ComfyUI Extension: Hermes Nodes
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.
WAN Dual LoRA Loader
README
Hermes Nodes
A collection of useful custom nodes for ComfyUI.
Feature requests and bug reports are welcome - please open an issue on GitHub.
Installation
Clone the repo to custom_nodes folder
Nodes
WAN Dual LoRA Loader
A specialized LoRA loader designed for WAN 2.2 architecture, which uses separate HIGH and LOW noise models.

Key Features
- Two-Column Layout - Separate LoRA lists for HIGH and LOW noise models
- I2V/T2V Folder Structure - Organize LoRAs by type:
models/loras/wan/ ├── I2V/ │ ├── HIGH/ │ └── LOW/ └── T2V/ ├── HIGH/ └── LOW/ - Shared Folders - LoRAs placed in
I2V/orT2V/root (not in HIGH/LOW subfolders) appear in both lists - Preset System - Save and load LoRA+prompt configurations
- Multiple Preset Lists - Organize presets into categories (Default, I2V, T2V)
- Wildcard Prompt Processing - Built-in wildcard support in the prompt field
- Auto-Match - Automatically finds matching LoRAs between HIGH/LOW folders
Usage
- Add "WAN Dual LoRA Loader" node (category:
loaders) - Connect your WAN HIGH and LOW models
- Select folder type (I2V or T2V)
- Click "+ Add" to add LoRAs to each column
- Use the preset system to save/load configurations
- Enter your prompt (supports wildcards)
Wildcard Syntax
{option1|option2|option3}- Random choice from options__filename__- Random line fromwildcards/filename.txt<random:0.5:1.0>- Random float in range
Outputs
| Output | Description |
|--------|-------------|
| model_high | HIGH model with LoRAs applied |
| model_low | LOW model with LoRAs applied |
| prompt | Processed prompt with resolved wildcards |
License
MIT License
Run ComfyUI workflows without the setup
No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.