将100多个图像组合在一起的更优化方法

Question

Python3.4 皮尔

我结合了100+16x16瓷砖一起随机生成一个测试应用程序的地形，并保存为存储.png文件

我在测试时担心的是我会收到以下完成时间：

• 100x100地图（1600x1600像素）=1.16分钟
• 200x200地图（3200x3200像素）=2.43分钟
• 300x300地图（4800x4800像素）=5.27分钟
• 500x500地图（8000x8000像素）=13.14分钟

对于移动/桌面应用程序，这太长。有没有更理想的方法来创建这些地图

注：我不打算使用任何其他以外的16x16瓷砖。不过，我对其他Python模块持开放态度

from datetime import datetime
from os import listdir, chdir
from random import randint
from time import strftime

from PIL import Image

x_size = 4800
y_size = 4800

x_tile_size = 16
y_tile_size = 16

x_tile_amount = int(x_size/x_tile_size)
y_tile_amount = int(y_size/y_tile_size)

new_im = Image.new('RGB', (x_size, y_size))

file_path = './assets/terrain/plains/'
print("Generating map.")
start_time = strftime("%H:%M:%S")
for i in range(x_tile_amount):
    for j in range(y_tile_amount):

    dice_roll = randint(1,7)

    if dice_roll == 1:
        terrain_file_path = 'grass_1.png'                        
    elif dice_roll == 2:
        terrain_file_path = 'grass_2.png'
    elif dice_roll == 3:
        terrain_file_path = 'grass_3.png'                            
    elif dice_roll == 4:
        terrain_file_path = 'grass_4.png'
    elif dice_roll == 5:
        terrain_file_path = 'white_flower.png'
    elif dice_roll == 6:
        terrain_file_path = 'red_flower.png'
    elif dice_roll == 7:
        terrain_file_path = 'blue_flower.png'

    old_im = Image.open(file_path + terrain_file_path)
    new_im.paste(old_im, (i*x_tile_size,j*y_tile_size))


end_time = strftime("%H:%M:%S")
t_format = "%H:%M:%S"
t_delta = datetime.strptime(end_time, t_format) - datetime.strptime(start_time, t_format)
print('Time to generate map = ' + str(t_delta))
new_im.save('pillow_test.png', 'PNG')

Answer 1

您正在打开新图像n^2次。这使得它非常慢

在执行for循环之前，应该先预加载这些图像

grass_1 = Image.open(file_path + 'grass_1.png')
grass_2 = Image.open(file_path + 'grass_2.png')
...
for i in range(x_tile_amount):
    for j in range(y_tile_amount):
         dice_roll = random.randint(1, 7)
         if dice_roll == 1:
              etc.

你可以做的另一件事是把你的照片设置成dict

my_dict = {
    1: 'grass_1',
    2: 'grass_2',
    ...
}

那你就这么做吧

dice_roll = random.randint(1,7)
new_im.paste(my_dict[dice_roll], (i*x_tile_size,j*y_tile_size))

试试看这是否对你的时间有帮助

主要的问题是，当你生成一张100x100的地图时，你要加载10000次图像。没有理由加载相同的7个图像文件10k次，所以您只需预加载它们并根据需要添加

Answer 2

您正在为每个随机选择加载每个地形图像。最好先预加载所有图像，然后简单地使用random.choice()为您选择一个图像，如下所示：

from datetime import datetime
import random
from time import strftime

from PIL import Image

x_size = 4800
y_size = 4800

x_tile_size = 16
y_tile_size = 16

x_tile_amount = int(x_size/x_tile_size)
y_tile_amount = int(y_size/y_tile_size)

new_im = Image.new('RGB', (x_size, y_size))
file_path = './assets/terrain/plains/'
images = ['grass_1.png', 'grass_2.png', 'grass_3.png', 'grass_4.png', 'white_flower.png', 'red_flower.png', 'blue_flower.png']
terrain = [Image.open(os.path.join(file_path, image)) for image in images]

print("Generating map.")
start_time = strftime("%H:%M:%S")

for i in range(x_tile_amount):
    for j in range(y_tile_amount):
        old_im = random.choice(terrain)
        new_im.paste(old_im, (i*x_tile_size,j*y_tile_size))

end_time = strftime("%H:%M:%S")
t_format = "%H:%M:%S"
t_delta = datetime.strptime(end_time, t_format) - datetime.strptime(start_time, t_format)
print('Time to generate map = ' + str(t_delta))
new_im.save('pillow_test.png', 'PNG')    

这种方法应该给你一个巨大的速度