- Python 53.7%
- JavaScript 38%
- CSS 8.3%
|
|
||
|---|---|---|
| .github/workflows | ||
| web | ||
| __init__.py | ||
| LICENSE | ||
| README.md | ||
ComfyUI-Librarian
ComfyUI-Librarian is a ComfyUI extension that caches model files when ComfyUI loads them.
Use this if you have your model library on a big slow drive but have a smaller faster drive available to keep recently used models.
Use for spinning HDD, external drive or network storage paired with a local NVME or fast SSD.
Used models will remain cached on the fast disk until the cache storage is exhausted. Models remain cached even through a comfyUI restart.
https://github.com/user-attachments/assets/79adbca6-aff3-4328-886c-5eca05850e4e
Installation
Clone into you custom_nodes folder:
git clone https://github.com/Comfy-Org/ComfyUI-Librarian ComfyUI/custom_nodes/ComfyUI-Librarian
Restart ComfyUI and refresh your comfy browser tab if already running.
Usage
On first usage the configuration will popup. Navigate to your fast drive with "browse" and set the amount of storage to budget for the cache. 256GB is a good size. Smaller is ok if running workflows repeatedly. Larger is better if jumping around lots of models.
Then use ComfyUI normally. No extra nodes needed. (this is integrated with the standard native file loader).
The Librarian configuration menu is available in the main ComfyUI menu if the configuration does not automatically pop up.