如何使用Python将多张图片对角合并为一张图片

Question

I'm trying to merge multiple images diagonally into a single one using Python. I checked a lot of questions but didn't find something similar to my need.

All I can do right now is a simple merge of files on top of each other:

from PIL import Image

import numpy as np

img = Image.open("1.png")

background = Image.open("2.png")

background.paste(img, (0, 0), img)

background.save('result.png',"PNG")

Here are the pictures to test :

image1 , image2 , image3

I need the pictures to be arranged diagonally to fit into a final 900 x 1200 px size picture with white Background. Probably they need to be sized down a bit and fit ? At least that's the process I am doing in Photoshop, manually (time consuming).

Sometimes there's 2 pictures to fit, sometimes could be 4 or 5.

Answer 1

This should do the job:

from PIL import Image

images = ['1.png', '2.png', '3.png']

# shift between images
offset = (200, 100)
target_size = (900, 1200)

images = [Image.open(fn) for fn in images]
no_img = len(images)
image_size = [s+no_img*o for s, o in zip(images[0].size, offset)]

#create empty background
combined_image = Image.new('RGBA', image_size)

# paste each image at a slightly shifted position, start at top right
for idx, image in enumerate(images):
  combined_image.paste(image, ((no_img - idx - 1) * offset[0], idx * offset[1]), image)

# crop to non-empty area
combined_image = combined_image.crop(combined_image.getbbox())

# resizing and padding such that it fits 900 x 1200 px
scale = min(target_size[0] / combined_image.size[0], target_size[1] / combined_image.size[1])
combined_image = combined_image.resize((int(combined_image.size[0] * scale), int(combined_image.size[1] * scale)), Image.BICUBIC)
img_w, img_h = combined_image.size

finale_output = Image.new('RGB', target_size, (255, 255, 255))

offset = ((target_size[0] - img_w) // 2, (target_size[1] - img_h) // 2)
finale_output.paste(combined_image, offset, combined_image)

# display
finale_output.show()

EDIT: I added the code for resizing and padding such that the output is exactly of your wanted size (whilst maintaining the aspect ratio).