nodes/MoonHugo-ComfyUI-BiRefNet-Hugo
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

add load local model way

Browse Source
This commit is contained in:
yueliang
2024-09-21 18:11:58 +08:00
parent 37a20e516c
commit 06b00c6135
9 changed files with 354 additions and 477 deletions
Download Patch File Download Diff File Expand all files Collapse all files
BiRefNet_node_config.py
-148
View File
@@ -1,148 +0,0 @@
### config.py
import os
import math
os.environ['HOME'] = os.path.expanduser("~")
class Config():
def __init__(self) -> None:
# PATH settings
self.sys_home_dir = os.environ['HOME'] # Make up your file system as: SYS_HOME_DIR/codes/dis/BiRefNet, SYS_HOME_DIR/datasets/dis/xx, SYS_HOME_DIR/weights/xx
# TASK settings
self.task = ['DIS5K', 'COD', 'HRSOD', 'DIS5K+HRSOD+HRS10K', 'P3M-10k'][0]
self.training_set = {
'DIS5K': ['DIS-TR', 'DIS-TR+DIS-TE1+DIS-TE2+DIS-TE3+DIS-TE4'][0],
'COD': 'TR-COD10K+TR-CAMO',
'HRSOD': ['TR-DUTS', 'TR-HRSOD', 'TR-UHRSD', 'TR-DUTS+TR-HRSOD', 'TR-DUTS+TR-UHRSD', 'TR-HRSOD+TR-UHRSD', 'TR-DUTS+TR-HRSOD+TR-UHRSD'][5],
'DIS5K+HRSOD+HRS10K': 'DIS-TE1+DIS-TE2+DIS-TE3+DIS-TE4+DIS-TR+TE-HRS10K+TE-HRSOD+TE-UHRSD+TR-HRS10K+TR-HRSOD+TR-UHRSD', # leave DIS-VD for evaluation.
'P3M-10k': 'TR-P3M-10k',
}[self.task]
self.prompt4loc = ['dense', 'sparse'][0]
# Faster-Training settings
self.load_all = True
self.compile = True # 1. Trigger CPU memory leak in some extend, which is an inherent problem of PyTorch.
# Machines with > 70GB CPU memory can run the whole training on DIS5K with default setting.
# 2. Higher PyTorch version may fix it: https://github.com/pytorch/pytorch/issues/119607.
# 3. But compile in Pytorch > 2.0.1 seems to bring no acceleration for training.
self.precisionHigh = True
# MODEL settings
self.ms_supervision = True
self.out_ref = self.ms_supervision and True
self.dec_ipt = True
self.dec_ipt_split = True
self.cxt_num = [0, 3][1] # multi-scale skip connections from encoder
self.mul_scl_ipt = ['', 'add', 'cat'][2]
self.dec_att = ['', 'ASPP', 'ASPPDeformable'][2]
self.squeeze_block = ['', 'BasicDecBlk_x1', 'ResBlk_x4', 'ASPP_x3', 'ASPPDeformable_x3'][1]
self.dec_blk = ['BasicDecBlk', 'ResBlk', 'HierarAttDecBlk'][0]
# TRAINING settings
self.batch_size = 4
self.IoU_finetune_last_epochs = [
0,
{
'DIS5K': -50,
'COD': -20,
'HRSOD': -20,
'DIS5K+HRSOD+HRS10K': -20,
'P3M-10k': -20,
}[self.task]
][1] # choose 0 to skip
self.lr = (1e-4 if 'DIS5K' in self.task else 1e-5) * math.sqrt(self.batch_size / 4) # DIS needs high lr to converge faster. Adapt the lr linearly
self.size = 1024
self.num_workers = max(4, self.batch_size) # will be decrease to min(it, batch_size) at the initialization of the data_loader
# Backbone settings
self.bb = [
'vgg16', 'vgg16bn', 'resnet50', # 0, 1, 2
'swin_v1_t', 'swin_v1_s', # 3, 4
'swin_v1_b', 'swin_v1_l', # 5-bs9, 6-bs4
'pvt_v2_b0', 'pvt_v2_b1', # 7, 8
'pvt_v2_b2', 'pvt_v2_b5', # 9-bs10, 10-bs5
][6]
self.lateral_channels_in_collection = {
'vgg16': [512, 256, 128, 64], 'vgg16bn': [512, 256, 128, 64], 'resnet50': [1024, 512, 256, 64],
'pvt_v2_b2': [512, 320, 128, 64], 'pvt_v2_b5': [512, 320, 128, 64],
'swin_v1_b': [1024, 512, 256, 128], 'swin_v1_l': [1536, 768, 384, 192],
'swin_v1_t': [768, 384, 192, 96], 'swin_v1_s': [768, 384, 192, 96],
'pvt_v2_b0': [256, 160, 64, 32], 'pvt_v2_b1': [512, 320, 128, 64],
}[self.bb]
if self.mul_scl_ipt == 'cat':
self.lateral_channels_in_collection = [channel * 2 for channel in self.lateral_channels_in_collection]
self.cxt = self.lateral_channels_in_collection[1:][::-1][-self.cxt_num:] if self.cxt_num else []
# MODEL settings - inactive
self.lat_blk = ['BasicLatBlk'][0]
self.dec_channels_inter = ['fixed', 'adap'][0]
self.refine = ['', 'itself', 'RefUNet', 'Refiner', 'RefinerPVTInChannels4'][0]
self.progressive_ref = self.refine and True
self.ender = self.progressive_ref and False
self.scale = self.progressive_ref and 2
self.auxiliary_classification = False # Only for DIS5K, where class labels are saved in `dataset.py`.
self.refine_iteration = 1
self.freeze_bb = False
self.model = [
'BiRefNet',
][0]
if self.dec_blk == 'HierarAttDecBlk':
self.batch_size = 2 ** [0, 1, 2, 3, 4][2]
# TRAINING settings - inactive
self.preproc_methods = ['flip', 'enhance', 'rotate', 'pepper', 'crop'][:4]
self.optimizer = ['Adam', 'AdamW'][1]
self.lr_decay_epochs = [1e5] # Set to negative N to decay the lr in the last N-th epoch.
self.lr_decay_rate = 0.5
# Loss
self.lambdas_pix_last = {
# not 0 means opening this loss
# original rate -- 1 : 30 : 1.5 : 0.2, bce x 30
'bce': 30 * 1, # high performance
'iou': 0.5 * 1, # 0 / 255
'iou_patch': 0.5 * 0, # 0 / 255, win_size = (64, 64)
'mse': 150 * 0, # can smooth the saliency map
'triplet': 3 * 0,
'reg': 100 * 0,
'ssim': 10 * 1, # help contours,
'cnt': 5 * 0, # help contours
'structure': 5 * 0, # structure loss from codes of MVANet. A little improvement on DIS-TE[1,2,3], a bit more decrease on DIS-TE4.
}
self.lambdas_cls = {
'ce': 5.0
}
# Adv
self.lambda_adv_g = 10. * 0 # turn to 0 to avoid adv training
self.lambda_adv_d = 3. * (self.lambda_adv_g > 0)
# PATH settings - inactive
self.data_root_dir = os.path.join(self.sys_home_dir, 'datasets/dis')
self.weights_root_dir = os.path.join(self.sys_home_dir, 'weights')
self.weights = {
'pvt_v2_b2': os.path.join(self.weights_root_dir, 'pvt_v2_b2.pth'),
'pvt_v2_b5': os.path.join(self.weights_root_dir, ['pvt_v2_b5.pth', 'pvt_v2_b5_22k.pth'][0]),
'swin_v1_b': os.path.join(self.weights_root_dir, ['swin_base_patch4_window12_384_22kto1k.pth', 'swin_base_patch4_window12_384_22k.pth'][0]),
'swin_v1_l': os.path.join(self.weights_root_dir, ['swin_large_patch4_window12_384_22kto1k.pth', 'swin_large_patch4_window12_384_22k.pth'][0]),
'swin_v1_t': os.path.join(self.weights_root_dir, ['swin_tiny_patch4_window7_224_22kto1k_finetune.pth'][0]),
'swin_v1_s': os.path.join(self.weights_root_dir, ['swin_small_patch4_window7_224_22kto1k_finetune.pth'][0]),
'pvt_v2_b0': os.path.join(self.weights_root_dir, ['pvt_v2_b0.pth'][0]),
'pvt_v2_b1': os.path.join(self.weights_root_dir, ['pvt_v2_b1.pth'][0]),
}
# Callbacks - inactive
self.verbose_eval = True
self.only_S_MAE = False
self.use_fp16 = False # Bugs. It may cause nan in training.
self.SDPA_enabled = False # Bugs. Slower and errors occur in multi-GPUs
# others
self.device = [0, 'cpu'][0] # .to(0) == .to('cuda:0')
self.batch_size_valid = 1
self.rand_seed = 7
# run_sh_file = [f for f in os.listdir('.') if 'train.sh' == f] + [os.path.join('..', f) for f in os.listdir('..') if 'train.sh' == f]
# with open(run_sh_file[0], 'r') as f:
# lines = f.readlines()
# self.save_last = int([l.strip() for l in lines if '"{}")'.format(self.task) in l and 'val_last=' in l][0].split('val_last=')[-1].split()[0])
# self.save_step = int([l.strip() for l in lines if '"{}")'.format(self.task) in l and 'step=' in l][0].split('step=')[-1].split()[0])
# self.val_step = [0, self.save_step][0]
README.md
+13 -14
View File
@@ -22,6 +22,16 @@ ___
示例工作流放置在`ComfyUI-BiRefNet-Hugo/workflow`中<br/>
The demo workflow placed in `ComfyUI-BiRefNet-Hugo/workflow`
加载模型支持两种方式，一种是自动下载远程模型并加载模型，另外一种是加载本地模型。加载本地模型的时候需要把load_local_model设置为true，并把local_model_path设置为本地模型所在路径，例如：H:\ZhengPeng7\BiRefNet<br/>
Loading the model supports two methods: one is to automatically download and load a remote model, and the other is to load a local model. When loading a local model, you need to set 'load_local_model' to true and 'local_model_path' to the path where the local model is located, for example: H:\ZhengPeng7\BiRefNet.
![](./assets/e21c32bf-ab98-444a-8055-54975ac47da3.png)
模型下载地址：https://huggingface.co/ZhengPeng7/BiRefNet/tree/main<br/>
Model download address: https://huggingface.co/ZhengPeng7/BiRefNet/tree/main
___
工作流workflow.json的使用<br/>
The use of workflow.json
@@ -41,6 +51,8 @@ ___
![](./assets/demo3.gif)
![](./assets/demo4.gif)
## 社交账号 | Social Account Homepage
- Bilibili：[我的B站主页](https://space.bilibili.com/1303099255)
@@ -49,20 +61,7 @@ ___
感谢BiRefNet仓库的所有作者 [ZhengPeng7/BiRefNet](https://github.com/zhengpeng7/birefnet)
Thanks to BiRefNet repo owner [ZhengPeng7/BiRefNet](https://github.com/zhengpeng7/birefnet)
```
library_name: birefnet
tags:
- background-removal
- mask-generation
- Dichotomous Image Segmentation:二分图像分割(DIS)，指分割成前景与背景，二个集合，分割出高精度效果。
- Camouflaged Object Detection:伪装物体检测(COD)，偏工程向，旨在识别“无缝”嵌入到周围环境中的物体，例如野生动物保护、军事侦察或者工业自动化。
- Salient Object Detection:显著性目标检测(SOD)，自动检测图像中最具视觉吸引力的部分。
- pytorch_model_hub_mixin
- model_hub_mixin
repo_url: https://github.com/ZhengPeng7/BiRefNet
pipeline_tag: image-segmentation
license: mit
```
部分代码参考了 [ZHO-ZHO-ZHO/ComfyUI-BiRefNet-ZHO](https://github.com/ZHO-ZHO-ZHO/ComfyUI-BiRefNet-ZHO) 感谢！
Some of the code references [ZHO-ZHO-ZHO/ComfyUI-BiRefNet-ZHO](https://github.com/ZHO-ZHO-ZHO/ComfyUI-BiRefNet-ZHO) Thanks!
__init__.py
+2 -1
View File
@@ -1,3 +1,4 @@
from .nodes import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS']
WEB_DIRECTORY = "./js"
__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS","WEB_DIRECTORY"]
assets/2de5b085-1125-46f9-8ef3-06706743f182.png
BIN
View File
Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 509 KiB

After

Width:  |  Height:  |  Size: 285 KiB

assets/d0a22b2a-ceb3-4205-9b4e-f6a68e4337c7.png
BIN
View File
Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 1.3 MiB

After

Width:  |  Height:  |  Size: 809 KiB

assets/demo3.gif
BIN
View File
Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 239 KiB

After

Width:  |  Height:  |  Size: 280 KiB

nodes.py
+21 -9
View File
@@ -4,19 +4,11 @@ from torchvision import transforms
import numpy as np
from PIL import Image
import torch.nn.functional as F
from .BiRefNet_node_config import Config
import comfy.model_management as mm
Config()
import os
torch.set_float32_matmul_precision(["high", "highest"][0])
birefnet = AutoModelForImageSegmentation.from_pretrained(
"ZhengPeng7/BiRefNet", trust_remote_code=True
)
transform_image = transforms.Compose(
[
transforms.Resize((1024, 1024)),
@@ -25,6 +17,11 @@ transform_image = transforms.Compose(
]
)
current_path = os.getcwd()
## ComfyUI portable standalone build for Windows
model_path = os.path.join(current_path, "ComfyUI"+os.sep+"models"+os.sep+"BiRefNet")
def tensor2pil(image):
return Image.fromarray(np.clip(255. * image.cpu().numpy().squeeze(), 0, 255).astype(np.uint8))
@@ -68,8 +65,12 @@ class BiRefNet_Hugo:
return {
"required": {
"image": ("IMAGE",),
"load_local_model": ("BOOLEAN", {"default": False}),
"background_color_name": (colors,{"default": "transparency"}),
"device": (["auto", "cuda", "cpu", "mps", "xpu", "meta"],{"default": "auto"})
},
"optional": {
"local_model_path": ("STRING", {"default":model_path}),
}
}
@@ -80,13 +81,24 @@ class BiRefNet_Hugo:
def background_remove(self,
image,
load_local_model,
device,
background_color_name,
*args, **kwargs
):
processed_images = []
processed_masks = []
device = get_device_by_name(device)
if load_local_model:
local_model_path = kwargs.get("local_model_path", model_path)
birefnet = AutoModelForImageSegmentation.from_pretrained(local_model_path,trust_remote_code=True)
else:
birefnet = AutoModelForImageSegmentation.from_pretrained(
"ZhengPeng7/BiRefNet", trust_remote_code=True
)
birefnet.to(device)
for image in image:
orig_image = tensor2pil(image)
workflow/video_workflow.json
+160 -155
View File
@@ -3,20 +3,32 @@
"last_link_id": 16,
"nodes": [
{
"id": 11,
"type": "VHS_LoadVideo",
"pos": [
840,
-90
],
"id": 13,
"type": "VHS_VideoCombine",
"pos": {
"0": 1830,
"1": -180
},
"size": [
210,
480
360,
928.4444444444445
],
"flags": {},
"order": 0,
"order": 2,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 14,
"label": "图像"
},
{
"name": "audio",
"type": "AUDIO",
"link": null,
"label": "音频"
},
{
"name": "meta_batch",
"type": "VHS_BatchManager",
@@ -31,122 +43,47 @@
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
12
],
"shape": 3,
"label": "图像",
"slot_index": 0
},
{
"name": "frame_count",
"type": "INT",
"name": "Filenames",
"type": "VHS_FILENAMES",
"links": null,
"shape": 3,
"label": "帧计数"
},
{
"name": "audio",
"type": "AUDIO",
"links": null,
"shape": 3,
"label": "音频"
},
{
"name": "video_info",
"type": "VHS_VIDEOINFO",
"links": null,
"shape": 3,
"label": "视频信息"
"label": "文件名"
}
],
"properties": {
"Node name for S&R": "VHS_LoadVideo"
"Node name for S&R": "VHS_VideoCombine"
},
"widgets_values": {
"video": "6月28日(1).mp4",
"force_rate": 0,
"force_size": "Disabled",
"custom_width": 512,
"custom_height": 512,
"frame_load_cap": 0,
"skip_first_frames": 0,
"select_every_nth": 1,
"choose video to upload": "image",
"frame_rate": 30,
"loop_count": 0,
"filename_prefix": "AnimateDiff",
"format": "video/h265-mp4",
"pix_fmt": "yuv420p10le",
"crf": 22,
"save_metadata": true,
"pingpong": false,
"save_output": true,
"videopreview": {
"hidden": false,
"paused": false,
"params": {
"frame_load_cap": 0,
"skip_first_frames": 0,
"force_rate": 0,
"filename": "6月28日(1).mp4",
"type": "input",
"format": "video/mp4",
"select_every_nth": 1
"filename": "AnimateDiff_00004.mp4",
"subfolder": "",
"type": "output",
"format": "video/h265-mp4",
"frame_rate": 30
},
"muted": false
}
}
},
{
"id": 1,
"type": "BiRefNet_Hugo",
"pos": [
1110,
-60
],
"size": {
"0": 210,
"1": 60
},
"flags": {},
"order": 1,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 12,
"slot_index": 0,
"label": "image"
}
],
"outputs": [
{
"name": "image",
"type": "IMAGE",
"links": [
14
],
"slot_index": 0,
"shape": 3,
"label": "image"
},
{
"name": "mask",
"type": "MASK",
"links": [
15
],
"slot_index": 1,
"shape": 3,
"label": "mask"
}
],
"properties": {
"Node name for S&R": "BiRefNet_Hugo"
}
},
{
"id": 14,
"type": "MaskToImage",
"pos": [
1410,
30
],
"pos": {
"0": 1466,
"1": 93
},
"size": {
"0": 210,
"1": 30
@@ -169,9 +106,9 @@
"links": [
16
],
"slot_index": 0,
"shape": 3,
"label": "图像",
"slot_index": 0
"label": "图像"
}
],
"properties": {
@@ -181,13 +118,13 @@
{
"id": 15,
"type": "VHS_VideoCombine",
"pos": [
1410,
120
],
"pos": {
"0": 1409,
"1": 210
},
"size": [
360,
660
928.4444444444445
],
"flags": {},
"order": 4,
@@ -254,32 +191,20 @@
}
},
{
"id": 13,
"type": "VHS_VideoCombine",
"pos": [
1830,
-180
],
"id": 11,
"type": "VHS_LoadVideo",
"pos": {
"0": 826,
"1": -90
},
"size": [
360,
660
210,
466
],
"flags": {},
"order": 2,
"order": 0,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 14,
"label": "图像"
},
{
"name": "audio",
"type": "AUDIO",
"link": null,
"label": "音频"
},
{
"name": "meta_batch",
"type": "VHS_BatchManager",
@@ -294,39 +219,119 @@
],
"outputs": [
{
"name": "Filenames",
"type": "VHS_FILENAMES",
"name": "IMAGE",
"type": "IMAGE",
"links": [
12
],
"slot_index": 0,
"shape": 3,
"label": "图像"
},
{
"name": "frame_count",
"type": "INT",
"links": null,
"shape": 3,
"label": "文件名"
"label": "帧计数"
},
{
"name": "audio",
"type": "AUDIO",
"links": null,
"shape": 3,
"label": "音频"
},
{
"name": "video_info",
"type": "VHS_VIDEOINFO",
"links": null,
"shape": 3,
"label": "视频信息"
}
],
"properties": {
"Node name for S&R": "VHS_VideoCombine"
"Node name for S&R": "VHS_LoadVideo"
},
"widgets_values": {
"frame_rate": 30,
"loop_count": 0,
"filename_prefix": "AnimateDiff",
"format": "video/h265-mp4",
"pix_fmt": "yuv420p10le",
"crf": 22,
"save_metadata": true,
"pingpong": false,
"save_output": true,
"video": "6月28日(1).mp4",
"force_rate": 0,
"force_size": "Disabled",
"custom_width": 512,
"custom_height": 512,
"frame_load_cap": 0,
"skip_first_frames": 0,
"select_every_nth": 1,
"choose video to upload": "image",
"videopreview": {
"hidden": false,
"paused": false,
"params": {
"filename": "AnimateDiff_00004.mp4",
"subfolder": "",
"type": "output",
"format": "video/h265-mp4",
"frame_rate": 30
"frame_load_cap": 0,
"skip_first_frames": 0,
"force_rate": 0,
"filename": "6月28日(1).mp4",
"type": "input",
"format": "video/mp4",
"select_every_nth": 1
},
"muted": false
}
}
},
{
"id": 1,
"type": "BiRefNet_Hugo",
"pos": {
"0": 1076,
"1": -81
},
"size": {
"0": 319.6482238769531,
"1": 154.82546997070312
},
"flags": {},
"order": 1,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 12,
"slot_index": 0,
"label": "图像"
}
],
"outputs": [
{
"name": "image",
"type": "IMAGE",
"links": [
14
],
"slot_index": 0,
"shape": 3,
"label": "图像"
},
{
"name": "mask",
"type": "MASK",
"links": [
15
],
"slot_index": 1,
"shape": 3,
"label": "遮罩"
}
],
"properties": {
"Node name for S&R": "BiRefNet_Hugo"
},
"widgets_values": [
false,
"transparency",
"auto"
]
}
],
"links": [
@@ -367,10 +372,10 @@
"config": {},
"extra": {
"ds": {
"scale": 0.8390545288824037,
"scale": 1.6105100000000008,
"offset": [
-471.7394319558889,
296.200180526101
-674.0647486185928,
271.0286531806558
]
}
},
workflow/workflow.json
+158 -150
View File
@@ -2,159 +2,21 @@
"last_node_id": 5,
"last_link_id": 4,
"nodes": [
{
"id": 1,
"type": "BiRefNet_Hugo",
"pos": [
921,
256
],
"size": {
"0": 210,
"1": 46
},
"flags": {},
"order": 1,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 1,
"label": "image",
"slot_index": 0
}
],
"outputs": [
{
"name": "image",
"type": "IMAGE",
"links": [
2
],
"shape": 3,
"label": "image",
"slot_index": 0
},
{
"name": "mask",
"type": "MASK",
"links": [
3
],
"shape": 3,
"label": "mask",
"slot_index": 1
}
],
"properties": {
"Node name for S&R": "BiRefNet_Hugo"
}
},
{
"id": 4,
"type": "MaskToImage",
"pos": [
922,
352
],
"size": {
"0": 210,
"1": 26
},
"flags": {},
"order": 3,
"mode": 0,
"inputs": [
{
"name": "mask",
"type": "MASK",
"link": 3,
"label": "遮罩"
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
4
],
"shape": 3,
"label": "图像",
"slot_index": 0
}
],
"properties": {
"Node name for S&R": "MaskToImage"
}
},
{
"id": 5,
"type": "PreviewImage",
"pos": [
1299,
410
],
"size": [
210,
246
],
"flags": {},
"order": 4,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 4,
"label": "图像"
}
],
"properties": {
"Node name for S&R": "PreviewImage"
}
},
{
"id": 3,
"type": "PreviewImage",
"pos": [
1295,
74
],
"size": [
210,
246
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 2,
"label": "图像"
}
],
"properties": {
"Node name for S&R": "PreviewImage"
}
},
{
"id": 2,
"type": "LoadImage",
"pos": [
561,
254
],
"size": [
315,
314
],
"pos": {
"0": 561,
"1": 254
},
"size": {
"0": 315,
"1": 314
},
"flags": {},
"order": 0,
"mode": 0,
"inputs": [],
"outputs": [
{
"name": "IMAGE",
@@ -180,6 +42,152 @@
"image.webp",
"image"
]
},
{
"id": 3,
"type": "PreviewImage",
"pos": {
"0": 1312,
"1": 140
},
"size": [
410.6826949448018,
214.56675819839558
],
"flags": {},
"order": 2,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 2,
"label": "图像"
}
],
"outputs": [],
"properties": {
"Node name for S&R": "PreviewImage"
}
},
{
"id": 4,
"type": "MaskToImage",
"pos": {
"0": 933,
"1": 473
},
"size": {
"0": 210,
"1": 26
},
"flags": {},
"order": 3,
"mode": 0,
"inputs": [
{
"name": "mask",
"type": "MASK",
"link": 3,
"label": "遮罩"
}
],
"outputs": [
{
"name": "IMAGE",
"type": "IMAGE",
"links": [
4
],
"slot_index": 0,
"shape": 3,
"label": "图像"
}
],
"properties": {
"Node name for S&R": "MaskToImage"
}
},
{
"id": 1,
"type": "BiRefNet_Hugo",
"pos": {
"0": 921,
"1": 256
},
"size": [
315.4826217026143,
143.56669716323933
],
"flags": {},
"order": 1,
"mode": 0,
"inputs": [
{
"name": "image",
"type": "IMAGE",
"link": 1,
"slot_index": 0,
"label": "图像"
}
],
"outputs": [
{
"name": "image",
"type": "IMAGE",
"links": [
2
],
"slot_index": 0,
"shape": 3,
"label": "图像"
},
{
"name": "mask",
"type": "MASK",
"links": [
3
],
"slot_index": 1,
"shape": 3,
"label": "遮罩"
}
],
"properties": {
"Node name for S&R": "BiRefNet_Hugo"
},
"widgets_values": [
false,
"transparency",
"auto"
]
},
{
"id": 5,
"type": "PreviewImage",
"pos": {
"0": 1316,
"1": 425
},
"size": [
430.8826461166768,
261.3667765089424
],
"flags": {},
"order": 4,
"mode": 0,
"inputs": [
{
"name": "images",
"type": "IMAGE",
"link": 4,
"label": "图像"
}
],
"outputs": [],
"properties": {
"Node name for S&R": "PreviewImage"
}
}
],
"links": [
@@ -220,10 +228,10 @@
"config": {},
"extra": {
"ds": {
"scale": 1.2100000000000002,
"scale": 1,
"offset": [
-299.9616678351899,
94.8424540865177
-172.4824996323017,
67.43331809554974
]
}
},