ComfyUI Node
Person Selector (Match)
Matches a reference person in the current image using InsightFace (ArcFace) embeddings. Outputs similarity score, match boolean, and optional mask (face/head/body). Body mask requires SAM model from Impact Pack SAMLoader. Face/head masks use BiSeNet segmentation. threshold: minimum cosine similarity for a match (0.6-0.7 recommended) aggregation: how to combine scores across multiple reference images det_size: face detection resolution — higher finds smaller faces but uses more VRAM
Person Selector (Match)
- current_image
- reference_images
- sam_model
- similarity
- match
- mask
- best_reference
- face_count
- matched_face_index
- report
◄threshold0.65►
◄aggregation▾►
◄mask_mode▾►
◄mask_fill_holestrue►
◄mask_blur0►
◄det_size▾►
CategoryFVM Tools/Face
Inputs (9)
| Name | Type | Default | Description |
|---|---|---|---|
| current_image | IMAGE | — | |
| reference_images | IMAGE | — | |
| threshold | FLOAT | 0.650–1 | Minimum cosine similarity to count as match (0.6-0.7 recommended) |
| aggregation | COMBO | How to combine similarity scores across multiple reference images | |
| mask_mode | COMBO | none=no mask, face=skin only, head=face+hair, body=full body (needs SAM) | |
| mask_fill_holes | BOOLEAN | true | Fill holes inside the mask (closes gaps in segmentation) |
| mask_blur | INT | 00–100 | Gaussian blur radius for mask edges |
| det_size | COMBO | Face detection resolution — higher finds smaller faces but uses more VRAM | |
| sam_modelopt | SAM_MODEL | (opt) SAM model from Impact Pack SAMLoader — required for body mask mode |
Outputs (7)
| Name | Type | Description |
|---|---|---|
| similarity | FLOAT | — |
| match | BOOLEAN | — |
| mask | MASK | — |
| best_reference | IMAGE | — |
| face_count | INT | — |
| matched_face_index | INT | — |
| report | STRING | — |