BiRefNet 是什么？
BiRefNet 是用于背景去除和抠图的模型。
正式来说，它是用于 Dichotomous Image Segmentation (DIS) 的模型，也就是把图像分成 前景 和 背景 两部分。
它并不是像 SAM 那样，通过“这个点”“这个框”“这个物体”来指定并生成蒙版。BiRefNet 更擅长从头发、植物等细节复杂的对象中，生成高质量的切出蒙版。
模型下载
- birefnet.safetensors (444 MB)
📂ComfyUI/
└── 📂models/
└── 📂background_removal/
└── birefnet.safetensors
workflow
背景去除
{
"id": "32d5b916-1532-435e-8274-bab16b8cce42",
"revision": 0,
"last_node_id": 7,
"last_link_id": 8,
"nodes": [
{
"id": 1,
"type": "RemoveBackground",
"pos": [
1160.3888735981782,
421.65467672131837
],
"size": [
204.705078125,
46
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 8
},
{
"name": "bg_removal_model",
"type": "BACKGROUND_REMOVAL",
"link": 1
}
],
"outputs": [
{
"name": "mask",
"type": "MASK",
"links": [
3
]
}
],
"properties": {
"Node name for S&R": "RemoveBackground"
},
"widgets_values": [],
"color": "#232",
"bgcolor": "#353"
},
{
"id": 2,
"type": "LoadImage",
"pos": [
754.5683630708619,
179.7700993500816
],
"size": [
327.2964755928126,
350.7897563865719
],
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
8
]
},
{
"name": "MASK",
"type": "MASK",
"links": null
}
],
"properties": {
"Node name for S&R": "LoadImage"
},
"widgets_values": [
"0088_amodal_rebuild001_0089 (1).png",
"image"
]
},
{
"id": 3,
"type": "LoadBackgroundRemovalModel",
"pos": [
756.5990139230331,
598.8218881150212
],
"size": [
325.9727446183857,
58
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "bg_model",
"type": "BACKGROUND_REMOVAL",
"links": [
1
]
}
],
"properties": {
"Node name for S&R": "LoadBackgroundRemovalModel"
},
"widgets_values": [
"birefnet.safetensors"
],
"color": "#232",
"bgcolor": "#353"
},
{
"id": 4,
"type": "MaskPreview",
"pos": [
1442.8048400402731,
175.32623268583868
],
"size": [
477.1221995094718,
480.5791751839481
],
"flags": {},
"order": 3,
"mode": 0,
"inputs": [
{
"name": "mask",
"type": "MASK",
"link": 3
}
],
"outputs": [],
"properties": {
"Node name for S&R": "MaskPreview"
},
"widgets_values": []
}
],
"links": [
[
1,
3,
0,
1,
1,
"BACKGROUND_REMOVAL"
],
[
3,
1,
0,
4,
0,
"MASK"
],
[
8,
2,
0,
1,
0,
"IMAGE"
]
],
"groups": [],
"config": {},
"extra": {
"ds": {
"scale": 0.9229599817706424,
"offset": [
-436.5441715227812,
211.37047252024877
]
},
"frontendVersion": "1.44.19"
},
"version": 0.4
}
- 使用
Load Background Removal Model读取birefnet.safetensors。 - 将图像和模型输入
Remove Background后，会输出用于背景去除的MASK。