chflame163/ComfyUI_LayerStyle_Advance

The nodes detached from [ComfyUI Layer Style](https://github.com/chflame163/ComfyUI_LayerStyle) are mainly those with complex requirements for dependency packages.

What it solves

This project provides a set of advanced custom nodes for ComfyUI, specifically designed to handle complex image processing tasks that require heavy dependency packages. It extends the basic functionality of ComfyUI by adding capabilities for high-quality image captioning, visual language model (VLM) inference, and advanced image composition.

How it works

It integrates various specialized AI models and APIs into the ComfyUI node-based workflow. This includes:

  • Local VLM Inference: Implementing models like Llama 3.2 Vision, JoyCaption2, and SmolVLM for image-to-text tasks.
  • API Integration: Connecting to external services like DeepSeek, Gemini, ZhipuGLM4, and Jimeng for text and visual inference.
  • Image Manipulation: Providing tools for creating image collages and utilizing SAM2 (Segment Anything Model 2) for precise object masking.
  • Specialized Processing: Using models like Florence-2 for object recognition and cropping.

Who it’s for

  • ComfyUI users who want to automate the generation of detailed prompts from images (image-to-prompt).
  • Digital artists and AI creators who need advanced layering, masking, and image composition tools within their node-based workflows.
  • Users looking to integrate state-of-the-art VLMs and LLMs into their image generation pipelines.

Highlights

  • Advanced Captioning: Includes JoyCaption2 for highly detailed, customizable image descriptions (e.g., specifying lighting, camera angle, or aesthetic quality).
  • Multi-Model Support: Supports a wide range of local and API-based models including Llama 3.2 Vision, Qwen, and Gemini 2.0 Flash.
  • Smart Composition: A Collage node that can use Florence-2 for intelligent object recognition and cropping when creating image grids.
  • Efficient Resource Management: Split model loading nodes (like JoyCaption2Split) to share models across multiple nodes and reduce VRAM usage.

Related

  • Project
    aimclub/FEDOT

    FEDOT is an open‑source Python AutoML framework that automatically designs and optimises machine‑learning pipelines (including preprocessing, feature engineering and models) using evolutionary algorithms. It supports classification, regression, clustering, and time‑series forecasting, integrates with popular ML libraries, offers a simple high‑level API, and provides reproducible pipeline export.

  • Project
    X-GenGroup/Flow-Factory

    A unified framework for online RL and offline fine-tuning of diffusion and flow-matching models across image, video, and audio-video modalities.

  • Project
    NomaDamas/CozyClay

    A browser-based previs studio for blocking scenes and posing characters to create high-control reference assets for AI video generation.

  • Project
    pnnx/pnnx

    pnnx is an open‑source tool that optimises PyTorch models and exports them to a lightweight, dependency‑free format (PNNX) and to ncnn/ONNX‑zero files, enabling fast inference on edge devices.