Social media generational loss simulation

I just simulated a picture being uploaded to social media, then saved, then re-uploaded 2,000 times using Imagemagick, using the following command, iterated over and over again:

magick 0000.jpg -filter triangle -resize nxn -unsharp 0.5x0.5+0.1+0 0001.jpg

Below is the raw file if you want to play raw, or I am kind of curious to see what you think of such a crazy experiment. I know that the resolutions at social media are lower than the random range around 5 megapixels I used, but I wanted to also see how far the degradation would propagate as well from the resizing and jpeg encoding.

DSC03169.ARW (23.9 MB) DSC03169.ARW.pp3 (13.7 KB)

Crazy processing encouraged.
Edit: Photo released under creative commons liscense.

Edit 2: also adding the basic python script I used to generate a batch file featuring all the Imagemagick commands. (Probably a better way to do this using some API stuff, but I feel somewhat intellectually lazy when it comes to programming stuff.)

import random

f = open("resize_gen_3.cmd", "a")
image_1='"E:\\Photoshop Requests\\Jpeg testing\\boat.tif"'
q=' -quality 80 '
print('magick ' + image_1 + ' -quality 100 ' + '0000.jpg', file=f)
fr=" -filter triangle -resize "
unsh=" -unsharp 0.6x0.6+0.4+0 "

for i in range(2000):
    imj_n= zeros[:-digits_n] + str(n) + '.jpg'
    imj_i= zeros[:-digits_i] + str(i) + '.jpg'
    sca_n= random.randrange(1920,3840,16)
    sca_c= str(sca_n) + 'x' + str(sca_n)
    print('magick '+ imj_i + fr + sca_c + unsh + q + imj_n, file=f)

Result with a bit more sharpening.

DSC03169.ARW.xmp (12.6 KB) DSC03169_01.ARW.xmp (16.7 KB)

@afre @Reptorian @Joan_Rake1 sounds like your works are needed!

Result with more jpeg compression on each save.

Hardened survivors.

1 Like

Ten iterations of guided filter with 20x1 kernel plus unsharp mask.