FC_Segment
FC Segment — the plain face mask, no options attached
- source_image
- MASK
FC Segment is the simplest segmentation node in ComfyUI-FaceChain: image in, one face mask out, no parameters to touch. It runs the same human-parsing model as the pack's full-featured FC FaceSegment node, but strips away every option - no ksize expansion, no include_neck, no warp mask, no second human mask. Just a clean face mask, exactly once.
It's one of the pack's older/underscore-class names. In the current node set, FC FaceSegment is the segmentation workhorse and returns three outputs (seg_image, soft_mask, human_mask) with expansion controls; this class is the minimal predecessor. You'll hit it in old workflow JSONs and older tutorials, and the fix is almost always "swap in FC FaceSegment."
The schema
- source_image (
IMAGE) - the photo to segment. Required, and the only input. - MASK output - the face mask for the image.
That's the entire surface area. The engine is ModelScope's damo/cv_resnet101_image-multiple-human-parsing pipeline, downloaded on first use from modelscope.cn - the same model FC FaceSegment uses, just called without any of the expansion knobs, so you get the tight, default face region.
What to do with it
If you're on a modern graph: replace it with FC FaceSegment. The current node gives you the same mask plus the ability to grow it (ksize, ksize1), include the neck, and get the whole-person human_mask - all strictly more useful, and its soft_mask output is a drop-in for this node's MASK.
If you're stuck with it in an old workflow and just need it to run: it works as-is. Feed the MASK into FC ReplaceByMask for a masked paste or into FC MaskOP if you want to blur or expand the edge afterward - the mask-editing nodes give you back the control this node omits.
Install
Same pack either way: ComfyUI Manager (search "ComfyUI-FaceChain") or git clone https://github.com/THtianhao/ComfyUI-FaceChain into custom_nodes/, restart. The segmentation model downloads on first run; the pack-wide startup dependency install (onnxruntime-gpu, mmdet, mmcv, modelscope, insightface, transformers…) applies to the whole pack regardless of which node you actually use.
The short version: it's a historical skeleton of FC FaceSegment, useful to recognize and quick to migrate off of. Don't build anything new on it.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| source_image | IMAGE | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| MASK | MASK | — |