Extensions/ComfyUI-SmartModelResolver
ComfyUI Extension

ComfyUI-SmartModelResolver

An intelligent, non-intrusive model locator and subfolder auto-linker for ComfyUI

By tahakhorsand·Created 2 days ago·Updated 2 days ago· 1
tahakhorsand/ComfyUI-SmartModelResolver
Nodes1
On cloudLocal install
CategorySmartModelResolver
Stars1
Updated2 days ago

Nodes (1)

Smart Model Path Finder (Deep Subfolder Locator)

The node that tells you where your models actually live

SmartModelResolver
Readme

ComfyUI-SmartModelResolver

An intelligent, lightning-fast model locator and subfolder auto-linker for ComfyUI.

Eliminates workflow loading failures, missing subfolder paths, and red widget errors with a modern, glassmorphic UI.

📸 Preview & Interaction

1. ⚡ Consolidated Batch Resolver

When opening a complex workflow with multiple missing models or subfolders, SmartModelResolver groups all issues into a single, non-blocking glassmorphic card. Review exact subfolder matches and similar suggestions with checkboxes, then link them all in one click:

<p align="center"> <img src="assets/consolidated_dialog_preview.png" alt="Consolidated Smart Model Resolver Dialog" width="850" /> </p>

2. 💡 Intelligent Similarity Detection

When a workflow specifies a slightly different revision or naming format (e.g., _pruned or revision differences), the assistant highlights authentic candidates in your local library with match percentage badges:

<p align="center"> <img src="assets/similar_model_card_preview.png" alt="Similar Model Suggestion Card" width="850" /> </p>

3. ✅ Seamless Canvas Linking

Selected paths are instantly mapped directly to your graph and subgraphs without reloading ComfyUI, leaving your nodes clean and ready to queue:

<p align="center"> <img src="assets/auto_resolved_success_preview.png" alt="Successfully Resolved Models" width="850" /> </p>

✨ Features

  • 🔍 Automatic Subfolder Resolution: Instantly discovers and links models stored inside local subfolders.
  • 💡 Smart Model Suggestions: When a workflow requests a model you do not have (e.g., minimax_h3_fl2va_pruned_int8_convrot), the assistant detects authentic alternatives in your library (e.g., minimax_h3_fl2va_int8_convrot with 89%+ match) and lets you replace them in one click.
  • ⚡ Consolidated Batch Resolver: Groups all missing models and suggestions into a single unified card with selection checkboxes and a single-click Replace Selected button.
  • ⚡ Blazing Fast Indexing (<50ms): Directly leverages ComfyUI's native in-memory cache system and background daemon threads with 0ms startup lag.
  • 🧩 Full Subgraph / Group Node Support: Recursively traverses both root graphs and inner subgraphs, keeping promoted inputs and internal definitions perfectly in sync.
  • 🛡️ Strict File Protection: Only matches authentic neural model formats (.safetensors, .ckpt, .pt, .pth, .bin, .sft). Never matches .py, .json, or non-model files.

🚀 Installation

  1. Open a terminal inside your ComfyUI custom_nodes directory: 
    git clone https://github.com/tahakhorsand/ComfyUI-SmartModelResolver.git
  2. Restart ComfyUI.
  3. Open any workflow. If models are in subfolders or close alternatives exist, the glassmorphic assistant will appear in the top-right corner.

⚙️ How It Works

  1. Backend Interceptor: Transparently hooks into folder_paths.get_full_path to resolve paths on-the-fly during execution without breaking native ComfyUI caching.
  2. Frontend Sync: Automatically updates node widgets across graphs and subgraphs, redrawing the canvas seamlessly.
  3. Smart Similarity Engine: Employs token-aware string similarity to suggest genuine model revisions without false positives.

📄 License

This project is licensed under the GNU General Public License v3.0 (GPL-3.0).