有没有一种快速的方法可以将图像文件保存在python上？

Question

I made a program to filter an images with only 20 colors, and i want to save the process that the program is making until it finished, i've successfully saved the images however it takes a lot of time.

Lets say the user gives me an image of 800x600, it will take about 15 seconds or less to my program to finish editing the image, but if i save after every step, it will take around 10 min.. and that is because after every save, the program is waiting for the file to be ready altho it doesn't need the image it can move on and let winodws handle it.

I've tried doing it with threading and i get the same result, about 1s per image..

from PIL import Image
from PIL import ImageGrab

im = Image.open('braw.png') # Can be many different formats.
rgb_im = im.convert('RGB')

pix = rgb_im.load()

height, width = im.size
image = ImageGrab.grab()
path = os.getcwd()

def read_color(height,width, COLORS):
    for x in range(height):
        for y in range(width):
            r,g,b = rgb_im.getpixel((x,y))
            color = closest_color(r, g, b, COLORS)
            back_work(color,x,y)
            save_frame(x, '/frames')

def save_frame(x, location):
    try:
        rgb_im.save(path + location+'/ark_ai' + str(x) + '.jpeg')
    except OSError:
        sleep(0.5)

def back_work(color, x, y):
    pix[x,y] = color

I've expected the program to keep saving images after every loop even tho its not done, is there a faster way of doing so that i am not aware of?

Answer 1

It looks like you're saving every time you update a pixel, which means you're saving 480,000 images, with a total of 691 GB written to the disk. Granted, not all of that will be saved to the hard drive at once, since you're overwriting 99% of the files. Nonetheless, that's a lot of file I/O, and there simply isn't much you can do to speed that up.

Why don't you try saving after each column is updated, rather than each pixel? That would reduce the amount of saving you do by a factor of 600. And the output should be the same, since you're only skipping saving the frames that would have been overwritten anyway.

def read_color(height,width, COLORS):
    for x in range(height):
        for y in range(width):
            r,g,b = rgb_im.getpixel((x,y))
            color = closest_color(r, g, b, COLORS)
            back_work(color,x,y)
        save_frame(x, '/frames')