アルファチャンネルをグレースケールに変換するノードと、アルファチャンネルをRGBにフラット化するノードを追加

__init__.py

@@ -4,6 +4,8 @@ from .nodes.binarize_image import BinarizeImage
 from .nodes.binarize_image_using_otsu import BinarizeImageUsingOtsu
 from .nodes.antialiasing_image import AntialiasingImage
 from .nodes.remove_white_background_noise import RemoveWhiteBackgroundNoise
+from .nodes.alpha_to_grayscale import AlphaToGrayscale
+from .nodes.alpha_flatten import AlphaFlatten
 
 NODE_CLASS_MAPPINGS = {
     "BrightnessTransparency": BrightnessTransparency,
@@ -12,6 +14,8 @@ NODE_CLASS_MAPPINGS = {
     "GrayscaleImage": GrayscaleImage,
     "AntialiasingImage": AntialiasingImage,
     "RemoveWhiteBackgroundNoise": RemoveWhiteBackgroundNoise,
+    "AlphaToGrayscale": AlphaToGrayscale,
+    "AlphaFlatten": AlphaFlatten,
 }
 
 NODE_DISPLAY_NAME_MAPPINGS = {
@@ -21,6 +25,8 @@ NODE_DISPLAY_NAME_MAPPINGS = {
     "GrayscaleImage": "GrayscaleImage",
     "AntialiasingImage": "AntialiasingImage",
     "RemoveWhiteBackgroundNoise": "RemoveWhiteBackgroundNoise",
+    "AlphaToGrayscale": "AlphaToGrayscale",
+    "AlphaFlatten": "AlphaFlatten",
 }
 
 __all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]

nodes/alpha_flatten.py

@@ -0,0 +1,59 @@
+import numpy as np
+import torch
+
+
+class AlphaFlatten:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "images": (
+                    "IMAGE",
+                    {
+                        "tooltip": "Input images to flatten alpha channel into RGB colors."
+                    },
+                ),
+            }
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    FUNCTION = "run"
+    CATEGORY = "ComfyUI-Image-Toolkit"
+
+    def run(self, images):
+        is_tensor = isinstance(images, torch.Tensor)
+
+        if is_tensor:
+            if images.device.type != "cpu":
+                images = images.cpu()
+            images_np = images.numpy()
+        else:
+            images_np = images
+
+        result_images = []
+
+        for image in images_np:
+            # アルファチャンネルがある場合のみ処理
+            if image.shape[2] == 4:
+                # RGB値を取得
+                rgb = image[:, :, :3]
+                # アルファ値を取得
+                alpha = image[:, :, 3]
+                
+                # 白背景とのalpha合成処理
+                # result = foreground_color * alpha + background_color * (1 - alpha)
+                # 白背景：background_color = 1.0
+                flattened_rgb = rgb * alpha[:, :, np.newaxis] + 1.0 * (1.0 - alpha[:, :, np.newaxis])
+                
+                result_images.append(flattened_rgb)
+            else:
+                # アルファチャンネルがない場合はそのまま返す
+                result_images.append(image)
+
+        result_np = np.stack(result_images)
+        result_tensor = torch.from_numpy(result_np)
+
+        if is_tensor and images.device.type != "cpu":
+            result_tensor = result_tensor.to(images.device)
+
+        return (result_tensor,)

nodes/alpha_to_grayscale.py

@@ -0,0 +1,68 @@
+import numpy as np
+import torch
+
+
+class AlphaToGrayscale:
+    @classmethod
+    def INPUT_TYPES(s):
+        return {
+            "required": {
+                "images": (
+                    "IMAGE",
+                    {
+                        "tooltip": "Input images to convert alpha to grayscale."
+                    },
+                ),
+            }
+        }
+
+    RETURN_TYPES = ("IMAGE",)
+    FUNCTION = "run"
+    CATEGORY = "ComfyUI-Image-Toolkit"
+
+    def run(self, images):
+        is_tensor = isinstance(images, torch.Tensor)
+
+        if is_tensor:
+            if images.device.type != "cpu":
+                images = images.cpu()
+            images_np = images.numpy()
+        else:
+            images_np = images
+
+        result_images = []
+
+        for image in images_np:
+            # アルファチャンネルがある場合のみ処理
+            if image.shape[2] == 4:
+                # アルファ値を取得
+                alpha = image[:, :, 3]
+                
+                # アルファ値をグレースケール値に変換（BrightnessTransparencyの逆変換）
+                # BrightnessTransparency: alpha = 1.0 - grayscale
+                # AlphaToGrayscale: grayscale = 1.0 - alpha
+                grayscale = 1.0 - alpha
+                
+                # RGB画像を作成（透明度なし）
+                rgb_image = np.zeros(
+                    (image.shape[0], image.shape[1], 3), dtype=np.float32
+                )
+                
+                # 全チャンネルにグレースケール値を設定
+                rgb_image[:, :, 0] = grayscale
+                rgb_image[:, :, 1] = grayscale
+                rgb_image[:, :, 2] = grayscale
+                
+                result_images.append(rgb_image)
+            else:
+                # アルファチャンネルがない場合はそのまま返す
+                result_images.append(image)
+
+        result_np = np.stack(result_images)
+        result_tensor = torch.from_numpy(result_np)
+
+        if is_tensor and images.device.type != "cpu":
+            result_tensor = result_tensor.to(images.device)
+
+        return (result_tensor,)
+