ComfyUI Extension: ComfyUI-WJNodes

ComfyUI-WJNodes explain

• Ready to use upon download. No need to install dependencies for the time being.
• If there are new functions or suggestions, please provide feedback.
• Attention! The delfile node is not recommended for use on servers. I am not responsible for any losses incurred.
• To disable the delfile node, change the 'DelFile=True' of node/file.py to 'DelFile=False'

Node list:

• Image: WJNode/Image
  • ✅load image from path : Load image from path
  • ✅save image to path : Save image by overwriting the path
  • ✅save image out : Save image to output and output the path
  • ✅select images batch : Batch selection and recombination of batches
  • ✅load image adv : Load image with mask inversion and path output
  • 🟩Load value feature recognition model (e.g., nsfw, aesthetic score, AI value, time)
  • 🟩Input recognition model and image batch, output batch and corresponding feature values
  • 🟩Sort image batches through specified arrays (e.g., feature value arrays)
• Mask Editing: WJNode/MaskEdit
  • ✅load_color_config : Load color configuration for color block to mask, currently supports loading ADE20K pre-processing color data
  • ✅color_segmentation : Color block to mask, currently supports pre-processing ADE20K and SAM2 data
  • ✅color_segmentation_v2 : Color block to mask v2, uses keys in color configuration to select masks, only supports ADE20K data
  • ✅mask_select_mask : Mask selection within a mask batch (intersection represents selection)
  • 🟩coords_select_mask : Coordinate selection of masks, used to assist SAM2 video keying (under development)
  • ✅mask_line_mapping : Mask line mapping, can automatically calculate maximum and minimum values when input is -1 or 256, can map to specified values
  • ✅mask_and_mask_math : Mask to mask operations, supports addition/subtraction/intersection/multiplication operations, Adjustable cv2 and torch modes, if cv2 is not installed, automatically switches to torch
  • 🟩Accurate_mask_clipping : Precise search for mask bbox boundaries (under development)
• Image Editing: WJNode/ImageEdit
  • ✅adv crop : Advanced cropping: can quickly crop/expand/move/flip images, can output background masks and custom filling (Usage method included in the node, known bug: expansion size more than 1 times cannot use tiling and mirror filling)
  • ✅mask detection : Mask detection: detect whether there is a mask, detect whether it is all hard edges, detect whether the mask is pure white/pure black/pure gray and output values 0-255
  • ✅InvertChannelAdv : Invert/separate image channels Image ⭐ RGBA to mask batch Replace channels Any channel to RGBA
  • ✅Video_fade : Two video segments can choose two ways to fade in and out, Mask: Local fade in and out under development... Exponential: Exponential gradient under development...
• Others: WJNode/Other-functions
  • ✅any_data : Group any data, known bug: nested grouping will split
  • ✅show_type : Display data type
  • ✅array_element_count : Display the number of array elements
• Plugins: WJNode/Other-plugins
  • ✅WAS_Mask_Fill_Region_batch : Optimize WAS plugin's WAS_Mask_Fill_Region (mask cleanup) to support batches
  • ✅SegmDetectorCombined_batch : Optimize impack-pack plugin's SegmDetectorCombined (segm detection mask) to support batches
  • ✅bbox_restore_mask : Add impack-pack plugin's seg decomposition, restore cropped images through cropping data (SEG editing)
  • ✅Sam2AutoSegmentation_data : Add Sam2AutoSegmentation (kijia) node's color list/coordinate output, used to assist SAM2 video keying
• Path: WJNode/Path
  • ✅comfyui path : Output comfyui common paths (root, output/input, plugins, models, cache, Python environment)
  • ✅path append : Add prefix/suffix to strings (reference KJNode)
  • ✅del file : Detect whether file or path exists, whether to delete file, operation requires input signal, deletion requires write permission
  • ✅split path : Path slicing, input path, output: disk symbol/path/file/extension + detect whether it is a file

ComfyUI-WJNodes介绍

• 下载即用，暂时无需安装依赖(可选安装cv2)，有新功能或建议请反馈。
• 注意！delfile节点不建议在服务器上使用，产生任何损失与本人无关
• 修改node/file.py的“DelFile = True”为“DelFile = False”即可禁用delfile节点

节点列表

• 图像：WJNode/Image
  • ✅load image from path : 从路径加载图片
  • ✅save image to path : 通过路径覆盖保存图片
  • ✅save image out : 保存图片到output并输出该路径
  • ✅select images batch : 批次选择和重新组合批次
  • ✅load image adv : 带遮罩反转和路径输出的加载图片
  • 🟩加载值特征识别模型(例如nsfw,美学分数,AI值,time)
  • 🟩输入识别模型和图像批次，输出批次和对应特征值
  • 🟩通过指定数组(例如特征值数组)排序图片批次
• 遮罩编辑：WJNode/MaskEdit
  • ✅load_color_config : 加载颜色配置，用于色块转遮罩,目前支持加载 ADE20K 预处理颜色数据
  • ✅color_segmentation : 色块转遮罩，目前支持预处理 ADE20K 和 SAM2 数据
  • ✅color_segmentation_v2 : 色块转遮罩v2，使用颜色配置内的key来选取遮罩，仅支持ADE20K数据
  • ✅mask_select_mask : 遮罩选择遮罩批次内的遮罩(有交集即代表选择)
  • 🟩coords_select_mask : 坐标选择遮罩，用于辅助SAM2视频抠图(待开发)
  • ✅mask_line_mapping : 遮罩线映射，当输入为-1或256时可自动计算最值，可映射到指定值
  • ✅mask_and_mask_math : 遮罩与遮罩的运算，支持加/减/(交集)/乘运算， 可调cv2和torch两种模式,若未安装cv2则自动切换到torch
  • 🟩Accurate_mask_clipping : 精确查找遮罩bbox边界 (待开发)
• 图像编辑：WJNode/ImageEdit
  • ✅adv crop : 高级裁剪:可快速裁剪/扩展/移动/翻转图片,可输出背景遮罩和自定义填充
  • (节点内附使用方法,已知bug:扩展尺寸超过1倍时无法使用平铺和镜像填充)
  • ✅mask detection : 遮罩检测:检测是否有遮罩,检测是否是全硬边,检测遮罩是否是纯白/纯黑/纯灰并输出值0-255
  • ✅InvertChannelAdv : 翻转/分离图像通道⭐ 图像RGBA转遮罩批次 替换通道 任意通道合成RGBA
  • ✅Video_fade : 两段视频可选两种方式渐入渐出， 遮罩:局部渐入渐出开发中... 指数:指数渐变开发中...
• 其它：WJNode/Other-functions
  • ✅any_data : 将任意数据打组，已知bug:嵌套打组会裂开
  • ✅show_type : 显示数据类型
  • ✅array_element_count : 显示数组元素数量
• 插件：WJNode/Other-plugins
  • ✅WAS_Mask_Fill_Region_batch : 优化WAS插件的的WAS_Mask_Fill_Region(遮罩清理)支持批次
  • ✅SegmDetectorCombined_batch : 优化impack-pack插件的的SegmDetectorCombined(segm检测遮罩)支持批次
  • ✅bbox_restore_mask : 增加impack-pack插件的seg分解后，通过裁剪数据恢复裁剪后的图像（SEG编辑）
  • ✅Sam2AutoSegmentation_data : 增加Sam2AutoSegmentation(kijia)节点的颜色列表/坐标输出，用于辅助SAM2视频抠图
• 路径：WJNode/Path
  • ✅comfyui path : 输出comfyui常用路径(根,输出/输入,插件,模型,缓存,python环境)
  • ✅path append : 给字符串增加前缀后缀(参考KJNode)
  • ✅del file : 检测文件或路径是否存在,是否删除文件,运行需输入信号,删除需有写入权限
  • ✅split path : 路径切片,输入路径,输出:盘符/路径/文件/扩展名+检测是否是文件