Danbooru

Identical Twins

Posted under General

Would it make sense to have a tag for identical twins? Sepcifically I'm thinking of cases where you have two people, and can tag them both as independant characters, but because they are visually identical you can't tell them apart. (e.g. post #1147539)

We have twins already and once you exclude brother/sister twins (Kagamine Len and Kagamine Rin would never count) identical twins are still very prevalent almost to the point of redundancy (there are very few fraternal same-sex twins it seems; maybe Hiiragi Kagami and Hiiragi Tsukasa are?).

The reason I add the "can't tell them apart" clause above would be to exclude easily differentiable pairs like Futami Ami and Futami Mami, and thus maybe give the tag greater power to narrow things down and mitigate the redundancy. I could see us going with either definition though -- "all identical twins", or "all visually identical twins".

As an aside, this is a circumstance where brother and sister is legitimately useful for non-spoiler reasons in excluding obviously fraternal twins.

Updated by jxh2154

my opinion on this matter is i still oppose another creation of new family relationship tags. they're unnecessary. it doesn't matter if one can't point which one of the twins is which. what matters is they are locatable/present in the given post and searchable in any query. and we already have look-alike tag.

uhm, the look-alike wiki states: Two or more different characters whose physical attributes greatly resemble each other. what we call identical twins can fall into this, without implying family/blood relationships. look-alike, imho, is neutral and objective. though some posts may need some cleaning.

if somehow the look-alike wiki isn't enough (though i'm not sure why), how about institute a new neutral tag like spitting_image or a better one.

EDIT: or remove the alias of appearance_connection and retain the alias look-alike and lookalike. those that are similar in a way or hint any resemblance will fall into appearance_connection. those that are greatly similar that one can't set them apart will fall into look-alike and lookalike.

Updated

Just to point out, while you may be in opposition to the creation of "new" family relationship tags you're missing some points. These tags are a subset of an already existing relationship tag, so adding them has no more impact than better sorting the already existing tag. Whether you like it or not, family and blood relationship is actually an important concept. Notably when the relationship is openly given, it greatly influences how common the characters are depicted together and how they interact, which in turn influence what to expect in the depictions.

Adding related characters to the look-alike tag would likely be more detrimental than beneficial, as it would serve to flood the tag with twin sibling images. While you could argue that we could separate sibling "look-alikes" from non-sibling "look-alikes," the only methods to do it would rely on the very same relationship tags you're trying to get rid of. There is value in non-related characters looking similar, and I believe that the look-alike tag should be restricted as to not include related characters.

I do support separating fraternal twins from identical twins, so I would support an identical twins tag. As for what goes under the tag, I get the feeling just using it to cover all identical twins would make it easier. We could always make a subtag of identical twins to cover truly/completely identical twin depictions where you can't distinguish them, though another approach might be a neutral tag somewhat like ghostrigger suggests. We could possibly simply just use identical to cover the concept of indistinguishable character appearance.

Two questions though. One, where would "identical fraternal twins" go? Such as like in post #603959 or post #646872 (a good example of being indistinguishably identical). Two, would a clone passed off as a identical twin fall under the identical twin tag? And in turn the twin tag itself?

Just as a side note: I personally think there is several separate groupings of "look-alikes" that exist (not including clone): Identical twins, identical fraternal twins, look-alike relatives (post #28814, I consider K-on's Ui under this when needed post #493727), and look-alike non-relatives (which most commonly relies on crossovers).

just for clarity, my opposition alone can never stop the creation of "new" family relationship tags or any tags for that matter. i'm just one voice. what i'm arguing here is the worth/value/usefulness of family relationship tags or in this thread the *twins tags. it doesn't matter if i like them or not, because forces greater than me or the community will ultimately decide.

the already existing relationship tags don't make my argument moot since their very existence is contestable back in forum #46910 . unless that thread ends in a resolution then probably i won't bother airing my sentiments.

i'm open to know how exactly family relationships are actually important (in danbooru) and worthy of tags that aid users in their queries.

i honestly believe that adding blood-related characters to look-alike tag is not detrimental. the wiki exists for a reason. it's not limited to blood-related. there's no reason why not to use it. and we are not forced to rely on family relationships if we want to separate siblings and non-siblings. since you know the characters' names, then use them in searches. if you happen to know their gender and count, then add them as well. the only "good" thing i see for family tags is that they make searches shorter by one tag.

i don't see why you ended up "flooding" look-alike tag when post like post #1147539) above perfectly fits into it.

Updated

The twin tags aren't only a relationship issue, so I think that's the first thing you need to stop trying to claim, as you're wrong. When you have two identical or nearly identical characters, the first assumption that would be made is that they're twins. With no additional information the most likely assumption is that the characters are twin siblings, as such the twins tag is not restricted to simply a relationship. Built on this visual information alone, identical and fraternal twin tags would still arise as fraternal would be when "identical" twins are different genders and identical twins would be those that are the same gender.

You question the value of relationship tags because they're built on outside knowledge, but the look-alike tag itself is built upon outside knowledge and would not exist without it. The look-alike tag requires knowing who the characters are, as the degree of similarity that is used under the tag for most cases does not reach a level that we would assign a tag for it. Without external knowledge, something like post #1125388 would not be tagged look-alike and we would not create a tag to mark such a degree of similarity between characters. Additionally, the only time that such similarity reaches a point that we would really tag it would be when characters appear identical. Without outside knowledge of who the characters are, an image like post #1028956 would be tagged twins as the characters are so similar that we'd believe they're twins. Somewhere just under identical appearance (post #850595), if we had to tag them, images would likely be tagged "sisters" or "brothers" without external information identifying the characters. As we have a tendency to consider that when characters are very similar and are together, it is because they're related. The last thing we'd assume without external information, is that the characters are completely unrelated.

Using the look-alike tag the way you want to, ruins its original function and as such is detrimental. The tag's function is to label images that are a crossover or parody based on character similarity, it is not exactly what the wiki defines it as.

Furthermore you're trying to lump identicalness with just similarity. Identicalness is the highest degree of similarity to the extent that we would consider them the same or at least a twin character and as such would require its own tag (which may or may not be subservient to a tag based on similarity). An issue though is that any level of similarity that is below near identicalness starts becoming an issue of subjectivity. At what extent does a character count as being similar or count as being not similar? It starts becoming rather gray, which makes it a poor choice as a tag without external information to tie characters closer together.

You might not like basing tags on outside information, but that outside information is important. It tells us not simply who is related to a character, is also tells us who isn't related. We could easily build family tags based only on visual information using depictions and similarities of characters, but it's the outside information that corrects us from incorrectly assigning similarity as due to being related. If it wasn't for that outside information, I don't think it would be that odd to find every image of Akiyama Mio only with Nakano Azusa tagged "sisters."

Updated

twins wiki states, "Two siblings from the same birth, in anime usually identical girls." you are tagging also the blood relations aside being identical (the usual case). if two characters share the same features but not blood related they are most likely not twins. twins even implicates siblings. relations. it's all about relationships. clearly. this is nothing i claim, this is what we currently have. i don't get why i become wrong in this aspect, but that doesn't make you right either.

look-alike is not built on outside knowledge. we tag what we see, right there on the given post. if two characters share the same features, regardless who they are, then they're look-alike. when you search for look-alike you expect what? of course, a look-alike - siblings or not. unless you are searching for a specific character to trim a search, then you go add the character's name.

post #1125388 doesn't fit in look-alike's definition. they have similarities (but not really strong) in the way the post was drawn. appearance_connection might be better for them or if there exist a tag for tangential resemblance. no outside information necessary. you rely wholly in the image.

post #1028956 doesn't need a twins tag. they're look-alike regardless of who they are. what's exactly your point here? are you implying that non-sibling posts will pollute the look-alike tag? or the opposite? look-alikes should be look-alikes whoever they are.

or if you're dissatisfied with the look-alike's wiki which already mentions "greatly resemble each other", then maybe add 'very' or 'exactly' in *_look-alike tag to reflect a higher degree of similarity than plain look-alike.

please substantiate your claim how exactly i'm ruining the look-alike tag. how detrimental truly it is? the wiki states so, so how come i'm doing an infraction here. if its purpose is entirely different as you claim, then how come it's not reflected there and nobody among the mod team corrected it? 6 long months since last revision. nothing.

and just in case, if ever it's truly about crossovers/parodies based on character similarity, then we could use a new neutral tag. something like very_look-alike or exactly_look-alike or whichever would do.

this thread is about characters that "can't tell them apart". having a modifier like 'exactly' or 'very' tells us one level higher. or we can just use can't tell them apart if we ran out of ideas. in any case, i don't see your justification to use relationship tags.

the outside information of family relations is not important in tagging if the characters are look-alike or not. it didn't help us correct anything for 'mistagging' Akiyama Mio and Nakano Azusa as sisters. Akiyama Mio and Nakano Azusa share resemblance but not strong enough based on the wiki. and nobody will use a sisters tag if both of them are present if we don't have family relationships tags in the first place.

and if somehow, an artist was able to draw the two characters with great resemblance, then i don't see why we should not tag it as look-alike.

Updated

Wow, this is more controversial than I would have thought. I do mostly agree with NWF Renim here though. Look-alike as it exists right now (defined as such or not) is a nice collection of close, but not identical characters usually spanning series. Because a huge proportion of the population of twins looks alike (even same-sex fraternal twins usually to within the variance in current use), we would simply be adding redundancy and swamping out the interesting posts and making thing redundant by using that tag to describe this phenomenon.

I'd actually suggest we amend the wiki for look-alike to specifically exclude twins or siblings, since they are so common as to be mundane.

As for the family tags, I still don't understand why some people are so averse to them. "Tag as you see" is a good rule of thumb, but is by no means hard and fast here, in lots of cases we tag either by meta-knowlege of the image itself or the series depicted. Subjectivity is something that we do rule against, and rightly so, but identifying two people as indistinguishable is hardly subjective.

Furthermore, I don't see how family tags are harmful. Yes they should be omitted where spoilers are concerned, but for most series these relationships are out in the open and completely obvious the moment the characters are introduced. If they don't harm, and could potentially be useful (for people looking for images of people interacting as family), what's the compelling reason to purge them?

Ghostrigger, you seriously need to get over wiki definitions. They change, easily. It all depends on who decided to go at the definition. They're not written in stone and can be changed, and we seriously don't have the manpower to be constantly monitoring every single wiki. As for the definition of look-alike, I could simply do a reversion of the look-alike wiki and that would bring in the clause that the characters are non-related that was removed from one of the earlier versions of the wiki. Furthermore what is written in the wikis does not necessarily match usage and this can be due to one of at least 2 reasons, one the definition might go against common assumption and two the definition might be written by someone who never bothered to look at how the tag is being used. I've seen enough cases during my time here where someone has written a wiki going on assumptions of what was under the tag, only for it end up being wrong or half-right.

The look-alike tag is built on outside knowledge, so as much as you stubbornly stick to your claim, it still is based on such knowledge. 94.94% of all images under the look-alike tag have 2 or more copyrights (and I can assure you, that nearly all of those is not a master copyright and a subset copyright). Of the remaining 5.06%, nearly all the images are referencing outside similarities (post #737112 looks like touhou) or you have single characters cosplaying other characters from the same series (post #923221). This tag was built for the purpose of referencing characters across series that looked similar or similar looking characters parodying/dressed up as another (28.45% of the images under the look-alike tag are also tagged solo, and of the images that are under look-alike parody they normally lack a crossover with a similar appearing character). So this tag was built to indicate reference and indicate outside information, it was not built to indicate only visual similarity.

The twins tag in a stand-alone setting is a visual based tag that indicates identicalness or near identicalness (twins original). Much like with the looser similarity assumed by normal siblings, we naturally assume a family relationship between two identical or near-identical characters. It is unnatural to assume that they're unrelated. It is only due to outside information or contextual information that any assumption other than a sibling relationship would be assumed between two near-identical or identical characters.

As much as you attack family relationships, if we used no outside information, these tags would still have visual and contextual cues to define them. In a stand-alone setting, there is information that identifies relationship, and those relationships are naturally family based. When an artist draws characters, to depict a family relationship the artist is forced to draw similarities between the characters. This similarity isn't to the level that you recognize for your tag, but it certainly exists (post #147688, post #153856) and relational tags help to label these depictions. We naturally assume a family relationship when characters are not the same but similar enough and the situation the characters are depicted in can only help fuel these assumptions. Azusa and Mio in many of their depictions would naturally draw people to conclude a sibling relationship between them without outside contextual information (post #715711). When two somewhat similar characters are acting very close, we more than likely would associate it due to be siblings. Furthermore, you can't mistag these depictions of them as "sisters" if there is no outside information, it is only outside information that labels them something other than sisters to begin with (post #740129). In short, family relationship tags when not being driven by outside knowledge become tags that indicate a level of character similarity and type of situations and depictions to expect in the images. In those cases, it is only outside knowledge in turn that denies certain images from receiving these classifications of similarity and depiction.

An artist draws characters dissimilar to indicate unrelatedness, an artist draws characters similar to indicate relatedness. This is the natural state of images with no outside knowledge. Outside information is just as necessary to indicate unrelatedness between similar characters as it is to indicate relatedness between dissimilar characters. A tag that only indicates complete identicalness is certainly possible, but the natural assumption without context is that they're twins and that they're siblings. It is outside information that defines them as not being twins, it is significantly rarer that it is outside information that is needed to define identical characters as twins.

Updated

Shinjidude said:

I'd actually suggest we amend the wiki for look-alike to specifically exclude twins or siblings, since they are so common as to be mundane.

Personally when someone says look-alike to me it excludes identical twins.

Look-a-like is someone who appears very similar to another person (related or not) but is not identical.

Does anyone really say identical twins are "look-alikes"?

As for the family tags, I still don't understand why some people are so averse to them. "Tag as you see" is a good rule of thumb, but is by no means hard and fast here, in lots of cases we tag either by meta-knowlege of the image itself or the series depicted. Subjectivity is something that we do rule against, and rightly so, but identifying two people as indistinguishable is hardly subjective.

Ya I forget what the tag was recently that was discussed but it was all based on meta knowledge.

Anyway the definition (which wikis should be using more then they do here) is thus:

look-alike
[look-uh-lahyk]   Origin
look-a·like
   [look-uh-lahyk] Show IPA
noun
1.
a person or thing that looks like or closely resembles another; double.

Very similar but not identical.

While identical_twins would use some meta knowledge so does the twins, sisters, brothers and any family tag.

I think we have to be a little flexible here.

(part 1 of 2)
before anyone misunderstands my position again, i'm stating that family relationship tags are unnecessary and replaceable by existing tags, as far as tagging in danbooru is concerned. it's the status quo though. so creating identical_twins and fraternal_twins does have precedence. there's nothing i can do about it. i've accepted this fact.

however, if twins is to be replaced, what tags should we use to locate posts? character names, genders, and counts should be enough to cover them. if this is impossible to do or there's one post that can't be found without using twins please let me know.

but do all twins look alike? alike in the sense of greatly resembles each other, as stated in look-alike wiki? some would fall but others won't. post #1147539 seems to fall into it.

the question is what tag to use for "can't tell them apart" like the given post above. i suggested look-alike as the wiki states so. counter to this is the wiki is entirely and absolutely wrong! look-alike should be exclusively for crossovers and parodies alone. i suggested removing the alias of appearance_connection and updating the wikis to accommodate two different concepts but sharing similarities. this is to prevent "flooding" or the "detrimental" effect what NWF Renim claims. first of all, there's no "flooding" because not everything fits.

if somehow this won't work, for possible (unlikely) confusion, i suggested yet again another alternative to use very_look-alike or exactly_look-alike just for "can't tell them apart". or we can just use can't tell them apart as a tag by itself. everything i've stated answers the problem of the OP. and everything in the solution presented doesn't mention or imply family relationships. i've demonstrated the family relationships tags are unnecessary.

so what makes me wrong? the wiki? where is the "flooding"? the detrimental effect? every new problem thrown against me has a corresponding solution. so what is left unsolved? however, what i propose is just one solution. others may also offer a better one.

@Shinjidude: you may not believe me (you don't have to), but i'm not trying to be a devil's advocate here. i'm doing this in good faith. what i advocate is closing family relationship tags because they're unnecessary. if we have pointless pools and wikis, we also have pointless tags. and these are a whole bunch of them. in fact it also has a tag group. i've yet to see where a family relationship tag is actually useful in searches and impossible to replicate without using other tags.
(part 1 of 2)

Updated

(part 2 of 2)
@NWF Renim: wikis can be changed easily. sure. it's not like i'm treating it as gospel truth. in fact, modifying them is one of my solutions. but since we have a working definition, we should enforce them. if a post is covered by the tag's wiki, then we use that tag. what wikis are for?

and please don't give the canned excuse that we don't have manpower. there was a thread for it before forum #67204 which touched upon it. the wiki was bumped 3 times the first day, almost after every hour. 3 times the next day. and yet another. if a mod team failed to see it, yet for another week it was bumped again. if no mod team member was able to see it in the month of october, we have a 6 long months grace period. how could you miss something so major for so long? if you believe that look-alike should exclusively be used for the old lookalike (before the alias) and nothing more (like the current definition), it should be reflected there because it will cause major confusion or "flooding" of unnecessary posts. in any case, i already offered a solution for that.

we have threads like this to settle conflicting concepts and ideas. if a wiki is written by someone who never bothered to look at how the tag is being used, then i'm open for revision. you are exactly the one who don't want revisions. you are using the outdated definition of lookalike before the alias, treating it as gospel truth. and yet again, i have offered solution that addresses this if you don't like look-alike to be used the way i initially suggested.

contrary to your claim, look-alike (the current definition) is not built on outside knowledge. you don't need outside info to know if "Two or more different characters whose physical attributes greatly resemble each other." you just have to use your eyes. the old lookalike wiki relies knowing other copyrights but it didn't justify your position to use family relationship tags for can't tell them apart. it should be pointed out that i'm not against every single outside information in the world. then artist names, character names, copyrights shouldn't be tagged at all unless they're written on the post itself.

what exactly is your point? do you still have anything to say that i didn't address? if you strongly believe look-alike should be use based on old lookalike wiki solely and no questions asked, i've offered solutions and solutions after solutions for that.

when you search for family, you do exactly expect to see or filled with posts like post #147688 ? there's nothing filial about the post. its just 2girls 2boys child sex or rating:e, though this might fail without adding the necessary tags first. age_difference size_difference shota works though. post #153856 is pool material. can be replaced by 2girls 1boy child dinner or eating (this might fail again without adding the necessary tags first). child dinner is enough though for the benefit of basic members. no need to establish if the girl in the middle is their aunt, mother, an old maid, a childless neighbor or a fairy godmother in human form. no outside knowledge used. family relationships unnecessary.

why do you always assume that posts involving akiyama_mio and nakano_azusa will be tagged as siblings or sisters? is it because they look alike? in fact they are but not too strong. why not use clone or dual persona instead following your line of reasoning? you always insist that siblings tag makes the world a better place, but in fact it is not. the presence or absence of siblings doesn't aid the user in searches. when you see two people looking alike, they are look alike. if they look alike and share the same surname, then you would assume they are twins. you don't assume any two persons looking alike to be twins (or if a lesser degree, siblings) outright.

post #740129 is 2girls symmetry symmetrical_hand_pose height_difference or interlocked_fingers, though again may fail without adding necessary tags first.

the fact that you were able to identify the personalities, then you know they are not siblings. they have names. they don't even share the same surname, so why will people tagged it siblings? if you don't know them use tagme or character_request. again, no family tags used.
(part 2 of 2)

Updated

I hate the family relationship tags but since twin already does exist, I don't see harm in using identical_twins. I dislike fraternal_twins though, for the same reason I dislike the other relationship tags: the only time they impart any significant information is when they're spoiling something, which itself is bad.

As for identical_twins, I don't think a difference in hairstyle or clothing should disqualify otherwise identical twins from the tag.

1