Image scoring nodes for ComfyUI using PickScore to predict which images in a batch best fit a given prompt.
Clone the repository to custom_nodes:
git clone https://github.com/Zuellni/ComfyUI-PickScore-Nodes custom_nodes/ComfyUI-PickScore-Nodes
| Name | Description |
|---|---|
| Loader | Loads scoring models from Hugging Face or a given directory. The default one is around 4GB. |
| Processor | Takes images/text and converts them to inputs for the Selector node. |
| Selector | Selects up to limit best images and passes them to other nodes along with latents/masks of the same shape. Interrupts processing if the threshold isn't reached. |
The example workflow is embedded in the image below and can be opened in ComfyUI.
