import numpy as np import pyheif import cv2 import whatimage import time from PIL import Image import requests import torch def image2format(image: bytes): fmt = whatimage.identify_image(image) if fmt in ["heic", "avif"]: i = pyheif.read_heif(image) image = Image.frombytes(mode=i.mode, size=i.size, data=i.data) image = cv2.cvtColor(np.asarray(image), cv2.COLOR_RGB2BGR) else: image = np.asarray(bytearray(image), dtype=np.uint8) image = cv2.imdecode(image, cv2.IMREAD_COLOR) return image class image_url_node: def __init__(self): pass @classmethod def INPUT_TYPES(s): return { "required": { "image_url": ("STRING", {"default": "null"}), }, } RETURN_TYPES = ("IMAGE",) FUNCTION = "download_image" CATEGORY = "image" def download_image(self, image_url): res = requests.get(image_url, timeout=3).content image = image2format(res) image = image.astype(np.float32) / 255. image = torch.from_numpy(image) image = image.unsqueeze(0) return (image,) NODE_CLASS_MAPPINGS = { "image_url_node": image_url_node } NODE_DISPLAY_NAME_MAPPINGS = { "image_url_node": "image_url_node" }