Danbooru

Image Sample Cleanup Project

Posted under General

@BrokenEagle98 On that note, Konachan has samples too, don't they? Aren't they running on a similar codebase as yande.re?

EDIT: At the very least, uploading from Konachan should be highly discouraged as most of their images that aren't wallpapers natively are modified quite heavily.

Updated

I support the notion, but thumbnail should probably have a wiki page written for it. It seems the usage is pretty clear though.

One thing to note though, I wonder if it would be more fitting to make it an alias than an implication? Arguably, not all image samples are thumbnails, but all thumbnails are definitely image samples. It's just if the distinction is that important to make.

It's about these two post:

Just uploaded this post #2641518 right now but the parent one which one is old post #2407411 looks like this. My question is the old post have lower resulotion as if it is sample image, it is also unfinished version of the main one so do we get to delete or flag these kind of unfinished sample images too? If that's the case please go ahead and flag it and I'll also flag this kind of in the future too.

Thanks for the answers in advance!

feline_lump said:

work in progress is the tag for these. I don't think they should be treated the same as sample images, since it's not always possible to tell whether an image is a work in progress or a sketch (which will never be revised/completed) at first glance. Some people also find it interesting to see the creative process behind a well-drawn image.

I endorse this view. To call that image a sample is a lie. It was an MD5 mismatch. Flaggers should only use 'sample' when it's valid tbh.

Updated

sweetpeɐ said:

I endorse this view. To call that image is a lie. It was an MD5 mismatch. Flaggers should only use 'sample' when it's valid tbh.

It's still treated as a sample since it's a downscaled md5 mismatch though (which suggests either it's an old sample or resampled from the uploader's computer), so I consider it valid. It's just that the original of the WIP wasn't uploaded yet.

Mikaeri said:

It's still treated as a sample since it's a downscaled md5 mismatch though (which suggests either it's an old sample or resampled from the uploader's computer), so I consider it valid. It's just that the original of the WIP wasn't uploaded yet.

Would it be a valid image sample if say an uploader took a screenshot of a picture from an artist's stream and then uploaded it and used the Twitter upload as a source? I think an image is a valid sample only if it's a downscale provided at the same source, for instance a 'large' or 'original' from Twitter. As we have defined image sample, it's only valid to flag an image as a sample if a) the image is not the fullest quality available from the source and b) a superior version of the image is already active on Danbooru. Appending 'sample' to a flag when other reasons are present (for example quality check + image sample) is not valid nor is flagging an image as sample when there is a misappropriated source as in the post Provence mentioned.

sweetpeɐ said:

Would it be a valid image sample if say an uploader took a screenshot of a picture from an artist's stream and then uploaded it and used the Twitter upload as a source? I think an image is a valid sample only if it's a downscale provided at the same source, for instance a 'large' or 'original' from Twitter. As we have defined image sample, it's only valid to flag an image as a sample if a) the image is not the fullest quality available from the source and b) a superior version of the image is already active on Danbooru. Appending 'sample' to a flag when other reasons are present (for example quality check + image sample) is not valid nor is flagging an image as sample when there is a misappropriated source as in the post Provence mentioned.

Under a broader set of circumstances I would say it is (because screencaps in my opinion are 'sampled' from the user's screen, and may widely vary depending on what software is being used and/or screen resolution at worse), and moreover that would be the fault of the uploader for not indicating the source correctly. But besides that, one problem with Twitter (and other sites in general) is that we don't know if in the past the site has changed their scheme for resizing and using image samples. So for Twitter, it would be extremely inconsistent -- we would have to have another way of calling something a downscaled md5 mismatch as a valid flag reason IF we know there are no other known places that host the same exact resized image, whether it's truly valid or not (as in if the image is no longer live from any other source by the original artist).

As for why 'sample' is put in the flag reason, I use that to indicate to anyone who would approve it to upload the original instead, so I don't read it as a 'reason' per se. Typically whenever I'm going through posts, I will usually reupload the ones that I think are worthy to the site, but then flag ones that I don't think belong with the foremost reason first.

I consider any image sample an image that has been resampled from the original, whether that's by the user or by the source site in either their past schemes or in present schemes. There are Twitter downscaled md5 mismatches, for example, that differ by just a single pixel in size, but that's enough of an indicator that it has been modified. One can even perform a diff to see that.

EDIT: Although I will say I should have been more clear when flagging post #1985860, as the sketch is quite high resolution. Perhaps I should've mentioned that I didn't see a real point in uploading the sketch when the original's already there, but on second thought I think it would've been of value to other users.

Updated

Mikaeri said:

create alias thumbnail -> image_sample

Link to request

Reasons aforementioned

I'd say an implication instead. Doing this erases information since there is no specific source qualifier like for image sample tags. A thumbnail may have been thumbnailed by a third party website for instance.

EDIT: This bulk update request has been rejected because it was not approved within 60 days.

EDIT: The bulk update request #1107 (forum #127512) has been rejected by @DanbooruBot.

Updated by DanbooruBot

Hmm, you do have a point there. I'll change it.

A good number of the posts in thumbnail were populated by henmere back when nicoseiga sample didn't exist yet, so I'll probably remove those also. Technically, nicoseiga samples (along with deviantArt samples and the like) are also thumbnails, but their *_sample tags best describe those posts.

Mikaeri said:

Hmm, you do have a point there. I'll change it.

A good number of the posts in thumbnail were populated by henmere back when nicoseiga sample didn't exist yet, so I'll probably remove those also. Technically, nicoseiga samples (along with deviantArt samples and the like) are also thumbnails, but their *_sample tags best describe those posts.

Well those could be changed manually I guess. Or just neglected and forgotten since they will be deleted anyway, seems like a waste to expend effort on them.

Updated

sweetpeɐ said:

I'd say an implication instead. Doing this erases information since there is no specific source qualifier like for image sample tags. A thumbnail may have been thumbnailed by a third party website for instance.

Does it really matter what site did the sampling? It's still an inferior downscale and ought to be replaced and deleted. Why make the distinction?

On the topic of works in progress, I'm of the opinion that they definitely should not be flagged just because there's a final version available. Whereas a sample is strictly inferior and recreatable, a work in progress is actually a distinct work of art. Usually the final version supersedes it, but sometimes the early version has a certain charm that is absent in the finished product (obviously subjective), and occasionally is higher digital quality (post #2142176 for example, it's not really common but it happens more than you'd think). I have saved a few work-in-progress images that I haven't felt confident enough to upload, like this early version of post #2031348 and this of post #2246499. It's not that I think they're better than the final version, I just like to have them around as well, for one reason or another. And as someone already mentioned, it can be interesting to see the steps to the final product, particularly if you're an aspiring artist yourself.

In any case, they certainly shouldn't be tagged as samples. I think image sample and downscaled may be getting used a bit too zealously right now, sometimes by people who aren't clear on their actual meaning.

shrug Mmm.... you have a point, which is what I was thinking about also. The thing is, thumbnail, even now, is used so sparingly that I don't think an implication is worth it over an alias (and especially given that most if not all images tagged thumbnail are on the road to deletion). So it will probably be changed back.

As for WIPs, I think that's about right. There's value to other users (and especially artists) in having those images to work with. But it makes me wonder though, and this might be a little bit off topic, what is exactly our stance on things like that? As in, image collections or ugoira tagged メイキング (making) on pixiv/seiga which describe the creative process? I'd be fine with approving them, but it may be a little subject to content given the 'high-quality content policy' as this would also include very rough sketches (take the middle pages from pixiv #61606173, for example). Partially uploading sketches is also something I'm not a huge fan of, so I'm wondering if it would require special criteria. Perhaps in cases like these, the final product's quality is the deciding factor for the whole set? Or would only 'very close to finished' images (like the aforementioned post replaced by sweetpea) be acceptable?

But yes, WIPs/sketches are directly distinct from samples and shouldn't be tagged as such unless the image itself is a sample. They're unrelated and serve different purposes. Though I do wonder because I haven't been paying attention all too closely, there are users that misappropriate image sample and downscaled? That's a little bit concerning.

☆♪ said:

Does it really matter what site did the sampling? It's still an inferior downscale and ought to be replaced and deleted. Why make the distinction?

Does that include post #1966693? The last time a conversation was started on twitter pngs, they decided twitter pngs, though visually lossless, were inferior in quality to conventional pngs. Following that train of thought, even if the resolution is the same and the filetype is lossless, md5 not matching is md5 not matching, and isn't the one uploaded by the artist and therefore sampled and inferior.

1 2 3 4 5 6 7 8 9 10 14