guojijun/ComfyUI_GJJ_Nodes

What it solves

ComfyUI_GJJ_Nodes is a massive collection of 328 custom nodes for ComfyUI designed to reduce dependency on third-party node packs while expanding capabilities across image, video, audio, and workflow management. It simplifies the user experience by providing built-in Chinese tooltips, intuitive 1-based slot numbering, and consolidated "all-in-one" nodes that replace complex multi-node workflows.

How it works

The project implements a wide array of specialized nodes that handle diverse AI tasks. It uses a "zero external dependency" philosophy (relying only on standard requirements) and includes its own internal runtime for complex models like WanVideo. It features advanced model discovery that recursively searches subdirectories and a stable "AnySwitch" mode for dynamic input/output slots to prevent connection loss during changes.

Who it’s for

It is designed for ComfyUI users—particularly those in the Chinese-speaking community—who want a comprehensive, streamlined toolkit for generative AI without the overhead of managing numerous disparate custom node repositories.

Highlights

• Massive Node Library: 328 registered nodes covering image processing, video generation, audio synthesis, and prompt engineering.
• Consolidated Workflows: "All-in-one" nodes for LTX2.3 and Wan2.2 video generation, reducing the need for sprawling node graphs.
• Advanced Model Management: Recursive subdirectory searching for models and a LoRA management panel with metadata and thumbnail previews.
• Multimodal Capabilities: Integrated support for voice cloning (CosyVoice, Fish Audio), lip-syncing (LatentSync, Bernini), and 3D Gaussian Splatting (TripoSplat).
• User-Centric Design: Built-in Chinese tooltips and a custom doodle canvas for direct sketching and ControlNet generation.