📂 IG Load Image

Documentation

• Class name: IG Load Image
• Category: 🐓 IG Nodes/IO
• Output node: False

The IG Load Image node is designed for loading images from a specified directory, providing functionality to handle image preprocessing such as orientation correction, conversion to a specific color mode, and normalization. It abstracts the complexities of image loading and preprocessing, making it easier to integrate image data into workflows.

Input types

Required

• image_path
  • The 'image_path' parameter specifies the path to the image to be loaded. It plays a crucial role in determining which image file will be processed, affecting the node's execution and the resulting image data.
  • Comfy dtype: STRING
  • Python dtype: str

Output types

• image
  • Comfy dtype: IMAGE
  • The 'image' parameter represents the preprocessed image data, ready for further processing or analysis.
  • Python dtype: torch.Tensor
• mask
  • Comfy dtype: MASK
  • The 'mask' parameter provides a mask for the loaded image, useful for operations that require segmentation or specific area processing.
  • Python dtype: torch.Tensor

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class IG_LoadImage:
    @classmethod
    def INPUT_TYPES(s):
        return {
            "required": {
                "image_path": ("STRING", {"forceInput": True}),
            },
        }

    RETURN_TYPES = ("IMAGE", "MASK")
    FUNCTION = "main"

    CATEGORY = TREE_IO

    def main(self, image_path: str, **kwargs):
        img = Image.open(image_path)
        output_images = []
        output_masks = []
        for i in ImageSequence.Iterator(img):
            i = ImageOps.exif_transpose(i)
            if i.mode == 'I':
                i = i.point(lambda i: i * (1 / 255))
            image = i.convert("RGB")
            image = np.array(image).astype(np.float32) / 255.0
            image = torch.from_numpy(image)[None,]
            if 'A' in i.getbands():
                mask = np.array(i.getchannel('A')).astype(np.float32) / 255.0
                mask = 1. - torch.from_numpy(mask)
            else:
                mask = torch.zeros((64,64), dtype=torch.float32, device="cpu")
            output_images.append(image)
            output_masks.append(mask.unsqueeze(0))

        if len(output_images) > 1:
            output_image = torch.cat(output_images, dim=0)
            output_mask = torch.cat(output_masks, dim=0)
        else:
            output_image = output_images[0]
            output_mask = output_masks[0]

        return (output_image, output_mask)

    @classmethod
    def IS_CHANGED(s, image_path: str, **kwargs):
        return is_changed_image(image_path, **kwargs)