New Tag Category - Meta

end_card and eyecatch could also be considered meta tags since their context isn't obvious from the image itself. They're similar in this way to official_art, game_cg, and screencap. Would these all fall under provenance?

edit: added eyecatch

I was also thinking of the namesake + object namesake, although it might be apparent in the image (if not, the tags themselves) that characters share the same name with either another person or an object.

Likewise, there are *_connection tags like color connection, weapon connection, seiyuu connection and such. The way I see them, they function as explanations for the logic behind particular crossover images, but they're not easily apparent in the images themselves unless you have prior knowledge or do research.

EDIT: How about out of character? You would have to know why that character is not behaving the way they normally do.

Shouldn't truth belong in the metatag group?

I would place truth in the same group as things like you're doing it wrong, failure, what, creepy, or when you see it. They may be subjective, but I wouldn't group them with meta tags.

Likewise with *_connection and namesake tags. They're meta in a sense, but tags like these aren't really what the meta category was intended for.

Format, dimension, etc.: bmp, thumbnail (all deleted, but even so)
Image source-derived info: making of, fanbox reward, source filmmaker & easytoon
Other available version info: thumbnail surprise
Derivative work: phone wallpaper, other subtypes; wallpaper has been made meta.
Potentially of the animated exception: music video, video, song and surely seizure.

+1 spoilers. Perhaps not the intention, but definitely meta information for over-the-horizon consideration.

I have changed spoilers and some of the others you linked - I'll wait for others to chime in for the remaining.

Seizure was going to be moved to "seizure warning" so that tag could be used for actual depictions of seizures, but the bulk request hasn't been approved yet. Not sure whether to wait until then and leave that tag as it is now.

I just wanted to voice my thoughts and say that jpeg artifacts seems like a great candidate for a metatag. You could make the argument that anyone can see that an image is high resolution, an image is animated, or that an image needs a translation just as much as you can see jpeg artifacts. The point is that from a system level, jpeg artifacts do not necessarily constitute intentional artist inclusion in the image.

Metatags should be the artifacts from the creation: formats, sizes, untranslated works, filetypes, special considerations, and things created from the export to upload pipeline. They're data about data.

Why am I adding feedback? Because when I learned that the meta tag system was introduced, I actually tried to add jpeg artifacts to an image purely because I didn't see it in the orange metatag section.

nonamethanks said:

I have changed spoilers and some of the others you linked - I'll wait for others to chime in for the remaining.

Seizure was going to be moved to "seizure warning" so that tag could be used for actual depictions of seizures, but the bulk request hasn't been approved yet. Not sure whether to wait until then and leave that tag as it is now.

It's actually epilepsy_warning.

That aside, I'd like to add that I'm in favor of Type-kun's methodology for metatagging.

It seems there's either confusion or disagreement on what Meta tags are fundamentally supposed to be.

My view is that they are non-content-related tags, that describe either technical information about the image file or information about its status on Danbooru or its original source, rather than what the image is or depicts.

The purpose is to create a practically useful distinction that makes gallery maintenance easier. To that end, the Meta tag category should be kept small (especially if it's being moved to the top) -- we don't want to end up with "General 1" and "General 2" and pointless philosophical arguments about which one a tag belongs in. It doesn't matter if something can be determined by a hypothetical uninformed viewer by looking at the image or not. It matters if the tag describes the content / identity of the image or its source / status / digital properties.

Some rules of thumb are:

• In the case of two posts of the same picture from different sources, could the tag be different between them?
• Could the applicability of the tag to a given post change as a result of something happening on Danbooru or elsewhere, without any change in the definition of the tag?

If one or both are a "yes", it's probably a good candidate for a Meta tag. If they're both "no", it's probably a content tag and should be left in the General category.

To elaborate, some examples / tags brought up so far and my thought process on each:

• highres, jpeg artifacts - Yes. These are properties of the image file, not the content of the picture. They might apply to one source and not another.
• translation request, weapon request - Yes. This reflects the status of the post on Danbooru and may change as the post is translated and tagged.
• bad id, md5 mismatch - Yes. These are source-dependent and may change as the original source is deleted or replaced.
• everyone - No. This describes what's in the picture.
• dakimakura - No. It's more about what the picture is (or why it was created) than its digital properties or status.
• hard translated - I'd say yes, but I don't feel strongly about it.
• comic, silent comic, 4koma - No. These describe what the post actually is / what appears in it, not affected by source or encoding.
• tag group:family relationships - No. Absolutely not. These describe what's in the image. It doesn't matter if they're contextual or not, they're content tags.
• spoilers - My instinct is yes, even though it doesn't really fit my criteria. I think this is just kind of a special case. I could probably be convinced of either way.
• game cg - I would say no, mainly because I don't think it would be of particular interest maintenance-wise and I'd rather keep the Meta section small. The question here is, what actual purpose does it serve to put this tag in the Meta section? I see none.
• key visual, novel illustration - Again, no, there just isn't much point. Even if it's technically source-related, I don't see it being relevant to site maintenance. It really describes what the picture is more than its status. You won't have one version of an image that's a key visual and one that's not.
• scan - Yes. This is not about what's in the image, but more of an external factor. I could see the argument for leaving it out to keep the Meta section small, sort of like game cg. However, the same picture often has both a digital source and a scan source, with notably different non-content visual properties as a direct result. There isn't really such a situation with game cg. So I'm still saying yes.
• work in progress - No. It doesn't quite describe the content of the image, but it's still a property of the picture itself and not the digital image. It's a tag of general interest, not of maintenance interest.
• twitter username - I want to say no, but it's a little point of conflict for me. They aren't really content in most cases and answer "yes" to my first rule of thumb (Pixiv uploads sometimes remove them), but subjectively I just don't think they're worth putting in the Meta section. They are occasionally placed in some creative way such that the image wouldn't be the same without them, so that's another point in favor of treating them like content tags.

fossilnix said:

Are meta tags supposed to be meta for the image file, or meta for the picture content?

I would say unequivocally the former. The latter is both too hard to define and contains too many tags, and is less useful to separate from content than the former. (This is sort of what I already said above, but I noticed this after writing that and it might be a more succinct expression of my position.)

And of course that's my opinion and there's room for others, but I didn't want to just write IMO before or after every sentence.

The question here is, what actual purpose does it serve to put this tag in the Meta section? I see none.

You cannot tell if a picture is game cg from the picture alone. There is absolutely no way to tell them apart from regular artist cgs without relying to external info related to the property of the picture itself (in this case being that the picture was distributed in the form of (possibly compressed) game files and consequently extracted from therein). This kind of information is by definition called "meta".

The same thing goes for key visual, novel illustrations and official art. There's no way to tell where they come from unless you're aware of something that not only is external to the picture itself, but also to the copyright and characters it's depicting. If spoilers and official art are meta tags then the other two should definitely be too.

Hillside_Moose said:

It's actually epilepsy_warning.

Ah, I didn't notice the request had been filled. Doesn't this mean most if not all of the pictures tagged as seizure should now be moved? Not sure why they weren't with the mass update. Or is it again a problem with the delayed jobs processing being stuck?

nonamethanks said:

Ah, I didn't notice the request had been filled. Doesn't this mean most if not all of the pictures tagged as seizure should now be moved? Not sure why they weren't with the mass update. Or is it again a problem with the delayed jobs processing being stuck?

The latter.

Work_in_progress is describing something that is put by artist.
This can range from flat_shading to simply sketches.
Take for instance a lot of nori_tamago sketches. One would tag them as sketches from a visual standpoint, but it's the artists themselves who say that the images are still in process...something you can't see by looking at the thumbnail alone.

i want to find an image for my facebook cover, what tag i should use?

just a random user that doesn't talk much here, but here's my 2 cents on it...

Type-kun and ☆♪'s lists of things to consider for meta are really helpful, but to really solidify the picture (no pun intended), i think we should all look at the definition of metadata - that is, "data about data". i'd say a tag would fall under the meta category when it describes a post as a file rather than as a picture, if that makes any sense.

I don't think the theoretical definition of the term "meta" is really important here. The original idea for the tag category's name was "technical", though that doesn't entirely capture it either. Since this discussion, I'm actually thinking of them more as "status" tags... wish I had thought of that name earlier.

The driving question is: Why do we want these tags to stand out or be separate from the General ones?

Some of them indicate actions that a user might want to take, or warnings, so they want to stand out. Another thing that sets them apart is that they're not intrinsic to the picture itself - the tags could differ on a different image file that's the same picture (like freecom said).

Whether or not you can tell something just by looking at the picture doesn't seem important in and of itself to me, but it's been mentioned several times in this thread. So can someone tell me, why does that matter? Not based on a theoretical definition of "meta", but why does making those tags stand out actually improve the experience of users on the site?

nonamethanks said:

If spoilers and official art are meta tags then the other two should definitely be too.

I agree that official art should be the same as game cg, but I would have said "no" for both. I don't feel strongly about that, however, it was more just to keep the category small. Spoilers like I said feels like a special case, although I wouldn't say it's the same type of tag as game cg or official art anyway.

Having thought about it more, I'm backing off a little bit on my stance. A strict maintenance-utility view would probably exclude even tags like highres, and I do feel like those should be included. So maybe game cg and friends can be included as well - there are quite a few of them, but only one or two will apply to each post. I still think it's important to keep the section small, though. The average post should not have more than 5 tags in the Meta category, IMO, or it'll be diluted.

I see a few types of non-content tags, and we need to decide which of them "count".

• Status of the post on Danbooru: translation request, tagme
• Status of the post at source: bad id, md5 mismatch
• Technical properties of the image file: jpeg artifacts, highres, image sample
• Modification (usually undesirable): photoshop, upscaled, hard translated
• Source information with major visual+technical implications (analog vs. digital origin): scan
• Source information with minor visual+technical implications: screencap, end card, video crop
• Source information with no visual+technical impact: official art, game cg, key visual, novel illustration, album cover
• Format or layout of the picture: comic, 4koma, dakimakura
• Level of completion / polish: work in progress, making of, sketch

I think everyone agrees that status and technical / modification tags should be included. Most of the disagreement is over the source info ones. I think the analog/digital distinction is important enough that scan should be included. I'd also include screencap and friends for the imperfection they imply. The other source info tags, I could go either way on. The last two categories, I'd omit - they're part of the identity of the picture, unrelated to source or technical properties.

So what do others think? Which of those categories should be included? (Or am I missing a category?) From there it should be pretty easy to decide which individual tags count.

what am i doing with my life

Jpeg_artifacts should be treated as an exeption, because they might be technical aspects, but they are clearly visible on the image per se.

As an announcement, tags with post counts > 10000 must now be approved though a Bulk Update Request

But, they don't have to be greater than 10,000 to be changed that way. Therefore, I'd like to propose that any tags that are contentious be brought up and changed with a BUR. Tags which share a high degree of similarity could be proposed in the same topic, much in the same way that mass updates are currently done for potentially contentious tag changes. This will allow the different sets of tags to be argued individually on their own merits instead of trying to solve everything with tag categories on a single thread/post.

Edit:

Just tried to create a BUR, and found out that this is currently broken. Will update this thread once it gets fixed.

@☆♪ if we're not taking the word "meta" that literally... reading over your new list, it kind of sounds like they're all properties of a picture (like right click a file -> "Properties...", that sort of property), so maybe they can be named as such instead of meta?

edit: i know this tag category is a relatively new thing, but i went through help:tags and help:cheatsheet to see if there's anything about the new category - turns out there's already a metatags section in the latter. so that's going to be a little confusing if this new category of tags is still going to be called meta...

In general, I wanted to chime in and say that the thoughts @☆♪ presented on this issue are extremely well thought out and I agree with them. If the goal is to have them stand out for being actionable, then I would say we should move to only track things that we absolutely care about as far as an action goes (translations, for example).

If the goal isn't, then I would say that technical properties, status at source, status on danbooru, modifications, and quality/source info should be in the meta group. And that's all.

Re: forum #139258, forum #139341

I think this is well-written, and I pretty much agree with this approach. Just my short 2c for what's missing in either tag group:metatags or not categorized into meta yet, though:

• bad revision -- along with its family of subtags. It's not included in tag group:metatags yet, but the adoption seems normal now given the presence and support it has. Partly related to this.
• All paid reward tags also need to be included. But I'll propose a BUR for that soon.
• jpeg artifacts is still a gentag. So is scan artifacts, and lossy-lossless.

I haven't been around the forum as much since I'm trying to stay lowkey :P