ComfyUI-JoyCaption

ComfyUI-JoyCaption

Joy Caption is a ComfyUI custom node powered by the LLaVA model for efficient, stylized image captioning. Caption Tools nodes handle batch image processing and automatic separation of caption text.

Joycaption_node

News & Updates

• 2025/10/27: Added Custom Models v2.0.2 ( update.md )

• 2025/10/20: Update and bug fix ComfyUI-JoyCaption to v2.0.1 ( update.md )

• 2025/08/22: Update ComfyUI-JoyCaption to v2.0.0 ( update.md ) Joycaption GGUF

  • Added GGUF model support for better performance and compatibility
  • Fixed critical parameter handling issues (top_k, image format)
  • Improved console output (no more base64 spam)
  • Enhanced error handling and stability
  • Added JoyCaption GGUF and JoyCaption GGUF (Advanced) nodes
  • Enhanced GGUF Support: Comprehensive support for 12 quantization levels (Q2_K to F16)
  • llama_cpp_install folder: Complete installation guides and automated scripts for llama-cpp-python
  • Simplified Installation: One-click llama-cpp-python installation with automatic CUDA support
  • Cross-Platform Support: Windows, macOS, and Linux installation guides
  • Performance Improvements: Optimized model loading and memory management

• 2025/06/15: Update ComfyUI-JoyCaption to v1.2.0 ( update.md )

  • Enhanced CUDA performance optimization
  • Improved memory management strategies
  • Added Global Cache mode for faster processing

v1 2 0

• 2025/06/07: Update ComfyUI-JoyCaption to v1.1.1 ( update.md )

• 2025/06/05: Update ComfyUI-JoyCaption to v1.1.0 ( update.md )

• Add Caption tools

Joycaption_node

Features

• Simple and user-friendly interface
• Multiple caption types support
• Flexible length control
• Memory optimization options
• GGUF model support - Efficient quantized models for better performance and lower memory usage
• Automatic model download - Models will be downloaded automatically and properly renamed on first use
• Support for multiple caption types
• Support for advanced customization options
• Multiple precision options (fp32, bf16, fp16, 8-bit, 4-bit)
• Enhanced memory management with Global Cache mode
• Clean console output (no debug spam)
• Robust error handling and parameter validation
• llama_cpp_install folder - Complete installation guides and automated scripts for llama-cpp-python

Installation

1. Clone this repository to your ComfyUI/custom_nodes directory:

cd ComfyUI/custom_nodes
git clone https://github.com/1038lab/ComfyUI-JoyCaption.git

2. Install the required dependencies:

cd ComfyUI/custom_nodes/ComfyUI-JoyCaption
pip install -r requirements.txt

3. For GGUF Models (Recommended): Install llama-cpp-python with CUDA support:

# Option 1: Automated installation (Recommended)
python llama_cpp_install/llama_cpp_install.py

# Option 2: Manual installation
pip install -r requ