Image Sample Cleanup Project

How would that work in regards to the person who originally uploaded the image, would it be removed from their uploads? Or would it be attributed to two uploaders.

That's the question I'm asking: should the uploader be kept or changed to the person replacing it? Having two uploaders isn't possible, it has to be one or the other.

Renim suggested changing the uploader, since technically they're not who uploaded that exact file any more. Thinking it over, I think the uploader should stay as is, since losing an upload could affect your deletion ratio and hurt your upload limit. Also I think people might get salty if they lose credit for things they uploaded.

evazion said:

That's the question I'm asking: should the uploader be kept or changed to the person replacing it? Having two uploaders isn't possible, it has to be one or the other.

Renim suggested changing the uploader, since technically they're not who uploaded that exact file any more. Thinking it over, I think the uploader should stay as is, since losing an upload could affect your deletion ratio and hurt your upload limit. Also I think people might get salty if they lose credit for things they uploaded.

From what I hear, only uploads within a certain period of time are counted for upload limits now (I think after Kikimaru complained about his limit?), but yeah I can totally see users getting salty for losing credit. If they didn't care about credit, they wouldn't be rushing to upload posts with 1-3 tags.

My personal opinion is credit should be handed to a non-user system account, but I don't feel particularly strongly about who gets credit for the post.

The only concern I have on retaining credit for the sample uploader is that a habitual sample uploader might be overlooked and potentially promoted because their uploads are getting corrected by others.

As far as losing credit would impact an uploader's upload limit, it's only recent uploads that count toward the upload limit. During that time period losing a post from their record should be less harmful to their upload limit than receiving a deleted post (such as with the current method of replacement), if I'm recalling the upload limit formula correctly.

NWF_Renim said:

My personal opinion is credit should be handed to a non-user system account, but I don't feel particularly strongly about who gets credit for the post.

So then DanbooruBot...? That seems like the best compromise. That way, people that get their uploads replaced can't complain that others are stealing their "credit".

NWF_Renim said:

My personal opinion is credit should be handed to a non-user system account, but I don't feel particularly strongly about who gets credit for the post.

The only concern I have on retaining credit for the sample uploader is that a habitual sample uploader might be overlooked and potentially promoted because their uploads are getting corrected by others.

As far as losing credit would impact an uploader's upload limit, it's only recent uploads that count toward the upload limit. During that time period losing a post from their record should be less harmful to their upload limit than receiving a deleted post (such as with the current method of replacement), if I'm recalling the upload limit formula correctly.

That was my main concern. Samples shouldn't be uploaded in the first place, but crediting the original uploader seems kinda contrary to the incentive of replacing samples. Crediting the person replacing them is that incentive -- if users don't want their uploads replaced by someone other than themselves, then they should take care to replace their own as soon as possible. But I do understand this is quite a hard stance.

I think DanbooruBot is the best idea, but if we're going that route then I also think DanbooruBot (the user/script behind it and the like) should just automate the process of replacing them periodically. That way, this project could pretty much be done and over with and no one would have to worry about things going lost (unless in the case of already present bad id's).

Mikaeri said:

I think DanbooruBot is the best idea, but if we're going that route then I also think DanbooruBot (the user/script behind it and the like) should just automate the process of replacing them periodically. That way, this project could pretty much be done and over with and no one would have to worry about things going lost (unless in the case of already present bad id's).

I'm a little leery about letting any kind of automated process do any kind of uploading. Seems like something prone to bugs and mistakes. Even RaisingK manually approves things that RazingK uploads.

BrokenEagle98 said:

I'm a little leery about letting any kind of automated process do any kind of uploading. Seems like something prone to bugs and mistakes. Even RaisingK manually approves things that RazingK uploads.

Can't agree more than this.

Fair enough, but I think even semi-automated would be nice (as in, there's some way to easily double check the retrieved original and the sample).

Just a PSA... I'm going to do the full gamut on Twitter starting today, so there might be more tag changes than normal going on. I'm starting one month out since those already fall within the range of the daily scans.

Full scan report

• Site: Twitter
• Range: 1Month->End
• Start time: 2017-04-09 22:20 Z
• End time: 2017-04-11 11:33 Z
• Tag changes:
• Twitter sample: 80
• MD5 mismatch: 246
• Resized: 110
• Downscaled: 96
• Upscaled: 10
• Stiched: 1
• Cropped: 11
• Bad Twitter ID: 2413
• Inactive account: 51
• Protected link: 473
• Bad source: 40

Edit:

• Post versions are working again, so I updated the above.

I think the #1 think to take away from the above info is the transient nature of Twitter posts, which makes this sample cleanup effort all the more important.

Thanks to the effort of @Qpax today, the ArtStation sample count has went down by a LOT. However, there are some that are now marked with bad artstation id AND artstation sample, which well... kinda sucks. I suppose we could remove the artstation sample tag from those posts since they're no longer 'active', but it does reveal that time is sometimes of the essence. Especially with more Twitter users being suspended now, or just cleaning out their old tweets so they don't get potentially reported, etc.

Anyways, the post counts are updated. Still a lot to go through, but if any user is interested in easy approvals, then go ahead and take a look through them. There's something like 4000 posts left? It definitely can't be done alone.

EDIT: ArtStation samples are completely wiped out! Yaay~

Eh well, "easy approval" is easy said.
Since the last bunch is Twitter, I'd rather stop uploading from there. To me, there is now a factor of uncertainty, since the remaining posts don't really convince me at all. Might be some good posts left, but the majority is more meh (at least for the queue).

I'll probably go through and flag some of them for review then. I agree that most of the ones left over from Twitter are ones that no one really cares to replace since they're just meh quality.

EDIT: Although, have you looked through the relevant MD5 samples? They're different from the genuine twitter sample ones and there's still a lot of them. You might be able to find something good.

You means md5_mismatch resized?

EDIt
And yes, that's a good idea. Maybe I'll disagree with one of the flags, but I think this won't happen.

provence. said:

You means md5_mismatch resized?

EDIt
And yes, that's a good idea. Maybe I'll disagree with one of the flags, but I think this won't happen.

Mmm, well there are those, but I mean the ones here:

• twitter_sample -status:deleted -- 855
• source:*twitter.com/ md5_mismatch -upscaled -status:deleted -- 2232
• source:*pbs.twimg.com/ md5_mismatch -upscaled -status:deleted -- 834

They're also in the OP, for future reference.

So many posts? Will take a lot of time with my 20 uploads to skim through ><.
Hope, other users will jump in, too.

provence. said:

So many posts? Will take a lot of time with my 20 uploads to skim through ><.
Hope, other users will jump in, too.

Well, yeah. That's exactly what I mean when I say that this thing can't be done alone. It's way too much stuff. Heck, I'm sure Qpax is already extremely exhausted after going through so many ArtStation samples. He says he'll get around to the deviantArt samples later in the future.

It's just that again, now there's the problem that there's too little to work with. Most of the good posts are already taken, leaving all the mediocre posts neglected. Someone needs to be a sacrificial pawn for this kind of stuff, sadly :x

Just looked through the first ten pages of the second search query. And I'm afraid that most posts, judging from their thumbnails, won't be re-uploaded by me. Still, I might take a closer look tomorrow^^.

I noticed recently that when some samples where deleted, users' favorites were moved to the parent in some cases, while they remained in the deleted version in others.

For example:

post #2451939 (user favorites were moved to the parent)
post #2618062 (user favorites remained in the deleted post)

I'm not sure how this works exactly, i.e. if it's done manually by the mod when deleting the post or what, but I thought I'd bring it up because in cases like these, where there's no revision or variation and it's just a matter of sample clean-up, it's certainly worth it to move the favorites over to the parent every time, provided the process is simple enough.

@Fred1515
It is only done manually.
Not sure if it still works after the image is deleted, but it seems that it was just forgotten here.