Sizes are different. Please check carefully before flagged for deletion. Thanks.

The size is the same.

Both image has same quality (99%).
See comment #213176

Usefull site.

When I said size, I meant vertical and horizontal size.
And the only difference between these differences in data size is the difference between baseline and progressive JPG compression formats, and all dots in the picture are exactly the same.
Also, as an addendum, even a difference of a few percent is considered the same here. So please remove it.
Thanks.

BattlequeenYume said:
The size is the same.

I saw the other one is 1.98 MB, and the one that I posted is 2.17 MB. Right?

That doesn't make any difference.

Basically what hiroimo2 said.

See also: comment #200502

even I think the pixiv version should be greater than the twitter version.

No longer the case when Twitter stopped re-compress (most of) JPEG images.

fireattack said:
Basically what hiroimo2 said.

See also: comment #200502

Noted.

Thank you for done it.
Do you know that there are two different compression methods for jpg baseline and progressive?
Because twitter uses progressive compression, the same picture will have a slightly smaller data size.

hiroimo2 said:
Thank you for done it.
Do you know that there are two different compression methods for jpg baseline and progressive?
Because twitter uses progressive compression, the same picture will have a slightly smaller data size.

I see... Thanks for your information.

If you have time, you can feed the URLs of similar images into this page:
https://duplicatebooru.herokuapp.com/

Basically an image from two different sources can share the same hash and jpeg quality.
It's tagged as a "pixel-perfect duplicate", and one will likely be rejected depending on which was posted earlier.

hiroimo2 said:
Because twitter uses progressive compression, the same picture will have a slightly smaller data size.

The différent file size come from the header (thumb or not, exif format, software meta...). jpeg data size is the same. Twitter delete exif and meta data on all picture and like this he replace the header.

That isn't contradictory to what he said.

Actually, headers etc. don't take that much space in most of cases (unless you have batch of Photoshop generated info like XMP), and we strip metadata anyway.

The progressive mode is exactly what causes the size difference here.

>and we strip metadata anyway.
yandere?
I strip all metadata from my posts, except in some rare cases, which so far I haven't found, because there's sometimes too much of metadata. And for archival purposes it is better to keep it.

Yes, yandere. Not at the very beginning btw, just recent years. Sans embedded color profile, of course.

So color profile stays intact; that's good, because I think I found the case where it should be left untouched. I will report in my future posts. FYI, see comment for post #991164. I still have to test properly, where colour profile matters.
In any case, without metadata it would be hard to tell post #29906 is a photograph. Sometimes images contain useful info, for me at least.

@Arsy, @Genex duplicatebooru.* is https "protected" by cloudflare and amazon, which both are VERY hostile platforms. There are fine offline FOSS tools like GraphicsMagick, ImageMagick and exiv2 for image comparison, converting and metadata manipulation. herokuapp is a well-known malware hoster.