65 lines
2.2 KiB
Python
65 lines
2.2 KiB
Python
import math, cv2, random, torch, torchvision, json, os
|
|
import numpy as np
|
|
from PIL import Image, ImageOps, ImageSequence
|
|
import nodes, folder_paths # 기본노드, 파일로드
|
|
|
|
|
|
class abyz22_SaveImage:
|
|
def __init__(self):
|
|
self.output_dir = folder_paths.get_output_directory()
|
|
self.type = "output"
|
|
self.prefix_append = ""
|
|
self.compress_level = 4
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(s):
|
|
return {
|
|
"required": {
|
|
"images": ("IMAGE",),
|
|
"folder_path": ("STRING", {"default": ""}),
|
|
},
|
|
"hidden": {"prompt": "PROMPT", "extra_pnginfo": "EXTRA_PNGINFO"},
|
|
}
|
|
|
|
RETURN_TYPES = ()
|
|
FUNCTION = "save_images"
|
|
|
|
OUTPUT_NODE = True
|
|
|
|
CATEGORY = "abyz22"
|
|
|
|
def save_images(self, images, folder_path, filename_prefix="ComfyUI", prompt=None, extra_pnginfo=None):
|
|
filename_prefix += self.prefix_append
|
|
_, filename, counter, subfolder, filename_prefix = folder_paths.get_save_image_path(
|
|
filename_prefix, self.output_dir, images[0].shape[1], images[0].shape[0]
|
|
)
|
|
|
|
if not os.path.exists(folder_path):
|
|
os.mkdir(folder_path)
|
|
full_output_folder = folder_path
|
|
|
|
file_list = os.listdir(folder_path)
|
|
file_list = [f for f in file_list if ".png" in f]
|
|
file_list = [int(f.split(".")[0].split("-")[1]) for f in file_list]
|
|
file_list.sort(reverse=True)
|
|
counter = 1 if len(file_list) == 0 else file_list[0] + 1
|
|
|
|
file_list = os.listdir(folder_path)
|
|
for file in file_list:
|
|
if ".zip" in file:
|
|
counter = 1
|
|
|
|
results = list()
|
|
for image in images:
|
|
i = 255.0 * image.cpu().numpy()
|
|
img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))
|
|
|
|
if folder_path != "no-data":
|
|
folder_num = folder_path.rsplit("/", 1)[1]
|
|
file = f"{folder_num}-{counter}.png"
|
|
img.save(os.path.join(full_output_folder, file), pnginfo=None, compress_level=self.compress_level)
|
|
results.append({"filename": file, "subfolder": subfolder, "type": self.type})
|
|
counter += 1
|
|
|
|
return {"ui": {"images": results}}
|