增加文档说明

README.md

@@ -2,10 +2,13 @@
 ComfyUI-Easyai is a powerful extension for ComfyUI that enables users to share workflows and models to easyai.
 
 ## ChangeLog
+- 20251102: add LoadImagesMulti，使用EasyAI平台多图上传（单任务）节点与该组件对接，可实现批量任务或者任意数量的多图上传和快速选择
+  ![LoadImage](./docs/loadImageMulti.png)
 - 202401027: init
 
 ## Features
 - share workflows to easyai
+- LoadImagesMulti
 
 ## Installation
 ### [method1] From Source
@@ -21,8 +24,6 @@ If you have a comfy-cli, you can simply execute `comfy node registry-install com
 ## Usage
 ![use](./docs/use.gif)
 - After installation, you can use the Easyai nodes in your ComfyUI workflows.
-- For more detailed usage, please refer to the [金华岩石](https://jinhuayanshi.cn) or [三景AI](https://easyai.jinhuayanshi.cn) websites.
-- Follow me: <img src="./docs/douyin.jpg" width="200" />
 
 ## License
 This project is licensed under the MIT License. See the LICENSE file for details.

__init__.py

@@ -1,6 +1,6 @@
 WEB_DIRECTORY = "js"
-from .nodes import NODE_CLASS_MAPPINGS
-__all__ = ['NODE_CLASS_MAPPINGS']
+from .nodes import NODE_CLASS_MAPPINGS,NODE_DISPLAY_NAME_MAPPINGS
+__all__ = ['NODE_CLASS_MAPPINGS', 'NODE_DISPLAY_NAME_MAPPINGS']
 
 from aiohttp import ClientSession, web
 from server import PromptServer

nodes.py

@@ -1,51 +1,126 @@
-import soundfile as sf
-import requests
-import io
+import os
+from PIL import Image, ImageOps, ImageSequence
 import numpy as np
 import torch
+import folder_paths
+import node_helpers
+import re
 
-class AudioLoadPath:
+class LoadImagesMulti:
     @classmethod
-    def INPUT_TYPES(s):
-        return {"required": { "path": ("STRING", {"default": "X://insert/path/here.mp4"}),
-                              "sample_rate": ("INT", {"default": 22050, "min": 6000, "max": 192000, "step": 1}),
-                              "offset": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1e6, "step": 0.001}),
-                              "duration": ("FLOAT", {"default": 0.0, "min": 0.0, "max": 1e6, "step": 0.001})}}
+    def INPUT_TYPES(cls):
+        input_dir = folder_paths.get_input_directory()
+        files = [f for f in os.listdir(input_dir) if os.path.isfile(os.path.join(input_dir, f))]
+        files = folder_paths.filter_files_content_types(files, ["image"])
 
-    RETURN_TYPES = ("AUDIO", )
-    CATEGORY = "Audio Reactor"
-    FUNCTION = "load"
+        return {
+            "required": {
+                "filenames": ("STRING", {
+                    "default": "filename1.png\nfilename2.png",
+                    "tooltip": "输入多个文件名，用逗号或者换行分隔，例如： 1.png, 2.jpg, dir/sub.png",
+                    "multiline": True  # 多行文本域
+                }),
+            }
+        }
 
-    def load(self, path: str, sample_rate: int, offset: float, duration: float|None):
-        if duration == 0.0:
-            duration = None
+    CATEGORY = "EasyAI"
+    RETURN_TYPES = ("IMAGE", "MASK", "STRING",
+                    "IMAGE", "IMAGE", "IMAGE", "IMAGE", "IMAGE", "IMAGE")
+    RETURN_NAMES = ("images", "masks", "filepaths",
+                    "image1", "image2", "image3", "image4", "image5", "image6")
+    INPUT_IS_LIST = False
+    OUTPUT_IS_LIST = (True,True,False,
+                      False,False,False,False,False,False)
+    FUNCTION = "load_images"
 
-        if path.startswith(('http://', 'https://')):
-            # 对于网络路径，直接从内存加载
-            try:
-                response = requests.get(path)
-                response.raise_for_status()
-                audio_data = io.BytesIO(response.content)
+    def load_images(self, filenames):
+        # 解析用户输入的多个文件名
 
-                # 使用 soundfile 从内存中读取音频数据
-                audio, file_sr = sf.read(audio_data)
+        filenames = re.split(r'[, \r\n]+', filenames)  # 按逗号、空格或任何换行符分割
+        filenames = [f.strip() for f in filenames if f.strip()]  # 去掉首尾空格和空字符串
 
-                # 如果需要重采样
-                if file_sr != sample_rate:
-                    # 这里需要添加重采样逻辑
-                    # 可以使用 librosa.resample 或其他方法
-                    pass
+        if len(filenames) == 0:
+            raise ValueError("未提供有效的文件名，请至少输入一个文件名。")
 
-            except Exception as e:
-                raise Exception(f"加载网络音频失败: {str(e)}")
-        else:
-            # 本地文件使用原有的 librosa 方式加载
-            audio, _ = librosa.load(path, sr=sample_rate, offset=offset, duration=duration)
+        output_images = []
+        output_masks = []
+        output_paths = []
 
-        # 转换为 torch tensor 并调整维度
-        audio = torch.from_numpy(audio)[None,:,None]
-        return (audio,)
+        excluded_formats = ["MPO"]
 
+        for fname in filenames:
+            # 支持子目录，如 "sub/my.png"
+            img_path = folder_paths.get_annotated_filepath(fname)
+
+            if not folder_paths.exists_annotated_filepath(fname):
+                raise FileNotFoundError(f"文件不存在: {fname}")
+
+            img = node_helpers.pillow(Image.open, img_path)
+
+            # frames_img = []
+            # frames_mask = []
+
+            w, h = None, None
+
+            for i in ImageSequence.Iterator(img):
+                i = node_helpers.pillow(ImageOps.exif_transpose, i)
+
+                if i.mode == "I":
+                    i = i.point(lambda x: x * (1 / 255))
+
+                rgb = i.convert("RGB")
+
+                # 统一尺寸
+                if w is None:
+                    w, h = rgb.size
+                elif rgb.size != (w, h):
+                    continue
+
+                # 转 tensor
+                rgb_tensor = torch.from_numpy(
+                    np.array(rgb).astype(np.float32) / 255.0
+                )[None,]
+
+                # Mask
+                if "A" in i.getbands():
+                    alpha = i.getchannel("A")
+                    mask_np = np.array(alpha).astype(np.float32) / 255.0
+                    mask_tensor = 1. - torch.from_numpy(mask_np)
+                elif i.mode == "P" and "transparency" in i.info:
+                    alpha = i.convert("RGBA").getchannel("A")
+                    mask_np = np.array(alpha).astype(np.float32) / 255.0
+                    mask_tensor = 1. - torch.from_numpy(mask_np)
+                else:
+                    mask_tensor = torch.zeros((64, 64), dtype=torch.float32)
+
+                output_images.append(rgb_tensor)
+                output_masks.append(mask_tensor.unsqueeze(0))
+
+            # if len(frames_img) > 1 and img.format not in excluded_formats:
+            #     image_tensor = torch.cat(frames_img, dim=0)
+            #     mask_tensor = torch.cat(frames_mask, dim=0)
+            # else:
+            #     image_tensor = frames_img[0]
+            #     mask_tensor = frames_mask[0]
+
+            # output_images.append(frames_img)
+            # output_masks.append(frames_mask)
+            output_paths.append(img_path)
+
+        # 合并为 batch（N, H, W, C）
+        # batch_images = output_images  # 保持 list，每个元素是不同尺寸的 tensor
+        # batch_masks = output_masks
+        # 前6张单图输出（如果不够就用None占位）
+        single_images = [output_images[i] if i < len(output_images) else None for i in range(6)]
+
+        return (output_images, output_masks, "\n".join(output_paths),
+                *single_images)
+
+# 节点导出
 NODE_CLASS_MAPPINGS = {
-    "AudioLoadPath": AudioLoadPath,
+    "LoadImagesMulti": LoadImagesMulti
+}
+
+NODE_DISPLAY_NAME_MAPPINGS = {
+    "LoadImagesMulti": "Load Images(input filenames)"
 }

requirements.txt

@@ -0,0 +1,2 @@
+numpy
+librosa