How to do sequential SEGS?

[mr-september]

I am playing around with your MakeTileSEGS_upscale_sdxl-lightning_ipadapter.png workflow.

I have an idea to expand it, so that the face is segmented by itself. Without knowing more about how the nodes work, I am thinking this:

1. Use the Simple Detector (SEGS) node to split background and subject (yolo_person) (unchanged from your tutorial).
2. Use another Simple Detector (SEGS) on the subject output to split face (yolo_face) and rest of the body.

However, there seems to be only 1 output from Simple Detector (SEGS): "SEGS". It doesn't seem to pipe the image, so it can't work sequentially?

Or perhaps there is another "correct" way to do this?

[ltdrdata]

Since yolo_face already detects only the face area, what’s the reason for trying to combine the two?

[mr-september]

I want to process the face, the body, and the background all differently. Use IPAdapter Face to maintain identity, while the body gets upscaled with ControlNets, allowing for higher denoise to fix very blur/lowres images.