddleocr pillow opencv-python numpy torch torchvision

3. Restart ComfyUI to load the new nodes.

Font Management

The package includes a comprehensive font library located in nodes/font_dir/:

• Multiple Chinese fonts (华文中宋, 华文仿宋, 华文彩云, etc.)
• Microsoft YaHei (微软雅黑) regular and bold
• Support for TTF and OTF font formats

To add custom fonts:

1. Place font files (.ttf or .otf) in the nodes/font_dir/ directory
2. Restart ComfyUI to refresh the font list

Use Cases

• Document Processing: Extract and analyze text from scanned documents
• Image Translation: OCR text extraction followed by translation overlay
• Content Creation: Add styled text overlays to images
• Data Extraction: Automated text extraction from images for data processing
• Multilingual Content: Handle text in various languages with appropriate fonts

Requirements

• ComfyUI
• Python 3.8+
• PaddleOCR
• PIL (Pillow)
• OpenCV
• NumPy
• PyTorch

Contributing

Contributions are welcome! Please feel free to submit issues, feature requests, or pull requests.

License

This project is licensed under the MIT License - see the LICENSE file for details.