ComfyUI Node
Region Feature Similarity
A ComfyUI node in ✨ LF Nodes/Analytics with 7 inputs and 1 output.
Region Feature Similarity
- features_a
- features_b
- masks
- metadata
- ui_widget
- similarity
◄model_filedino_v3_vit_h.safetensors►
◄cropnone►
Category✨ LF Nodes/Analytics
Inputs (7)
| Name | Type | Default | Description |
|---|---|---|---|
| features_a | CLIP_VISION_OUTPUT | Core CLIPVisionEncode output for A; pooled-only outputs must encode ordered region crops. | |
| features_b | CLIP_VISION_OUTPUT | Core CLIPVisionEncode output for B using the same model and crop setting. | |
| masks | MASK | Text Region Masks batch/list: one shared set or two sets in A/B order. | |
| metadata | JSON | Matching lf.text-region-masks.v1 metadata, in the same order as masks. | |
| model_file | STRING | dino_v3_vit_h.safetensors | Filename selected by CLIPVisionLoader, recorded as supplied by the workflow. |
| crop | COMBO | none | Must match both CLIPVisionEncode crop controls; none retains the full image. |
| ui_widgetopt | LF_TREE | [object Object] | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| similarity | JSON | Per-region feature cosine similarity, model filename and pooling method. |