https://github.com/guerreiro/comfyg-switch
Do you have a lot of models? Is it boring to change each config field for every testing?
https://github.com/guerreiro/comfyg-switch
ai comfyui comfyui-custom-node comfyui-nodes plugin python stable-diffusion user-experience
Last synced: 5 days ago
JSON representation
Do you have a lot of models? Is it boring to change each config field for every testing?
- Host: GitHub
- URL: https://github.com/guerreiro/comfyg-switch
- Owner: guerreiro
- Created: 2025-02-05T11:02:19.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2025-04-12T15:04:15.000Z (5 days ago)
- Last Synced: 2025-04-12T16:23:36.731Z (5 days ago)
- Topics: ai, comfyui, comfyui-custom-node, comfyui-nodes, plugin, python, stable-diffusion, user-experience
- Language: Python
- Homepage:
- Size: 34.2 KB
- Stars: 2
- Watchers: 1
- Forks: 2
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-comfyui - **Comfyg Switch** - specific settings from a JSON file (model_configs.json). (All Workflows Sorted by GitHub Stars)
README
# Comfyg Switch
Comfyg Switch is a custom node that dynamically selects model configuration parameters based on the chosen checkpoint.
It reads model-specific settings from a JSON file (model_configs.json).
## Inputs
- checkpoint_model: This value is used to determine which configuration to load.
- use_custom_input (BOOLEAN): Toggle between using manual inputs and automatically loaded configurations.
- steps (INT): Number of inference steps (default: 30).
- refiner_steps (INT): Number of inference steps for enhancement (default: 30).
- cfg (FLOAT): Classifier-free guidance scale (default: 7.0).
- sampler (SAMPLER)
- scheduler (SCHEDULER)
## Outputs
- MODEL_NAME
- STEPS (INT)
- REFINE_STEPS (INT)
- CFG (FLOAT)
- SAMPLER (SAMPLER)
- SCHEDULER (SCHEDULER)
## Example
See the file: ComfygSwitch-example.json
# Roadmap (or ideas)
- Maybe load configs from an external database or something like that, to avoid update the config file everytime;
- Import each model config dinamically from CivitAI API and use the config file as optional (maybe use LLM to read the model content and create the config object);
- When switch the model, load the config data into the node inputs (steps, cfg, etc...) and let us see the values before start queue, or change them too;
- Load more details about the selected model to help us with the workflow (result examples, tips, prompts...);
# Contributing
Contributions and suggestions are welcome! If you encounter any issues or have ideas for improvements, please open an issue or submit a pull request.
# License
This project is licensed under the MIT License.