model parameters

What parameters are working for you and what do you think why (e.g. bright / dark room, webcam resolution, etc.)?

---

Example Config:

• internal_resolution = 0.5
• multiplier = 0.5
• segmentation_threshold = 0.75

Works good with:

• 1280x720 Webcam
• artifical lighting
• untidy background
• Person in the middle of the image, not filling the whole image.

---

Configuration for https://github.com/fangfufu/Linux-Fake-Background-Webcam

segmentation_threshold: 0.75
blur: 30
erode: 0
dilate: 20
virtual_video_device: "/dev/video2"
real_video_device: "/dev/video0"
average_masks: 3
layers:
- "empty": [["image", "background.jpg"]]
- "foreground": []
multiplier: 0.75

The default parameters of Virtual Webcam Background seem to oversegment my face. I can't see my ear and parts of my hair. I wonder if it is sensible to default back to original blog post's parameter. I don't like the abrupt transition between the edge of the foreground and background, hence the dilation to make the mask bigger, follow by blur to smooth things out. (by fangfufu)