Danbooru

Filesize inconsistencies between pixiv and other sites

Posted under General

psich said:

Bump for resolution regarding Twitter/Pixiv parenting for PNGs.

If they are otherwise exactly the same (same resolution, no revisions, etc.), then the source that keeps the original file size should be the parent, so Pixiv.

EB said:

If they are otherwise exactly the same (same resolution, no revisions, etc.), then the source that keeps the original file size should be the parent, so Pixiv.

How does one tell what the original file size is if they don't have access to the original file?

CodeKyuubi said:

How does one tell what the original file size is if they don't have access to the original file?

post #2000942

Pixiv: http://i2.pixiv.net/img-original/img/2015/06/17/20/14/07/50945645_p0.png 853.82 KB

Seiga: http://seiga.nicovideo.jp/seiga/im4933786 853.82 KB

Tumblr: http://digitalflyer.jp/image/117928847779 853.82 KB

Twitter: https://twitter.com/un9man/status/594454266997116928/photo/1 (https://pbs.twimg.com/media/CD_tAV1VIAAK_x2.png:orig) 899.25 KB

Of course, the Twitter source is the only one that is different.

It's a known fact that Twitter resaves/resamples PNGs like it does with JPEGs in an attempt to reduce the size but utterly fails this since they don't convert the PNG into a JPEG. The end result is an increased filesize (sometimes it might actually be smaller than the original).

If you want further proof, take the example I used a above from either pixiv, seiga or tumblr, upload it to twitter yourself and check the :orig version's filesize. It will go from 853.82 KB to 899.25 KB.

Simply put, PNGs from Twitter will not have the original filesize.

CodeKyuubi said:

Given png's lossless nature, does it matter if it's resaved using a different algorithm? It feels like we're worrying too much about little things of no consequence.

Yes, because the algorithm could either reduce or increase the filesize as a result, thus changing the hash.

Whether or not the image quality changes is irrelevant. The fact of the matter is that the change in filesize can be considered a third-party edit and those have always been made the child of the original should it be uploaded.

I'd also like to make a sidenote on this thread's original topic: pixiv no longer strips the metadata from JPEG uploads, though I'm sure everyone's already figured that out.

Ars said:

Whether or not the image quality changes is irrelevant. The fact of the matter is that the change in filesize can be considered a third-party edit and those have always been made the child of the original should it be uploaded.

I'd also like to make a sidenote on this thread's original topic: pixiv no longer strips the metadata from JPEG uploads, though I'm sure everyone's already figured that out.

What are your opinions on scans with creases?

I feel like this discussion is a lot like a library complaining that two books of the same authorage and title have a different publisher, despite having the same content. On the opposite end, I can point to scan fixes as examples where the third-party edit should become the parent, given its completeness.

It's kind of like being anal about 'completeness of the database' and going on to pixiv and uploading identical, non-revision images with a pixel width difference of 1-10 (Just to give an example) just because their md5 tag has changed.

Edit:
tl;dr
I honestly don't think this issue is of much importance, given the identical picture quality.
Jpegs are an entirely different story, naturally.

Updated

CodeKyuubi said:

What are your opinions on scans with creases?

I feel like this discussion is a lot like a library complaining that two books of the same authorage and title have a different publisher, despite having the same content. On the opposite end, I can point to scan fixes as examples where the third-party edit should become the parent, given its completeness.

It's kind of like being anal about 'completeness of the database' and going on to pixiv and uploading identical, non-revision images with a pixel width difference of 1-10 (Just to give an example) just because their md5 tag has changed.

Why are you bringing up scans? Differences between two (or several) scans is an entirely separate matter and has nothing to do with the topic at hand.

CodeKyuubi said:

Edit:
tl;dr
I honestly don't think this issue is of much importance, given the identical picture quality.
Jpegs are an entirely different story, naturally.

Like I said, the quality is largely irrelevant. What we're discussing here is what to do in the event where a pixiv/seiga/nijie/blog/etc version of an image uploaded to danbooru is identical (resolution and no revisions) to an earlier upload from twitter. The non-twitter source is most likely the original and should be made the parent since we know for a fact twitter messes with images uploaded to its service.

Ars said:

Why are you bringing up scans? Differences between two (or several) scans is an entirely separate matter and has nothing to do with the topic at hand.

I'm not talking about differences in different scans, I'm talking about the difference between the original scan and a fixed one, which is the same image, but the second is edited to try and look as close to the original as possible, but is still a third-party edit of the first-party's (the publisher, for whom the artist has created and sold the artwork) image.

As to the second post:
I'm under the impression that the vast majority of users parent everything to the pixiv as it is, regardless of where it came from.

CodeKyuubi said:

Does the hash affect image quality?

The hash is the unique md5 code that each image has. Image quality will not be affected by saving it again as a png with a different algorithm but the image will no longer be exactly the same.

CodeKyuubi said:

I'm not talking about differences in different scans, I'm talking about the difference between the original scan and a fixed one, which is the same image, but the second is edited to try and look as close to the original as possible, but is still a third-party edit of the first-party's (the publisher, for whom the artist has created and sold the artwork) image.

Okay, I think I need to clarify something here: when I said "can be considered a third-party edit" earlier, I meant that the hosting service the artist uses to upload their work (that is, twitter and previously pixiv) made a change to the original file that the artist did not intend (stripping metadata, trying to compress it, whatever). When I say scans are irrelevant to this discussion, it's because we are strictly talking about digital sources. Pixiv, Seiga, Nijie, Tumblr, Tinami, Twitter, etc.

I nekro this thread because Toks wants me to do so.

In the TOS you can find this:

You can now upload watermarked, poorly compressed, and nude filter images, but they are discouraged and the original uncompressed artwork is preferred

We know that twitter recompresses all images, jpg and png, resulting in worse quality and/or bigger filesize. This is a third party edit, the image is no longer original.

All other sources like pixiv deliver the original files as they were made by the artists.

I read the TOS and see it's encouraged to upload the original file from another source and parent it. Toks told me to not upload the original art, IMHO that doesn't correlate with the TOS at all.

I told him so and he wants yet another discussion about it, so here it is. Let's throw light on the matter.

I'm talking specifically about pngs. pngs are identical between twitter and pixiv, so if the twitter one has already been uploaded to Danbooru there's no need to upload an exact duplicate. jpgs have quality loss on twitter so that's a different story.

*sigh*
I already told you the pngs are not identical. The Twitter ones are recompressed resulting in a bigger filesize than the original untampered with pngs other sources like pixiv have.

the original uncompressed artwork is preferred

It's many of the artists themselves who make the active decision to release on twitter first and sometimes the only release, which might mean they are comfortable with the upload being changed, if they're aware that it does. I have no way of knowing or proving that's the case all the time, but for the ones that are, does it still qualify it for being a third party edit?

Updated

Twitter recompresses everything and so is inferior anyway. There was even discussion if twitter uploads should be generally discouraged because of that. I don't think so since sometimes twitter is the only source and you never know if it becomes available elsewhere later. Parenting to an untampered original is the way to go.

1 2 3 4 5