[desuarchive]... jpg
(222.98 KB, 1080x720)

[4archive]... jpg
(219.14 KB, 1080x720)

[desuarchive]... jpg
(109.97 KB, 1080x720)

[4archive]... jpg
(109.97 KB, 1080x720)

 >>/24019/
My own observations on images from archives:
• Images not present or deleted in Desuarchive could have ended on Imgur through 4archive, what a save, right? No, nod really :DDDDD

• Looking at random 4archive thread (see below) today I can't but notice how there are surprisingly many images missing, and I unfortunately can't confirm whether they were deleted as part of current purge or earlier since neither of them apparently contain nudity, and they are not deleted in chronological order, some are still left.

• 4archive admin claims to have scraped 80% of images into his drives by May 18th[1].

• Imgur does not seem to transcode/compress small jpeg images under 1MB or it did not do so to old images. You can prove this by comparing them with imagemagicks's identify -quiet -format "%#" [file] command which I wrapped into crappy script for quick tests:


#!/bin/bash
HASH1=$( identify -quiet -format "%#" "$1" )
HASH2=$( identify -quiet -format "%#" "$2" )
	echo "$HASH1"
	echo "$HASH2"
if [[ "$HASH1" == "$HASH2" ]]
then
	echo "✅️ [Pixel data identical]"
else
	echo "❌️ [Pixel data NOT identical]"
fi


It extracts some sort of "hash" string for raw image disregarding whatever metadata junk was attached in the first place[2][3].

• I chose the following thread to illustrate above points: it both contains large amount of missing images, and has one image that although being 4 KB less in it's Imgur upload, is bitmap-identical to desuarchive version:
	https://desuarchive.org/r9k/thread/29000648/#29013159
	https://4archive.org/board/r9k/thread/29000648#p29013159
	File 570135280.jpg:
	https://desu-usergeneratedcontent.xyz/r9k/image/1459/19/1459195230779.jpg
	https://desuarchive.org/r9k/search/image/87GXo4Mao4VfkDkGtyabMg/
	https://i.imgur.com/viO5GPR.jpg

• Apparently 4chan is/was poor at disregarding junk, or technology has evolved, that using mat2[4] tool I was able to almost halve byte size for both source images. And although now identify produces different strings for each "cleaned" images and their respective originals, both resulting "cleaned" images are identical to each other, also at file level. I'm not sure why this happens and whether results are still true to originals as bit-maps. Otherwise seems like someone was using rare Elizas to smuggle steganographic messages or some shit, whopping 112 Kilobytes, even initial 4KB is as much as Endchan allows in each post.

[1] - https://archived.moe/talk/thread/1694/#q1737
[2] - https://softwarerecs.stackexchange.com/questions/9774/command-line-tool-to-check-whether-two-images-are-exactly-the-same-graphically
[3] - https://www.imagemagick.org/Usage/compare/#doubles
[4] - https://0xacab.org/jvoisin/mat2