cardenluo/ComfyUI-Apt_Preset

What it solves

This project addresses the complexity and "dense" connections of ComfyUI workflows, which often become difficult to operate and maintain. It simplifies the workflow by organizing nodes into a structured "Loader $\to$ Controller $\to$ Sampler" framework to improve consistency and reduce the need to relearn operation methods for every new saved workflow.

How it works

The plugin introduces specialized nodes that group related functions together to eliminate repetitive connections:

• Universal Loaders: A single node (Sum_load_adv) that supports various model combinations (XL, SD, Flux, SD3.5, etc.) and GGUF models, with the ability to save parameter presets.
• Controller Stacks: Centralized functional modules for specific tasks, such as image control (IPA, Redux, ControlNet), video generation (Wan), and animation (AnimateDiff).
• Simplified Samplers: Basic samplers that package built-in ComfyUI ports into a single port, and special function samplers that combine sampling with tasks like secondary repair or refine enlargement in one step.
• Utility Nodes: Dedicated tools for processing data, images, and masks (including segmentation for face, clothes, and body).

Who it’s for

Users of ComfyUI who want to organize their node-based workflows more efficiently and reduce visual clutter while maintaining support for a wide range of modern AI image and video generation models.

Highlights

• Preset System: Save and reuse model and sampling configurations as presets to avoid manual reconfiguration.
• Unified Loader: Supports a vast array of model types (SD, Flux, Wan, QwenEdit, etc.) in one node.
• Crammed Control Stacks: Groups complex control tools (like ControlNet and IPA) into single, manageable stacks.
• Integrated Image/Mask Processing: Includes built-in tools for pre-processing and post-recovery of images and masks.