79 lines
2.5 KiB
Python
79 lines
2.5 KiB
Python
import torch
|
|
from PIL import Image, ImageOps
|
|
import numpy as np
|
|
|
|
class REMADE_Batch_Image_Blend:
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
return {
|
|
"required": {
|
|
"images_a": ("IMAGE",),
|
|
"images_b": ("IMAGE",),
|
|
"blend_percentage": ("FLOAT", {"default": 0.8, "min": 0.0, "max": 1.0, "step": 0.01}),
|
|
},
|
|
}
|
|
|
|
RETURN_TYPES = ("IMAGE",)
|
|
RETURN_NAMES = ("images",)
|
|
FUNCTION = "batch_image_blend"
|
|
|
|
CATEGORY = "REMADE/Image"
|
|
|
|
def batch_image_blend(self, images_a, images_b, blend_percentage):
|
|
print(f"Input shapes: images_a {images_a.shape}, images_b {images_b.shape}")
|
|
print(f"Blend percentage: {blend_percentage}")
|
|
|
|
assert images_a.shape[0] == images_b.shape[0], "Number of images in set A must match number of images in set B"
|
|
|
|
blended_images = []
|
|
|
|
for i in range(images_a.shape[0]):
|
|
img_a = images_a[i]
|
|
img_b = images_b[i]
|
|
print(f"Processing image pair {i+1}")
|
|
print(f" Image A shape: {img_a.shape}")
|
|
print(f" Image B shape: {img_b.shape}")
|
|
|
|
# Convert images to PIL
|
|
pil_a = tensor2pil(img_a)
|
|
pil_b = tensor2pil(img_b)
|
|
|
|
print(f" PIL Image sizes: A {pil_a.size}, B {pil_b.size}")
|
|
|
|
# Blend image
|
|
blend_mask = Image.new(mode="L", size=pil_a.size,
|
|
color=(round(blend_percentage * 255)))
|
|
blend_mask = ImageOps.invert(blend_mask)
|
|
img_result = Image.composite(pil_a, pil_b, blend_mask)
|
|
|
|
print(f" Blended image size: {img_result.size}")
|
|
|
|
# Convert back to tensor and append
|
|
blended_tensor = pil2tensor(img_result)
|
|
print(f" Blended tensor shape: {blended_tensor.shape}")
|
|
|
|
blended_images.append(blended_tensor)
|
|
|
|
del pil_a, pil_b, blend_mask, img_result
|
|
|
|
# Stack all blended images into a single tensor
|
|
result = torch.stack(blended_images, dim=0)
|
|
print(f"Final result shape: {result.shape}")
|
|
|
|
return (result,)
|
|
|
|
def tensor2pil(image):
|
|
return Image.fromarray((image.cpu().numpy() * 255).astype(np.uint8))
|
|
|
|
def pil2tensor(image):
|
|
return torch.from_numpy(np.array(image).astype(np.float32) / 255.0)
|
|
|
|
# Node mappings
|
|
NODE_CLASS_MAPPINGS = {
|
|
"REMADE Batch Image Blend": REMADE_Batch_Image_Blend
|
|
}
|
|
|
|
NODE_DISPLAY_NAME_MAPPINGS = {
|
|
"REMADE Batch Image Blend": "REMADE Batch Image Blend"
|
|
}
|