So, here's something I learned recently - despite all the cyan_* tags being aliased to blue_*, they are actually not the same color. Cyan isn't just "light blue" or "greenish blue." It's as much a distinct color from blue as red is from yellow. In fact, it's one of three primary colors in subtractive coloring.

I first learned this after hearing that Russian and Italian speakers are apparently (slightly) better able to distinguish between blue and light blue/cyan. The reasoning for this is because those languages have a distinct word encompassing light blue and cyan, which is actually considered a primary color distinct from true blue, whereas in English (and most languages, actually) all shades of blue and cyan are generally considered shades of blue. In fact, in some languages, green, blue, and cyan are all considered to be the same color. This is similar to how, in English, "pink" is generally considered separate from red, and it encompasses not only light shades of red, but also much of the neighboring color magenta, which is, like cyan, also a subtractive primary color. There's even a Wikipedia article describing this distinction across many languages. From the Italian section:

Azzurro, the equivalent of the English azure, is usually considered a separate basic color rather than a shade of blu (similar to the distinction in English between red and pink). Some sources even go to the point of defining blu as a darker shade of azzurro.

So, I thought this was interesting, but I didn't think much of it otherwise. That is, anyway, until I opened Krita one day and had a look at the color wheel. Notice how this particular wheel I'm using as an example does not include gradations of saturation or lightness, which helps elucidate the distinction. Clearly, cyan is not just "light blue." It actually has quite a large section of the color wheel all to itself. The same piece of the pie as blue or red or yellow or green or magenta. And since lightness is not included in this wheel, you can see that what appears to be "light blue" is actually treated with the same luminosity as the apparently much darker "true blue".

Why? Well, that's a question for physics and psychology to answer, and is somewhat outside the scope of this thread. However, it is worth noting that different hues actually have inherently different luminosity values. And this article goes over the basic psychology behind why we perceive these colors differently.

Now, if we look at this color wheel taken from Wikipedia, we can see that there are essentially six primary colors if you take additive and subtractive coloring together: red, yellow, green, cyan, blue, and magenta. Basically, every primary color* is represented by our tags except for cyan, which is aliased to blue.
^{*"Magenta," in English, is generally considered a subset of pink, or sometimes purple on the darker/bluer side of the spectrum. This color is already pretty well represented by the pink_* and purple_* tags, but they generally include light red in addition. Sure, it would make more sense to simply tag light red as red, like how light_green_* and light_blue_* and so on are generally not given distinct color tags, but we all know that this will never work out in practice. Even if we aliased pink_* to magenta_*, people would continue using it for light shades of red since, at least in the English-speaking world, magenta is so commonly considered a shade of pink or light purple. However, I think it's fine to leave pink/purple as-is. It's "close enough" for our purposes. It actually increases the precision of color tagging at least somewhat, as magenta is being divided into the darker/bluer shades in purple and the lighter/redder shades in pink by the current system. At least magenta isn't aliased to red or blue, which would be just as fundamentally incorrect as the cyan -> blue alias is.}

With that all being said, I'd like to propose removing all the cyan -> blue aliases. In fact, there was some pushback against aliasing them in the first place when it was all done about six years ago, and this pushback was never really addressed. See the discussion at topic #16756 and topic #19170. The only argument against keeping cyan was apparently the fact that people often mistag cyan as blue anyway. However, this doesn't really address the fact that the "solution" to the problem was simply to enforce the mistagging through an alias. Instead of having some posts tagged correctly as blue and some tagged correctly as cyan with some others still mistagged as blue that are actually cyan, what we now have are all cyan posts mistagged as blue with no way to correctly tag the color. Even if mistagging is an issue, having at least some posts tagged correctly is preferable to the current situation where none of them are.

There is also the closely related issue of the teal_* tags. Technically, "teal" is either a mix of green and cyan, or commonly just dark cyan with no green. However, teal is officially included in the color variants of some tags such as teal eyes, teal hair, and teal scales, having implications in place to the base tag (like how teal hair implies hair). Meanwhile, a few other teal tags (namely teal_fur, teal_horn, and teal tail) are aliased to their green_* counterparts, even though teal is often just dark cyan, and in some cases may even be closer to blue than to green. It seems that there is no "official" stance on the teal tags, as some have been abolished while others remain valid with implications. I will say, however, that aliasing to either green or blue will likely cause mistags. Teal is not always green or always blue, but it is always some shade of cyan. So, I think aliasing all the teal_* tags to their cyan_* variants will not only avoid mistags, but also help populate cyan_* with correct posts.

Now, I created this thread to get some more discussion on this proposal before I go writing a BUR for it. If anyone has any further ideas/suggestions/objections, let's hear them.

You cannot directly compare importance of a color when dealing with modern subtractive(cymk) vs additive(rgb) color spaces, nor can you compare it to traditional(ryb) subtractive color spaces. For subtractive teal is a primary color, additive it is a secondary color(no red, max green/blue), and for traditional mixing, it is a tertiary color(green-blue with lightening). Also there are multiple types of color wheels, cylinders, and triangles.

Technically RYB(W) is wrong according to color theory, but cymk is a lot less intuitive. Also additive color is even less intuitive, with red+green = yellow. That still messes with my brain.

I say stick with RYB color priorities and naming. Just like Newtonian physics, it's good enough and much easier to understand, despite being slightly wrong. Color naming and separation is really subjective at times.

deadoon said:
You cannot directly compare importance of a color when dealing with modern subtractive(cymk) vs additive(rgb) color spaces, nor can you compare it to traditional(ryb) subtractive color spaces. For subtractive teal is a primary color, additive it is a secondary color(no red, max green/blue), and for traditional mixing, it is a tertiary color(green-blue with lightening). Also there are multiple types of color wheels, cylinders, and triangles.

Technically RYB(W) is wrong according to color theory, but cymk is a lot less intuitive. Also additive color is even less intuitive, with red+green = yellow. That still messes with my brain.

I say stick with RYB color priorities and naming. Just like Newtonian physics, it's good enough and much easier to understand, despite being slightly wrong. Color naming and separation is really subjective at times.

There’s still the fact that cyan takes up a considerable portion of the visible color spectrum, and is actually quite easily distinguishable from blue or green despite the fact that it’s often not considered a primary color (in English, anyway). What I’m saying is that we have a considerable gap in color tagging, whereas in other cases we actually differentiate between much smaller distinctions such as brown versus tan. And I don’t think “traditional” coloring is actually any easier to understand, it’s just what we’re mostly used to. However, having been a digital artist for a little while now, I’ve started to feel that RYB doesn’t really make sense. It’s really biased towards the warm side of the color spectrum, leaving the cooler colors somewhat neglected.

^{small digression: “traditional color space” (RYB) doesn’t really “exist.” It’s really just incorrect subtractive coloring. You can see an example of this by trying to create green, a secondary color in both subtractive (CMY) and “traditional” (RYB). You can create a bright, vivid green by combining cyan with yellow, which is simply a hue shift. One attempts to create green in RYB, however, by combining yellow with blue, which results in a desaturated green since the shade of blue used for this is almost the opposite of yellow and so they actually cancel each other out to an extent, including a lot of gray in the mix. Also, mixing all three (red, blue, yellow) evenly doesn’t produce gray, but brown.}

Basically, what I’m saying is that aliasing away cyan/teal isn’t really useful, even if you only consider it to be a tertiary color. Brown and tan are tertiary colors, and yet we tag them. So what do we gain from getting rid of cyan, exactly? We could stick with primary colors only and get rid of orange, aliasing it to red or yellow, but I similarly don’t see that being helpful to us at all.

Alternative suggestion:

seems to be more commonly understood to be a color in between blue and green than cyan does. While technically teal is just a darker shade of cyan (as opposed to cyan being a light shade of teal), perhaps it would be more pragmatic to alias all the cyan tags to teal instead? The teal tags, at least, are still mostly in use, and I think most people know what teal is so we can therefore avoid the potential problem of people mistagging blue as cyan/teal. Anyone have any thoughts on that idea?

Perhaps I know nothing about color theory, but wouldn't blue be better thought of as an umbrella term for cyan and indigo rather than one or the other?

clawstripe said:
Perhaps I know nothing about color theory, but wouldn't blue be better thought of as an umbrella term for cyan and indigo rather than one or the other?

I suppose it could be. All the color theory materials I’ve looked at distinguish them (and make no mention of indigo), but common usage of the word seems to agree with that. However, that isn’t very useful for tagging if we want to find things that are one color or the other.

_{An aside: additive coloring specifically always calls it blue as opposed to indigo - you hear the term “RGB” a lot if you deal much with computers or computer parts, as those are the three additive primary colors that are used in monitors to create all the other colors. Blue and green light together at equal values make cyan light. Indeed, cyan (if you go to look up a dictionary definition of it) is most often described as something like ”a moderate greenish-blue to bluish-green color,” referring to the fact that it is halfway between blue and green (or indigo and green, if you prefer).}

I’d say we should just have tags for both separately and have them both imply blue_*, but a) that sounds like a lot of work, and b) indigo_* has almost no usage so far. And even so, indigo tends to include colors that are closer to purple than blue, which would be a problem.

But anyway, the fact that all the cyan color tags were aliased away while most other colors outside the “official” colors are not aliased seems to indicate that people were actually using them - and it seemed that the reason for aliasing them was not because they were themselves being mistagged, but because some things that were cyan were being mistagged as blue. Teal_* still has some use as well, at least for eyes/hair - teal eyes has nearly 6000 posts, which is much more than at least one “official” color, tan eyes. But with teal fur being aliased away, we can’t compare that one. I don’t think this color (whether we call it teal or cyan) would go underutilized, at least.

Interesting how language can affect your color perception. To me, they are two distinct colors, with different meanings attached to them. If tan and brown are tagged separately, it would make sense to do the same for blue and cyan.

A big problem with going in-depth on colour tagging is making sure everyone's operating the same standards. You can set site standard definition but you can't make every tagger care to check what the standard is first.
An example I like to point to is googling maroon vs burgundy and watching each result flip which one is which. Unless you're Professional Colour Expert™ you're going to be making wrong judgement calls somewhere and having fights over which names are which, edgecase shades, etc

magnuseffect said:
A big problem with going in-depth on colour tagging is making sure everyone's operating the same standards. You can set site standard definition but you can't make every tagger care to check what the standard is first.
An example I like to point to is googling maroon vs burgundy and watching each result flip which one is which. Unless you're Professional Colour Expert™ you're going to be making wrong judgement calls somewhere and having fights over which names are which, edgecase shades, etc

Sure, though that’s not really the case with cyan. We can roughly define it as “blue-green,” and basically everyone will get the idea of what it is. Like I pointed out earlier, in a discussion that was had prior to the cyan tags being aliased away, someone pointed out that pretty much every post tagged as cyan was correct. The issue was apparently that some people were tagging things that should’ve been cyan as blue - presumably because they didn’t know it was (at the time) a valid color, or because they didn’t know the name for it. After all, we don’t typically have other “in-between” colors like, say, yellow-green (although cyan technically is not an in-between color, it’s often considered that way).

scaliespe said:
After all, we don’t typically have other “in-between” colors like, say, yellow-green (although cyan technically is not an in-between color, it’s often considered that way).

All colors are technically in-between colors. Color is a spectrum of wavelengths between infrared to ultraviolet, and every single wavelength of color is in-between them. Our eyes have cones that focus on one of three particular wavelengths of light (with overlapping edges), which are red, green, and blue. "Cyan" is a wavelength of light corresponding to ~495nm, which when striking our eyes, excites the cones tuned to green and blue equally. As the wavelength is reduced, it excites green less and blue more, until it becomes "blue", and continuing to reduce the wavelength turns it ultraviolet and then black as it leaves the visible range. When it comes to our perception of color, any color we see is a result of exciting the red-, green-, and blue-focused cones by varying amounts, and there's no hard line between any two mixes. Add to the fact that the way the brain interprets color is influenced by other nearby colors, and not everyone's monitor will be calibrated correctly, a user's ability to discern (and thus correctly tag) more specific colors can be limited.

Fun fact: the colors a person is able to easily perceive is also influenced by culture and language. Since it's on a spectrum, the separation of colors is largely arbitrary, and the colors we learn and talk about influence which ones we can more easily identify. English didn't used to have a word for the color orange, until we started referring to reddish-yellow things as "the color of a ripened orange" (the color is named after the fruit, not vice-versa). Japanese combines blue and green under a single word, 青 (ao), which can be used to refer to a green traffic light or blue eyes, for instance.

the fact that we have 3 tags for diffrent shades of a tertiary color orange_*, brown_*, and tan_* but no tag for cyan is eternally frustrating. there's so many times were a character who has fur/hair/whatever that is _directly_ between green and blue but I just can't tag the color properly because the proper tag does not exist.

here's the full RGB hue/saturation color wheel, can anyone look at this and still deny that cyan is distinct enough to be considered its own color?

magnuseffect said:
A big problem with going in-depth on colour tagging is making sure everyone's operating the same standards. You can set site standard definition but you can't make every tagger care to check what the standard is first.
An example I like to point to is googling maroon vs burgundy and watching each result flip which one is which. Unless you're Professional Colour Expert™ you're going to be making wrong judgement calls somewhere and having fights over which names are which, edgecase shades, etc

who's going to have a tag war over whether something's blue, cyan or green? if anything the lack of cyan_* tags is more likely to cause problems because we do not have a tag for one of the primary colors, the color that's directly between blue and green.

watsit said:
All colors are technically in-between colors. Color is a spectrum of wavelengths between infrared to ultraviolet, and every single wavelength of color is in-between them. Our eyes have cones that focus on one of three particular wavelengths of light (with overlapping edges), which are red, green, and blue. "Cyan" is a wavelength of light corresponding to ~495nm, which when striking our eyes, excites the cones tuned to green and blue equally. As the wavelength is reduced, it excites green less and blue more, until it becomes "blue", and continuing to reduce the wavelength turns it ultraviolet and then black as it leaves the visible range. When it comes to our perception of color, any color we see is a result of exciting the red-, green-, and blue-focused cones by varying amounts, and there's no hard line between any two mixes. Add to the fact that the way the brain interprets color is influenced by other nearby colors, and not everyone's monitor will be calibrated correctly, a user's ability to discern (and thus correctly tag) more specific colors can be limited.

Fun fact: the colors a person is able to easily perceive is also influenced by culture and language. Since it's on a spectrum, the separation of colors is largely arbitrary, and the colors we learn and talk about influence which ones we can more easily identify. English didn't used to have a word for the color orange, until we started referring to reddish-yellow things as "the color of a ripened orange" (the color is named after the fruit, not vice-versa). Japanese combines blue and green under a single word, 青 (ao), which can be used to refer to a green traffic light or blue eyes, for instance.

I thought cyan was one of those "emergent colors" that don't exist on a rainbow and only exist by mixing two wavelengths of light, like magenta.

This was a much deeper discourse than I was expecting from the site.

I guess I shouldn't be surprised so many art majors are here, huh?

darryus said:
here's the full RGB hue/saturation color wheel, can anyone look at this and still deny that cyan is distinct enough to be considered its own color?

Looking at that, cyan makes up as much, if not slightly more, of the wheel as purple does to me. I also notice a large portion of the wheel I would label as green since there's no distinct name (that I am aware of) for the yellow-green range; all the names I know are [adjective]-green.

cyan along with amber are colors that don't have any right to have tags as far as I'm concerned considering they vary from post to post enough to fall into 3 other standard colors.

riverinadryland said:
Looking at that, cyan makes up as much, if not slightly more, of the wheel as purple does to me. I also notice a large portion of the wheel I would label as green since there's no distinct name (that I am aware of) for the yellow-green range; all the names I know are [adjective]-green.

The only reason for that being that the color is equidistant from others on the RGB hue spectrum. As Lupine was saying, and from the physiology webpage mentioned in the OP post (the part explaining the nervous optical "filters"), it's not really a visible color for humans, as blue and green override eachother when looked at simultaneously.

It could also be as cultural as the sky we see on Earth as for why "-GB cyan" is light-blue to us.

In any case, I don't see a problem with letting Cyan have its own tag again; explicitly for blues that are closer to green/seagreen than blue. Yellow typically resides anywhere between orange-yellow ("Amber," as V was speaking of), and green is anywhere between yellowish-green to 255 green as is. The ambiguity of it allows users to search for nearby colors if they can't quite remember which color a post is tagged with.

edit: According to me this is how it breaks down on the hue spectrum. Wow, green and blue are disproportionately large.

watsit said:
All colors are technically in-between colors. Color is a spectrum of wavelengths between infrared to ultraviolet, and every single wavelength of color is in-between them. Our eyes have cones that focus on one of three particular wavelengths of light (with overlapping edges), which are red, green, and blue. "Cyan" is a wavelength of light corresponding to ~495nm, which when striking our eyes, excites the cones tuned to green and blue equally. As the wavelength is reduced, it excites green less and blue more, until it becomes "blue", and continuing to reduce the wavelength turns it ultraviolet and then black as it leaves the visible range. When it comes to our perception of color, any color we see is a result of exciting the red-, green-, and blue-focused cones by varying amounts, and there's no hard line between any two mixes. Add to the fact that the way the brain interprets color is influenced by other nearby colors, and not everyone's monitor will be calibrated correctly, a user's ability to discern (and thus correctly tag) more specific colors can be limited.

Well, yeah, sorta. There are a few different ways to interpret how color works. There’s the physical light wavelengths, the spectrum reflected by a surface, the way these wavelengths are perceived by the cones in the human eye, and the psychological perception of color (the colors that we are “taught” to identify, influenced by our language).

In the spectrum of physical light wavelengths, true, there isn’t technically any in-between color - but all the different facets of color perception have to be taken together. We identify red, green, and blue as primary colors of light because our cones primarily identify those three colors. Thus, any visible color can (and must) be composed of some combination of those three colors, since those are actually the only three colors we can truly “see.” Everything else we see is a combination of those three.

But that only applies to direct light. The pattern flips when white light (all wavelengths equally) is being reflected by a colored surface. Since these surfaces actually absorb some wavelengths of light and reflect others, some colors are being stripped away from white in order to create color - hence, subtractive coloring. In that case, what we see as primary colors are actually the opposite of the additive primary colors. A primary subtractive color is a surface that absorbs a primary color of light while reflecting the other two. That’s why the three subtractive primary colors are the exact opposites of the three additive primary colors - the colors that create the most contrast when juxtaposed. Cyan is the exact opposite of red because it absorbs all red light while reflecting all green and blue light. Likewise, magenta is the opposite of green, and yellow the opposite of blue for the same reason.

This makes cyan, magenta, and yellow subtractive primary colors because those three colors, and only those three colors are able to create any other visible color subtractively. This is why the outdated “red blue yellow” system doesn’t actually work - of those three, only yellow is a subtractive primary color, which means that those three colors together cannot create any visible color. It is impossible to create a vivid shade of green with just those three, as blue and yellow cancel each other out to some extent, meaning that the closest you can get is a dark, desaturated form of green. Cyan and yellow, however, create a perfect green when combined.

So… in that sense, you could consider an “in-between” color to be either:

a) any secondary color in either additive or subtractive coloring (cyan is “in between” green and blue in additive coloring, while green is “in between” cyan and yellow in subtractive coloring)
- and/or -
b) any tertiary color in either system (orange, for example, is a tertiary color in both). In fact, any tertiary color is, by definition, a tertiary color by either system.

Fun fact: the colors a person is able to easily perceive is also influenced by culture and language. Since it's on a spectrum, the separation of colors is largely arbitrary, and the colors we learn and talk about influence which ones we can more easily identify. English didn't used to have a word for the color orange, until we started referring to reddish-yellow things as "the color of a ripened orange" (the color is named after the fruit, not vice-versa). Japanese combines blue and green under a single word, 青 (ao), which can be used to refer to a green traffic light or blue eyes, for instance.

Right. As I stated in the OP, some languages (like Japanese) have a single word encompassing everything from green to cyan to blue, whereas other languages (Italian and Russian, primarily), unlike English, actually distinguish cyan and the lighter/greener shades of blue from actual blue. I imagine it must be strange for the Italian users of this site to notice that we have all sorts of tags for blu, but none for azzurro - or to see a bunch of things that are actually azzurro being tagged as blu.

darryus said:
the fact that we have 3 tags for diffrent shades of a tertiary color orange_*, brown_*, and tan_* but no tag for cyan is eternally frustrating. there's so many times were a character who has fur/hair/whatever that is _directly_ between green and blue but I just can't tag the color properly because the proper tag does not exist.

here's the full RGB hue/saturation color wheel, can anyone look at this and still deny that cyan is distinct enough to be considered its own color?

Yeah, exactly this. We do still have most of the teal tags, however, and teal is just dark cyan. It would make more sense, though, to just use the primary color (cyan) and alias all the teal tags to it.

It’s better than nothing, but it’s still a bit absurd. Imagine if red was aliased away to something else, and we had to use burgundy tags for all shades of red, even though burgundy is technically only dark red. That’s the current situation with cyan/teal.

lonelylupine said:
I thought cyan was one of those "emergent colors" that don't exist on a rainbow and only exist by mixing two wavelengths of light, like magenta.

Nope; in fact, only magenta has that distinction. There can only be one color like that due to the way the color spectrum works. The visible color wavelengths range from red at the lowest to violet at the highest. Wavelengths below red and above violet aren’t visible. However, humans perceive hue not linearly, but circularly. That’s why we have a “color wheel” and not a “color line.” Magenta is right between red and violet on our color wheel, but it doesn’t exist on the visible light spectrum. That’s because all colors are essentially just different combinations of red/green/blue, being the only wavelengths that we can actually perceive directly, while wavelengths in between those three, like yellow (in between red and green) are simply perceived as a combination of the two nearest wavelengths that we perceive directly. So yellow can be perceived in two different ways - either as a single wavelength in between red and green, or as a combination of two wavelengths, red and green together in the absence of blue. If our visual system was structured differently, we might actually perceive these as being two different colors, but we do not.

Magenta, therefore, can be perceived by the second method, but not by the first, simply because it does not have a pure wavelength like yellow or cyan do. Magenta is perceived as the equal combination of red and blue wavelengths in the absence of green light. That’s why green’s exact opposite is magenta. Magenta is 100% of the other two primary colors (red and blue) and 0% green.

Cyan, however, actually does exist on the visible color spectrum as its own wavelength. If you look at a rainbow closely enough, you can actually find it right between blue and green. Wikipedia defines cyan’s wavelength as the range between 490 and 520 nanometers on the visible color spectrum.

riverinadryland said:
Looking at that, cyan makes up as much, if not slightly more, of the wheel as purple does to me. I also notice a large portion of the wheel I would label as green since there's no distinct name (that I am aware of) for the yellow-green range; all the names I know are [adjective]-green.

The correct name for yellow-green is chartreuse. However, it is a tertiary color and not a commonly used word.

versperus said:
cyan along with amber are colors that don't have any right to have tags as far as I'm concerned considering they vary from post to post enough to fall into 3 other standard colors.

Uh… no, they don’t? Cyan is easily recognizable as a single color that exists between blue and green. Sure, English speakers who aren’t familiar enough with the concept of cyan will see it and call it blue - but as stated before, this is primarily a linguistic distinction, as some other languages actually do distinguish between blue and cyan and consider them to be discrete colors. Linguistics aside, however, cyan is actually a primary color and so is realistically much more important than some tertiary colors that we bother tagging, such as tan.

Remember the mnemonic Roy G. Biv? It's used to remember the colors of the rainbow.

That blue? It's cyan, hence why I suggesting thinking of "blue" as an umbrella tag for cyan and indigo (although tagging would likely be a different matter). Separating cyan and blue like that is relatively recent, though. The colors of the rainbow were numbered seven because that number seemed to be an important one (seven planets/days of the week, seven notes in an octave, seven virtues/deadly sins, etc.), but people usually just break it down into six colors.

Actually, I'm warming up towards adding the cyan tags back in. If we can break down species so well, we can do the same for colors of the spectrum.

Fun fact: The colors of the rainbow isn't exactly all of the visible light spectrum. A few select women are capable of seeing into the ultraviolet a little bit, though it doesn't appear to be all that interesting outside of flowers, and reptiles and birds can also see into the ultraviolet. Also, certain eye surgeries appear to allow a person to see a little bit into the near infrared.

clawstripe said:
Remember the mnemonic Roy G. Biv? It's used to remember the colors of the rainbow.
That blue? It's cyan, hence why I suggesting thinking of "blue" as an umbrella tag for cyan and indigo (although tagging would likely be a different matter). Separating cyan and blue like that is relatively recent, though. The colors of the rainbow were numbered seven because that number seemed to be an important one (seven planets/days of the week, seven notes in an octave, seven virtues/deadly sins, etc.), but people usually just break it down into six colors.

Yeah… I kept thinking about this. In some sense, the rainbow color naming is arbitrary, possibly maintained simply by the popularity of the mnemonic. “Indigo” is simply not a word that’s used very often, probably even less often than cyan. It would be easy enough to call it cyan and blue (ROYGCBV), but then you lose the mnemonic. As far as color theory goes, though, it’s pretty much always divided cyan/blue (RGB and CMYK).

^{An aside because I’m a music nerd: there are actually twelve notes in an octave; however, there are seven notes in the major scale and its six modes.}

Actually, I'm warming up towards adding the cyan tags back in. If we can break down species so well, we can do the same for colors of the spectrum.

Blue as an umbrella term would probably work, actually. It would require no retagging - only changing the cyan aliases into implications. In fact, I don’t see why we don’t just do that with all the sub-colors we have for things. There are already a number of magenta_* tags with a bit of use, and those could probably imply purple or pink - given that magenta is a fairly distinctive shade (it is a primary color, after all), I could imagine people would want to be able to find magenta things specifically rather than any old shade of purple. Same goes for cyan - in fact, that was partly my reason for creating this thread in the first place. Cyan is one of my favorite colors - much more so than any old blue. My OC wears cyan, but I can’t tag cyan_clothing, or search for art with cyan clothing in it. The best I can get is blue, which is insufficient. And as you pointed out, we have an extremely detailed tag structure for species in place. Given how important the color tags are, I think we ought to do the same for them. I think the idea behind aliasing away some of these tags was because they didn’t want these more obscure colors to be tagged instead of the more common colors - but that isn’t a problem if the niche colors imply the common colors. After all, any specific shade of a color is also an instance of the broader color, just like how any lizard is also an instance of a reptile.

I understand that not all instances of cyan will be tagged as cyan - many people will continue to just tag them as blue. But at least if we use blue as an umbrella tag for cyan and indigo, those won’t be mistags - referring to the original justification for aliasing cyan away in the first place. As long as cyan gets some usage, it’ll be useful even if it doesn’t cover everything. It’s already like this regarding our species tags - wolf currently has 201982 results, while arctic_wolf has only 1337 (lol), even though I’m certain that far more than 1337 of those wolves can be identified specifically as arctic wolves. Most people tend to go for the generic tag even when a more specific tag applies - but at least arctic_wolf exists for those who wish to find that species specifically, as opposed to just being aliased to wolf because it isn’t applied in all cases. It’s far more useful to keep a somewhat underutilized tag as an implication instead of getting rid of it completely, which doesn’t really help anyone.

scaliespe said:
Blue as an umbrella term would probably work, actually. It would require no retagging - only changing the cyan aliases into implications.

But cyan is literally as much blue as it is green. Why put it under the umbrella of blue and not green? Especially for shades of cyan that lean a touch more on the green side (or the color palette used makes pure cyan appear more greenish). That'd be like having yellow implicate green.

FWIW, I'm fine with having cyan as its own color tag. But I don't think we need to build color hierarchies, given the number of color-related tags there are.

Unlike so many other things when it comes to tagging. Color is fairly objective, just sample an area, take the hex code or RGB value and go look up what color it actually is. The color space of the image is good to take into consideration when doing this. And some lists are better than others, perhaps e621 should just make its own to be fair. For an example, cyan/teal is a 50/50 mix of green and blue.

(However, we should likely not talk about strictly "color", but rather more lean towards chroma instead, and one can argue about the saturation of colors as well. If we want to get really technical...)

In the end, if we can accept Red, Orange, Gold, Yellow, Green, Teal, Blue, Purple and Magenta as separate color tags (as well as the gray scale ones), then we should honestly have cyan aliased correctly.

Cyan should be aliased to the existing "Teal" since it is the same chroma. (teal is technically darker than cyan.) But why is Cyan aliased to blue? (if the tag should be Cyan or Teal is another debate.)
One can also ask why the rarely used magenta isn't aliased to purple... (same can be asked about gold and yellow.)

Dividing the spectrum into: Red, Orange, Yellow, Green, Cyan, Blue, Purple, White, Grey and Black, is an adequate amount of distinction for searching for artwork. Even adding the current, light/dark shades of these colors is honestly overkill in most situations as far as searching is concerned. (and honestly, orange is fascinating how it can squeeze in besides yellow without anyone second guessing it. While the existence of a tag for cyan/teal gets a whole discussion.)

nystemy said:
Unlike so many other things when it comes to tagging. Color is fairly objective, just sample an area, take the hex code or RGB value and go look up what color it actually is. The color space of the image is good to take into consideration when doing this. And some lists are better than others, perhaps e621 should just make its own to be fair. For an example, cyan/teal is a 50/50 mix of green and blue.

For 8-bit color components, that's 16.7 million distinct colors. That would be very impractical for tagging, so we have to pair them down into groups, where they will be be arbitrarily separated. The issue isn't being able to measure to the colors, but how to distinguish them. Most people will still see #00FFFE as "cyan", even though it's not a pure 50/50 mix. #00FFFD is less of a 50/50 mix, which people will still see as cyan. But what about #00FFD0? #00FFA0? When will it stop being cyan and become green (or something else in between cyan and green)? Besides, few people will have a color picker prepared to precisely measure the color value, and will tag according to what it looks like to them (miscalibrated monitor and all). Also, brown is weird. There are colors that don't exist on the color spectrum like that, but are perceived as a result of other colors around it.

The issue here is that some people were a bit overzealous with pairing cyan with blue, when many other people feel it's distinct enough to be separate.

nystemy said:
Dividing the spectrum into: Red, Orange, Yellow, Green, Cyan, Blue, Purple, White, Grey and Black, is an adequate amount of distinction for searching for artwork.

What about tan, pink, and brown? Those are pretty prominent colors, too.

watsit said:
For 8-bit color components, that's 16.7 million distinct colors. That would be very impractical for tagging, so we have to pair them down into groups, where they will be be arbitrarily separated. The issue isn't being able to measure to the colors, but how to distinguish them. Most people will still see #00FFFE as "cyan", even though it's not a pure 50/50 mix. #00FFFD is less of a 50/50 mix, which people will still see as cyan. But what about #00FFD0? #00FFA0? When will it stop being cyan and become green (or something else in between cyan and green)? Besides, few people will have a color picker prepared to precisely measure the color value, and will tag according to what it looks like to them (miscalibrated monitor and all). Also, brown is weird. There are colors that don't exist on the color spectrum like that, but are perceived as a result of other colors around it.

(Partly I want to point out how stupid the "but what if it isn't EXACTLY 50/50?!" when the whole discussion is about what tag applies in what realm of the color space. Ie, as we approach a 50/50 mix of green and blue it becomes more applicable to call it Cyan and less applicable to call it Red, among other colors. Obviously there would be problems with making definitions for discrete colors.)

This is why I did say that e621 should probably have a guide. As to say between what ratios a certain color tag is applicable.
Secondly, I did answer the "or something else in between cyan and green", here we use both tags, since people can argue for either color in that case, and would search for either color to find said piece. Ie, both colors are applicable as far as searchability goes. (and this is always the case regardless of how finely we divide our colors. Just use both tags in these situations, it is better for everyone.)

In regards to using a color picker, one don't have to be that fancy the vast majority of the time, and the few times one has to, then we are back at both tags being applicable. (And for a color tagging guide, we just have the "applicable" ranges of color codes overlap a bit. Personally I think the HSV color format is the best approach for such a guide. But this is debatable in itself.)

watsit said:
The issue here is that some people were a bit overzealous with pairing cyan with blue, when many other people feel it's distinct enough to be separate.

What about tan, pink, and brown? Those are pretty prominent colors, too.

Yes, Cyan is indeed not blue. And there is already Teal that is the same chroma as Cyan.

In regards to brown, prink and tan. These are at times indeed useful colors. So yes, including them isn't a bad idea.
Even if brown is partly a darker orange, and tan is somewhere lighter towards the yellow side. And pink effectively being a lighter purple/red color. I won't say these tags "shouldn't exist". (at least it isn't "indego" that by color experts is anything from a deep blue to a pink... Depending on what standards organization or art institution one asks. Yet again, different discussion.)

But in the end, Cyan should be the same tag as Teal, not Blue. (if the tag should be called Cyan or Teal is a different debate.)

nystemy said:
Personally I think the HSV color format is the best approach for such a guide. But this is debatable in itself.)

I'd just go with HV, since it's a 2 dimensional representation of white-to-color, color-to-black, and chroma. Saturation will complicate things even further with no benefit.

Even still, there's not a lot you can do to officiate which tags go to which colors, assuming it ever needs to change down the line, let alone color perception (eg. from nearby colors)

watsit said:
But cyan is literally as much blue as it is green. Why put it under the umbrella of blue and not green? Especially for shades of cyan that lean a touch more on the green side (or the color palette used makes pure cyan appear more greenish). That'd be like having yellow implicate green.

FWIW, I'm fine with having cyan as its own color tag. But I don't think we need to build color hierarchies, given the number of color-related tags there are.

I don’t really care which way it goes - as long as cyan exists as a color tag, I’ll be fine with the result. However, this suggestion was a result of what Clawstripe pointed out above. Cyan, in common English usage, is often considered to be blue. Even the ROYGBIV rainbow color mnemonic refers to pure cyan as “blue” and refers to “true blue” instead as “indigo.” And many dictionaries refer to cyan, the color, as cyan blue. So yes, while cyan is as different from green and “true blue” (or indigo or whatever you prefer to call it) as yellow is from red and green, a common, non-technical, non-Color Theory definition of blue includes the entire range belonging to cyan. This is why, before cyan was aliased away, many cyan things were mistagged as blue. Someone pointed out that Rainbow Dash (MLP) typically has cyan fur, but it was almost always being tagged as blue.

So, the best solution I can think of for avoiding the inevitable mistagging of cyan as blue would be to consider both cyan and “true blue” (there must be a better name we can find for it, though - I don’t really like using indigo as it often refers to hues stretching far into purple territory) as both valid subtypes of “blue” and have both imply it. The other solution would be to just invalidate all the blue_* tags, which would be far more work to maintain given how common blue is. Or, thirdly, just attempt to clean up all the blue_* tags regularly to fix cyan mistags, but given that we have 340840 blue_eyes, 203912 blue_body, 114459 blue_fur, and 103046 blue_hair, I don’t think that’s even feasible at this point. Perhaps if the original tags had never been aliased to blue, it would have been manageable.

Or, lastly, we just deal with the fact that there will be mistags and hope that the use of well-written wiki pages can mitigate the number of people mistagging cyan as blue.

aversioncapacitor' said:
I'd just go with HV, since it's a 2 dimensional representation of white-to-color, color-to-black, and chroma. Saturation will complicate things even further with no benefit.

We’d have to get rid of tan, then, which is just a desaturated yellow/orange color.

lonelylupine said:
I thought cyan was one of those "emergent colors" that don't exist on a rainbow and only exist by mixing two wavelengths of light, like magenta.

My understanding is magenta is the ONLY color like that. Magenta is what happens when you mix all colors except for the middle of the rainbow (green). EDIT: Someone already explained this. I should probably read threads before responding...

nystemy said:
Color is fairly objective, just sample an area, take the hex code or RGB value and go look up what color it actually is.

That's not how it works exactly. You need to consider the lighting. If something is in a shadow, then it'll be darker, so an orange object may consist of brown pixels. Furthermore, if the light source is colored (or reflecting off of something colored?), it's going to make the pixels closer to that color, even though our brains know to subtract the light color to understand the intended color of the object.

clawstripe said:
*color table*

Okay am I alone in thinking that indigo looks more blue than blue does in this example? Blue looks more like light blue or sky blue.

pheagleadler said:
Okay am I alone in thinking that indigo looks more blue than blue does in this example? Blue looks more like light blue or sky blue.

By “light blue or sky blue,” you’re referring to cyan, which… was the entire point of this thread.

Basically:

scaliespe said:
Cyan, in common English usage, is often considered to be blue. Even the ROYGBIV rainbow color mnemonic refers to pure cyan as “blue” and refers to “true blue” instead as “indigo.”

I remember when we combined these tags to begin with. The main problem we were having, and will likely run into again, is that for most users cyan and blue are interchangeable. This has lead to the tags being intermixed constantly and nothing we have tried worked to keep new users from tagging them that way.

I am not against broadening our color palette (I am cyan, not blue!) but there are definitely still drawbacks that will come with it.

Also, howdy!

rainbow_dash said:
Also, howdy!

Hi, RD! Long time no see!

rainbow_dash said:
I remember when we combined these tags to begin with. The main problem we were having, and will likely run into again, is that for most users cyan and blue are interchangeable. This has lead to the tags being intermixed constantly and nothing we have tried worked to keep new users from tagging them that way.

I am not against broadening our color palette (I am cyan, not blue!) but there are definitely still drawbacks that will come with it.

Yeah, I encountered that while looking through the old threads on the subject. I currently feel like the best solution, as I suggested above, is to simply use blue as an umbrella term for both cyan and “true blue” or “indigo” or whatever term we want to use to describe the non-cyan blues.

_{Admittedly, it would be unprecedented to have a certain color implied by other colors, and it would also sort of imply that we should do the same with things like tan, as tan is just a shade of brown. However, the tan/brown distinction seems to be fairly functional as-is, and that’s getting outside the scope of this thread anyway.}

I think the benefit of this unusual structure would be the fact that we’d avoid mistags. People tagging cyan things as blue would not be incorrect, they’d just be missing a degree of specificity - like how people tag wolf instead of arctic wolf.

Anyway, I’d like to get your thoughts on that idea! I still haven’t written a BUR for this simply because I’m not entirely sure how to go about the whole thing. I think this solution would work, but I’m hesitant to suggest such a major change to how colors have been handled for so long now.

Although, what if we had color implications for all the colors? Like magenta implying purple and chartreuse implying green. It could be interesting. Imagine the precision of tagging. I mean, we already tag things like species with a very high degree of precision. But, anyway…

scaliespe said:
_{Admittedly, it would be unprecedented to have a certain color implied by other colors, and it would also sort of imply that we should do the same with things like tan, as tan is just a shade of brown.}

^{And brown is just how we interpret dimmer orange with brighter surroundings.}

scaliespe said:
I think the benefit of this unusual structure would be the fact that we’d avoid mistags. People tagging cyan things as blue would not be incorrect, they’d just be missing a degree of specificity - like how people tag wolf instead of arctic wolf.

This would go the other way too, that people who mean to tag "true blue" or "indigo" will just tag "blue", probably even moreso than cyan. This would effectively make "blue" a mishmash of cyan and indigo anyway, with many posts not bothering to have the more specific color tagged (and it'd increase the number of tags on a post that have blue things, as every cyan_* and indigo_* tag on a post would have a corresponding blue_* with it). At least if they're kept as completely separate tags, if someone tags blue_* for something that should be cyan_*, people will be more likely to replace an incorrect tag than to leave a "good enough" tag as it is (e.g. someone tags a red fox as 'fox', and that post will most likely stay tagged as just 'fox' forever, whereas if someone tags a bird as dinosaur, someone will be more likely to fix it up). It's just a question of if enough people will start using cyan so it doesn't overwhelm people who would otherwise like to fix it.

scaliespe said:
_{Admittedly, it would be unprecedented to have a certain color implied by other colors, and it would also sort of imply that we should do the same with things like tan, as tan is just a shade of brown. However, the tan/brown distinction seems to be fairly functional as-is, and that’s getting outside the scope of this thread anyway.}

I think the benefit of this unusual structure would be the fact that we’d avoid mistags. People tagging cyan things as blue would not be incorrect, they’d just be missing a degree of specificity - like how people tag wolf instead of arctic wolf.

Anyway, I’d like to get your thoughts on that idea! I still haven’t written a BUR for this simply because I’m not entirely sure how to go about the whole thing. I think this solution would work, but I’m hesitant to suggest such a major change to how colors have been handled for so long now.

I'm not sure about more color specificity beyond adding cyan, at least for now. we don't want to get ahead of ourselves, cyan on it's own is already going to be a massive tagging project that's probably going to take a veritable taskforce to sort out the backlog of posts. we do live in an e621 where we have enough dedicated taggers that we could probably muster that taskforce, but going beyond that is probably going to be a really tough sell.

watsit said:
^{And brown is just how we interpret dimmer orange with brighter surroundings.}

You could essentially reduce everything down to only three colors by describing them in relation to other colors if you wanted, but that doesn’t seem to be very useful. As far as brown is concerned, it may technically be dark orange, but people generally recognize it as a color that is specifically not orange.

This would go the other way too, that people who mean to tag "true blue" or "indigo" will just tag "blue", probably even moreso than cyan. This would effectively make "blue" a mishmash of cyan and indigo anyway, with many posts not bothering to have the more specific color tagged (and it'd increase the number of tags on a post that have blue things, as every cyan_* and indigo_* tag on a post would have a corresponding blue_* with it).

That’s the idea. People who care about that added level of precision can add the missing cyan or indigo* tag, whereas the blanket “blue” would probably suffice for the average user. As long as cyan and indigo* are reasonably well tagged by the more advanced users, the fact that there would undoubtedly still be a bunch of posts tagged only with blue isn’t a problem, even if it’s not ideal. Red_fox is still useful and valid even though there are probably a lot of red foxes just tagged fox.

_{*if we were to go this route, we shouldn’t call it “indigo,” as the name “indigo” is very often used for shades of blue ranging far into purple territory as opposed to the shade of pure blue represented by the name “indigo” in ROYGBIV.}

At least if they're kept as completely separate tags, if someone tags blue_* for something that should be cyan_*, people will be more likely to replace an incorrect tag than to leave a "good enough" tag as it is (e.g. someone tags a red fox as 'fox', and that post will most likely stay tagged as just 'fox' forever, whereas if someone tags a bird as dinosaur, someone will be more likely to fix it up). It's just a question of if enough people will start using cyan so it doesn't overwhelm people who would otherwise like to fix it.

I’m not sure it’s comparable. Specific species tagging requires a high degree of familiarity with those species beyond what I think most people have. I certainly can’t tell the difference between a red_fox and a fox of a different species that happens to be red. If I were to ever use a more specific fox species tag than the generic fox, I’d probably have got the intended species directly from the artist. In comparison, it takes very little research to learn the difference between blue and cyan, and I’d certainly add the cyan tags whenever I notice them missing, as opposed to a missing red_fox tag that I probably wouldn’t even notice.

For the record, though, I’d be perfectly fine with this solution too. I’d actually prefer this if we can manage to keep up with the inevitable mistagging of cyan as blue. It would get the cyan tags more populated than my other idea. I’m just not sure if we’ll be able to keep up with it. Also, I’m trying to address the initial reason for these tags being aliased in the first place, which was the fact that cyan was mistagged as blue too rapidly for anyone to keep up with at that time. Maybe things are different now, though. More active experienced users than before, perhaps. I don’t know.

If we end up not using blue_X as umbrella tags, then we ought to disambiguate it, requiring replacement with cyan_X or something like deep_blue_X, azure_X, or true_blue_X. It may be better to stick with the umbrella rather than the disambiguation. At least with a blue umbrella, cyan and deep blue/azure/true blue would be tagged with something, even if not as specifically as we might like.

scaliespe said:
_{*if we were to go this route, we shouldn’t call it “indigo,” as the name “indigo” is very often used for shades of blue ranging far into purple territory as opposed to the shade of pure blue represented by the name “indigo” in ROYGBIV.}

_{I hardly think this caveat is a problem. After all, under ROYGBIV,} ^cyan _{is called "blue". There's no reason why} ^{deep blue/azure/true blue} _{has to be "indigo".}

I don't really see why we'd need to create a new tag family for blue we already have a tag family for blue, it's blue_* changing that is only going to make it more confusing, and, more importantly, _annoying_ to tag stuff.
I don't even see why we're even discussing indigo, it's such an an antiquated term that no one thinks about or uses in modern times, the only reason it was ever a relevant word was because it's the name of a natural blue dye, and the only reason that it still sticks around at all is because of the ROYGBIV mnemonic, which people don't use to think about color anymore, if they ever really did.

clawstripe said:
If we end up not using blue_X as umbrella tags, then we ought to disambiguate it, requiring replacement with cyan_X or something like deep_blue_X, azure_X, or true_blue_X. It may be better to stick with the umbrella rather than the disambiguation. At least with a blue umbrella, cyan and deep blue/azure/true blue would be tagged with something, even if not as specifically as we might like.

This is another possible solution. It would require more maintenance, but would probably net the highest precision in the cyan/blue distinction, as it’ll force users to pick one of the two colors as opposed to leaving the generic blue. I am always hesitant to create more maintenance work, though.

_{I hardly think this caveat is a problem. After all, under ROYGBIV,} ^cyan _{is called "blue". There's no reason why} ^{deep blue/azure/true blue} _{has to be "indigo".}

darryus said:
I don't really see why we'd need to create a new tag family for blue we already have a tag family for blue, it's blue_* changing that is only going to make it more confusing, and, more importantly, _annoying_ to tag stuff.
I don't even see why we're even discussing indigo, it's such an an antiquated term that no one thinks about or uses in modern times, the only reason it was ever a relevant word was because it's the name of a natural blue dye, and the only reason that it still sticks around at all is because of the ROYGBIV mnemonic, which people don't use to think about color anymore, if they ever really did.

On both counts, yes, I don’t like to call it indigo. I’d really prefer to just call it “blue,” as color theory generally just calls it “blue” without any qualifier, leaving cyan as an entirely distinct color. That leaves us without the umbrella tag option, though, and we’re back to the very beginning with having to regularly clean out the blue_* tags to correct all the mistakes due to the fact that the average person thinks that cyan is just a lighter or more greenish shade of blue.

Still, I actually like the idea of establishing color hierarchies in general to increase precision in tagging, at least optionally. Most “specific” color tags have not been aliased away, and some have even seen a bit of use. Like, magenta_hair has 143 results, all of which seem to be that specific shade of purple that we generally recognize as magenta.* I don’t see why we couldn’t just have that tag imply purple_hair instead of just aliasing it, as it’s a more specific kind of purple. Keeping the tag allows people to find that more specific color if that’s what they want, and that seems useful without any drawbacks. With a color tag hierarchy, these more niche colors don’t have to exist in opposition to the broader colors - you’re not tagging magenta in place of purple; you’re tagging it in addition. This also wouldn’t add any additional burden on taggers as the more specific color tags aren’t required, and we don’t have to clean up all these tags that people are using anyway, nor would we have to sort all the cyan/blue mistags. The only extra work that would have to be done is establishing the correct implications. Just throwing this idea out there. It’s worth considering…

_{*technically, as magenta is a primary/secondary color and purple is not, purple would be considered a bluish shade of magenta. But I don’t think that’s important here. We can get away with calling magenta a shade of purple.}

scaliespe said:
Still, I actually like the idea of establishing color hierarchies in general to increase precision in tagging, at least optionally. Most “specific” color tags have not been aliased away, and some have even seen a bit of use. Like, magenta_hair has 143 results, all of which seem to be that specific shade of purple that we generally recognize as magenta.* I don’t see why we couldn’t just have that tag imply purple_hair instead of just aliasing it, as it’s a more specific kind of purple. Keeping the tag allows people to find that more specific color if that’s what they want, and that seems useful without any drawbacks. With a color tag hierarchy, these more niche colors don’t have to exist in opposition to the broader colors - you’re not tagging magenta in place of purple; you’re tagging it in addition. This also wouldn’t add any additional burden on taggers as the more specific color tags aren’t required, and we don’t have to clean up all these tags that people are using anyway, nor would we have to sort all the cyan/blue mistags. The only extra work that would have to be done is establishing the correct implications. Just throwing this idea out there. It’s worth considering…

I guess I can kind of see the utility in that, but I also don't really see how it's really that relevant to the cyan situation, cyan isn't a shade of any other color. cyan is a primary additive color, smack dab in the middle of green and blue. it's not any more a shade of blue than yellow is a shade of red.

also I don't really mind the idea of having to clean tags up every so often, we have to do it all the time with other tags anyway, it's just a few more to the pile. the really big project is adding cyan_* tags to the backlog of existing posts in the first place.

darryus said:
....the really big project is adding cyan_* tags to the backlog of existing posts in the first place.

Hmm. Compared to some tagging projects, this one seems pretty manageable, despite being among the more common end of the 8 traditional colours+cyan itself.

Edit, related to a post I don't feel like quoting: But isn't purple also technically one of those colours that just can't be perceived well with the 3 colour-cone eye, making it an in-between of blue and red, rather than yellow (thus it appearing distinct)?

scaliespe said:
Still, I actually like the idea of establishing color hierarchies in general to increase precision in tagging, at least optionally.

Given the number of things we have color tags for, this would result in an absurd amount of color tags on posts, and people are already complaining about the number of color tags on posts (to the point there's a desire for a new tag category specifically for colors to more easily hide them from the tag list). Turning color tags into multiple color tags via implications would make the problem significantly worse.

Additionally, with colors like cyan it's as much blue as it is green. So you can have a slightly more greenish shade that's still cyan, and/or an image's color palette can make cyan be perceived more green than blue. Having cyan imply blue would cause these more greenish hues to be incorrectly tagged blue instead of green. Same with magenta, which is as much blue as it is red. You can have some shades that are more purple/blue, and others that are more red, so implying the more reddish magenta shades to purple would be incorrect.

watsit said:
Given the number of things we have color tags for, this would result in an absurd amount of color tags on posts, and people are already complaining about the number of color tags on posts (to the point there's a desire for a new tag category specifically for colors to more easily hide them from the tag list). Turning color tags into multiple color tags via implications would make the problem significantly worse.

See, it's good because people won't abuse colour tags for tag count anymore. ;) /j

darryus said:
I guess I can kind of see the utility in that, but I also don't really see how it's really that relevant to the cyan situation, cyan isn't a shade of any other color. cyan is a primary additive color, smack dab in the middle of green and blue. it's not any more a shade of blue than yellow is a shade of red.

Technically, yes - but as stated previously, cyan is considered blue in common parlance, even though it isn’t actually blue scientifically. Trust me, I’m not too happy about it myself (hence why I titled this thread “Cyan is Not Blue”) but I’m trying to account for how the average user likely perceives these colors. If you were to show the average person the color cyan, or even a shade of cyan leaning towards green, most people would identify it as blue. And that’s the problem we’re faced with: getting the average user to tag cyan separately from blue, which is difficult if most people think cyan is just a lighter or greener shade of blue.

So, the idea was to call actual blue something else, such as indigo or true blue or something along those lines, and have both that color and cyan imply blue simply because that’s what most people think blue is. The alternative is to try to keep up with mistagging by all the people conflating cyan for blue. Which, again, I’m not opposed to - but it was the reason for aliasing cyan to blue in the first place, all those years ago, and we’d have to be sure we can keep up with it this time.

funeralopolite said:
But isn't purple also technically one of those colours that just can't be perceived well with the 3 colour-cone eye, making it an in-between of blue and red, rather than yellow (thus it appearing distinct)?

The human eye directly perceives red, blue, and green - which is why those are the three additive primary colors. If instead our cones were adapted to perceive yellow, cyan, and violet wavelengths, then those would be our primary colors. The entirety of color theory is based not on the actual wavelengths of visible light themselves (which actually have no inherent colors - only wavelength) but on how the human eye perceives these wavelengths.

Given that red and blue are two of our primary colors, we simply perceive purple as the combination of those two. It’s not really any different than our perception of yellow, being the combination of red and green. What you might have heard is that magenta isn’t directly perceptible as it’s not an actual wavelength of light. It would hypothetically be higher than violet on the visible light spectrum, but that lands it outside the visible light spectrum, so we simply can’t see it. Rather, we only perceive it as the combination of red light and blue light in equal measure, whereas something like yellow can be perceived directly as the wavelength halfway between green and red, or alternatively as the combination of green light and red light in equal measure.

watsit said:
Given the number of things we have color tags for, this would result in an absurd amount of color tags on posts, and people are already complaining about the number of color tags on posts (to the point there's a desire for a new tag category specifically for colors to more easily hide them from the tag list). Turning color tags into multiple color tags via implications would make the problem significantly worse.

If we need to limit the tags somehow (technically there’s no limit to how many tags can exist, AFAIK, but I understand the point), I’d be much more in favor of doing away with the overly specific items being tagged. Like blue_fingers and red_toenails and brown_lips and so on. I’d find these more precise color tags to be far more useful applied to just major parts, like magenta_fur.

Additionally, with colors like cyan it's as much blue as it is green. So you can have a slightly more greenish shade that's still cyan, and/or an image's color palette can make cyan be perceived more green than blue. Having cyan imply blue would cause these more greenish hues to be incorrectly tagged blue instead of green. Same with magenta, which is as much blue as it is red. You can have some shades that are more purple/blue, and others that are more red, so implying the more reddish magenta shades to purple would be incorrect.

Technically true, again, but I’m going by common usage. Most people consider cyan to just be a kind of blue. I mean, people use the term “green-blue” or “greenish blue” all the time to refer to what is actually greenish cyan. Turquoise, for example, is a precise greenish shade of cyan, but most people will call it blue. In fact, it can get a bit more green than that and still be called blue. In that sense, tagging a greenish shade of cyan “blue” would still agree with what most people think “blue” is.

Likewise, magenta technically covers everything between red and blue that isn’t one or the other, but in common parlance, most people just call that purple. What I’m proposing for a magenta tag is a precise shade such as the example given by Wikipedia, hex #FF00FF, or something close to that, while reserving “purple” for the entire range between red and blue. As I said earlier, yes, purple is technically a shade of magenta, but most people think of it the other way around.
If you look through the results for magenta_hair, you’ll see that it mostly coincides with what I’m proposing anyway. Most of those are pretty close to #FF00FF, with a few deeper purples mixed in that I’d just call mistags. But a really reddish shade of magenta that’s closer to red than purple simply wouldn’t be tagged as magenta at all - it’d be tagged as red, because I’m not actually using magenta in its technical, color theory context here, but a much more limited version of it.

In my opinion, we ought to have more color options. I swear I feel like we don't have enough color tags. Using base colors as an UMBRELLA is kinda dumb.
I'm not saying we ought to straight-up start using Crayola Crayon name colors or something like "Mucus green" or "electric blue"

But I feel like when I SEE a color and I KNOW the color, I want to TAG the color...so yea, cyan should be tagged cyan. Hot Pink should be tagged Hot Pink
and that's about it...eyup!

post #3055993

closetpossum said:
In my opinion, we ought to have more color options. I swear I feel like we don't have enough color tags. Using base colors as an UMBRELLA is kinda dumb.
I'm not saying we ought to straight-up start using Crayola Crayon name colors or something like "Mucus green" or "electric blue"

But I feel like when I SEE a color and I KNOW the color, I want to TAG the color...so yea, cyan should be tagged cyan. Hot Pink should be tagged Hot Pink
and that's about it...eyup!

post #3055993

Then why shouldn’t “hot pink” imply “pink?” It is a type of pink, after all, and “pink” as it stands currently already covers everything in the hot pink range. Hence, color hierarchies. If you want to search for pink, you get all the different pinks. If you just want a specific kind of pink, you use that one tag for it. Better than having all these separate color tags divided up so that you can’t just get anything that’s any kind of green, but rather having to search for “chartreuse” instead, even though it’s still green.

scaliespe said:
Then why shouldn’t “hot pink” imply “pink?” It is a type of pink, after all, and “pink” as it stands currently already covers everything in the hot pink range. Hence, color hierarchies. If you want to search for pink, you get all the different pinks. If you just want a specific kind of pink, you use that one tag for it. Better than having all these separate color tags divided up so that you can’t just get anything that’s any kind of green, but rather having to search for “chartreuse” instead, even though it’s still green.

hot pink is another tertiary color, it's hue is 330^o, putting it directly half-way between red and magenta. we already kinda have magenta split in half as it is with the lower half (~270-300^o) being purple and the upper half (~300-330^o) being pink, although "pink" also includes really low saturation reds as well and at higher saturation hot pink is already pretty close to being taggable as red, so it's a bit messy.

we probably have enough tags in the 270-30^o area already, honestly. we've got purple, pink, red, orange, tan, and brown in that one third of the hue circle. more than twice what we have in the other two thirds of the circle, yellow, green, blue.

The color wheel isn't a pizza; language is not a pizza cutter. We don't care if our definitions cut it in equal slices or not. Peach is pink. Mario is red. Waluigi is purple—everyone knows that. You'll get in trouble if you paint your face brown thinking it's tan—everyone knows that. We need those tags, doesn't matter if they're similar hues. Cyan is a shade of blue. Language is a bitch, but nobody's invented anything better. Blue jeans are always blue, no matter how light. Sky is blue, not cyan.

Could there be a cyan tag? Sure, but cyan could mean anything from light blue to jade green, especially without terms like aqua or turquoise. To make it even more ambiguous, most scenes have lighting, so you can't just check where pixels lie on a color wheel. Our perception of an object's color depends on context. It's subjective. Consistently tagging two very similar colors, based on a loose definition, in unknown lighting conditions—is impossible. If the line's so thin that taggers need to check the color balance of their monitor, it's too thin. And implementation? If cyan is not considered blue, users will miss art they were searching for. If cyan is considered blue, may as well just search blue.

I get it, OP. tl;dr, you watched a Tom Scott video, learned a basic fact about a culture other than your own, and thought it was pretty neat. You have to understand, you can't just rip something from another language and expect it to work. My character is between blue and purple, but he's called purple and that's fine. According to our language, everything approaching purple is purple. Lavender is to purple as pink is to red, but lavender is still considered purple—pink is not considered red. Don't worry about "fixing" this. The purpose of tagging is to help people find images, nothing more.

darryus said:
hot pink is another tertiary color, it's hue is 330^o, putting it directly half-way between red and magenta. we already kinda have magenta split in half as it is with the lower half (~270-300^o) being purple and the upper half (~300-330^o) being pink, although "pink" also includes really low saturation reds as well and at higher saturation hot pink is already pretty close to being taggable as red, so it's a bit messy.

we probably have enough tags in the 270-30^o area already, honestly. we've got purple, pink, red, orange, tan, and brown in that one third of the hue circle. more than twice what we have in the other two thirds of the circle, yellow, green, blue.

Yeah, I’m not too concerned about keeping hot pink, personally. I was just using the example given. I think we need cyan, I think magenta and perhaps a few others may be useful, but I don’t care to have tags for every exact color in existence. Just some of the major ones that we’re missing.

nothing_is_possible said:
The color wheel isn't a pizza; language is not a pizza cutter. We don't care if our definitions cut it in equal slices or not. Peach is pink. Mario is red. Waluigi is purple—everyone knows that. You'll get in trouble if you paint your face brown thinking it's tan—everyone knows that. We need those tags, doesn't matter if they're similar hues. Cyan is a shade of blue. Language is a bitch, but nobody's invented anything better. Blue jeans are always blue, no matter how light. Sky is blue, not cyan.

And that’s why I’m proposing an implication. Cyan implies blue so that the tagging agrees with language while not ignoring an entire primary/secondary color.

Could there be a cyan tag? Sure, but cyan could mean anything from light blue to jade green, especially without terms like aqua or turquoise. To make it even more ambiguous, most scenes have lighting, so you can't just check where pixels lie on a color wheel. Our perception of an object's color depends on context. It's subjective. Consistently tagging two very similar colors, based on a loose definition, in unknown lighting conditions—is impossible. If the line's so thin that taggers need to check the color balance of their monitor, it's too thin.

Which is also why I’m not proposing any incredibly precise color distinctions, such as between, for example, maroon and burgundy. Cyan, however, is a very broad color, and there will certainly be no problem telling it apart from “true” blue even with weird lighting and miscalibrated monitors - at least no more issue than we’d have with any other primary color. Cyan is as different from “true” blue as yellow is from red, and somehow we still have room for orange. We are generally more sensitive to the warm side of the color spectrum, and that’s fine, but it’s still very easy under any conditions to identify pure cyan. It can get a little vague around the green and blue edges, but that’s true of the edges of any color.

And implementation? If cyan is not considered blue, users will miss art they were searching for. If cyan is considered blue, may as well just search blue.

I’ll just point to my implication suggestion again. We could simply have it so that you could search for either.

I get it, OP. tl;dr, you watched a Tom Scott video,

For the record, I have no idea who Tom Scott is.

learned a basic fact about a culture other than your own, and thought it was pretty neat. You have to understand, you can't just rip something from another language and expect it to work.

How about ripping an idea from color theory?

My character is between blue and purple, but he's called purple and that's fine. According to our language, everything approaching purple is purple. Lavender is to purple as pink is to red, but lavender is still considered purple—pink is not considered red. Don't worry about "fixing" this. The purpose of tagging is to help people find images, nothing more.

Yes, and what if you want to find images with cyan? Heck, we could have lavender imply purple as well. This would be much more useful for searching, as far as I can tell. Wouldn’t it?

scaliespe said:
For the record, I have no idea who Tom Scott is.

dude, you're missing out.

scaliespe said:
And that’s why I’m proposing an implication. Cyan implies blue so that the tagging agrees with language while not ignoring an entire primary/secondary color.

Except that would break for more greenish shades of cyan, make the issue of cyan being undertagged worse than if not implicated (since as I mentioned before, people are more likely to leave a "good enough" tag alone, vs fixing an incorrect tag), and further bloat color tags on posts by ensuring every use of cyan has an additional blue tag.

scaliespe said:
I’ll just point to my implication suggestion again. We could simply have it so that you could search for either.

But that's the issue, "If cyan is considered blue, may as well just search blue". An implication wouldn't fix (and would in face exacerbate) the issue of cyan things being tagged as just blue, so either you search for cyan and miss many results since they were just tagged blue, or you search blue, and get all blues with cyan just like now.

watsit said:
Except that would break for more greenish shades of cyan, make the issue of cyan being undertagged worse than if not implicated (since as I mentioned before, people are more likely to leave a "good enough" tag alone, vs fixing an incorrect tag), and further bloat color tags on posts by ensuring every use of cyan has an additional blue tag.

If it’s green enough to be tagged green rather than blue, then it shouldn’t be tagged as cyan, only green - that way, blue isn’t added by implication. If it is cyan enough to not be tagged as green, then the blue implication would also be valid (if we regard cyan as a type of blue). As I pointed out earlier, people who don’t really acknowledge the existence of cyan will see a greenish cyan color and call it “greenish blue” even though it has no actual blue in it. Even the Wikipedia article for turquoise calls it “greenish-blue.”

But that's the issue, "If cyan is considered blue, may as well just search blue". An implication wouldn't fix (and would in face exacerbate) the issue of cyan things being tagged as just blue, so either you search for cyan and miss many results since they were just tagged blue, or you search blue, and get all blues with cyan just like now.

The idea is to have a color like indigo that identifies “true blue” and which also implies the generic “blue” tag. That way, if you want cyan, search cyan; if you want actual blue, search indigo (or whatever we call it); if you don’t understand or care about the distinction between actual blue and cyan (what I expect of the average user, to be honest), simply search blue.

You are right, though, in that cyan (and perhaps even worse with indigo) would often not be tagged, leaving generic “blue.” This, however, would still be preferable to the current situation where the two colors are lumped together in one tag. Even if a search for cyan_thing doesn’t return all valid results, it will at least return enough to be usable as a tag. And to be honest, that’s the best most tags can hope for. There are still plenty of cases of color tags not being applied at all, meaning there will be always be missing results from any color search. Example: hair -*_hair -monochrome. So many results for that that the site stopped counting pages.

All it comes down to, I think, is whether or not we can plausibly keep up with cyan being mistagged as blue. If we can, perhaps keeping blue and cyan separate would be preferable. I think I’d prefer that. But, again, they were aliased to blue in the first place because taggers 6 years ago couldn’t keep up with it. Perhaps we could start by reinstating them normally, and just see if we can keep up with the mistags. If, after a while, it becomes evident that users tagging cyan as blue overwhelms our ability to keep up with it, then we could make cyan imply blue and establish distinct indigo tags for true blue.

Besides that, the only way I can think to keep them separate would be to just invalidate all the blue tags so that users are forced to pick between cyan and true blue - but even then, I’m not certain we won’t just end up with a ton of invalid tags that we can’t keep up with.

I think generally colors shouldn't be put into hierarchies in tags. Firstly, because it's going to make tagging colors more complicated. Secondly, because what colors we categorize as the same is very dependent on language and culture. IMO, rather than basing the tags on the weird subtleties of the English language, we should base it on what we actually see and use.

Our eyes use RGB color, and so do computer displays. In RGB color, the three primary colors are Red, Green, and Blue. The secondary colors are Cyan, Magenta, and Yellow. The fact is, on a fundamental, mechanical, measurable level, cyan is a much more distinct color than pink or tan. (Pink/tan/brown are distinct enough as concepts, I don't think we need to remove them or anything, I just think that cyan is no less valid of a color.)

Traditionally, the word "blue" spans a wide range of hues from indigo to cyan. But in digital art, we talk about color knowing that we're using an RGB color space, so we don't tend to use "blue" in the traditional sense. We use it to refer to the "ultramarine" color that strongly activates our blue cones, and which we use as one of the primary colors in RGB displays.

I think the most straightforward solution at the moment is to allow cyan to be a valid color tag, instead of aliasing it to blue. Variations of cyan such as teal, aqua, or turquoise can be aliased to cyan. That would be most consistent with how color is used in digital art and graphics, which is especially relevant given that this is an art website.

scaliespe said:
If it’s green enough to be tagged green rather than blue, then it shouldn’t be tagged as cyan, only green

Why? If it's close to cyan, it should be tagged as cyan, even if it's on the greenish end a bit more than the blueish end. Again, true cyan is 50/50 blue/green, so as much as it's capable of looking blueish by having a little bit more blue than green, it's capable of looking greenish by having a little bit more green than blue. Add in the fact that the surrounding colors can further create a more greenish perception of what's otherwise clearly cyan, this would prevent certain cyan things from being tagged cyan, which is the very problem you want to fix by unaliasing it. It seems a little self-serving to get cyan unaliased from blue because you say it's too useful separate from blue, but then say certain instances of cyan can't use it because it doesn't fit your idea of a tag hierarchy.

scaliespe said:
As I pointed out earlier, people who don’t really acknowledge the existence of cyan will see a greenish cyan color and call it “greenish blue” even though it has no actual blue in it.

But people who do acknowledge the existence of cyan will see a greenish cyan color and call is "closer to cyan than green" (just as they'd see a blueish cyan color and call is "closer to cyan than blue").

scaliespe said:
The idea is to have a color like indigo that identifies “true blue” and which also implies the generic “blue” tag. That way, if you want cyan, search cyan; if you want actual blue, search indigo (or whatever we call it); if you don’t understand or care about the distinction between actual blue and cyan (what I expect of the average user, to be honest), simply search blue.

Taggers also play a role here too. It's all well and good to say "if you want cyan, search cyan", but if the tagger simply tagged it "blue" (because that shade of cyan looked blueish, and no one bothered to add the more precise cyan tag because "blue" was good enough), that search will fail. So if you really want to find cyan things, you need to search blue anyway and get all the blues next to cyan, making it no better than now (if not worse actually, since some shades of cyan will be more green, so searching blue will get you some cyan posts that look green, which can't be fixed because it is cyan and the cyan tag would be correct).

scaliespe said:
All it comes down to, I think, is whether or not we can plausibly keep up with cyan being mistagged as blue. If we can, perhaps keeping blue and cyan separate would be preferable. I think I’d prefer that. But, again, they were aliased to blue in the first place because taggers 6 years ago couldn’t keep up with it. Perhaps we could start by reinstating them normally, and just see if we can keep up with the mistags. If, after a while, it becomes evident that users tagging cyan as blue overwhelms our ability to keep up with it, then we could make cyan imply blue and establish distinct indigo tags for true blue.

If they're kept as separate tags and enough people are able to keep cyan separate, then it's all good. But if you can't, if too many cyans end up being tagged blue (or green) to make it not useful to search, then an implication should not be used because that will create a worse problem; the alias should be reinstated.

watsit said:
Taggers also play a role here too. It's all well and good to say "if you want cyan, search cyan", but if the tagger simply tagged it "blue" (because that shade of cyan looked blueish, and no one bothered to add the more precise cyan tag because "blue" was good enough), that search will fail. So if you really want to find cyan things, you need to search blue anyway and get all the blues next to cyan, making it no better than now (if not worse actually, since some shades of cyan will be more green, so searching blue will get you some cyan posts that look green, which can't be fixed because it is cyan and the cyan tag would be correct).

I just realized that even if cyan doesn't implicate blue, we may need an exception for blue_eyes. We almost always call cyan eyes "blue", so there could be a special case for eye color. Maybe "deep blue eyes" and "cyan eyes" could be used, and both could fall under "blue eyes". I don't think this would be needed for other attributes, just eyes. Kind of like how we describe hair colors in special ways, like "blonde" for yellow, and "red" and "auburn" for red-orange and reddish brown.

watsit said:
Why? If it's close to cyan, it should be tagged as cyan, even if it's on the greenish end a bit more than the blueish end. Again, true cyan is 50/50 blue/green, so as much as it's capable of looking blueish by having a little bit more blue than green, it's capable of looking greenish by having a little bit more green than blue. Add in the fact that the surrounding colors can further create a more greenish perception of what's otherwise clearly cyan, this would prevent certain cyan things from being tagged cyan, which is the very problem you want to fix by unaliasing it. It seems a little self-serving to get cyan unaliased from blue because you say it's too useful separate from blue, but then say certain instances of cyan can't use it because it doesn't fit your idea of a tag hierarchy.

Again, this is true, but only if you regard cyan as being separate from blue. Which I do, but which the average English speaker probably does not, and which the site currently does not. I’m not suggesting that greenish cyan shouldn’t be tagged as cyan - only that, if we regard cyan as simply a type of blue and that we structure our tagging system to reflect that, then even greenish shades of cyan are blue. In fact, this is how it’s being done currently. Whenever I see a greenish cyan color, such as a shade like turquoise, it’s always tagged as blue, even though green would be technically more accurate. That’s just how the blue tags are used, since blue currently includes the entirety of cyan due to the current aliases.

Taggers also play a role here too. It's all well and good to say "if you want cyan, search cyan", but if the tagger simply tagged it "blue" (because that shade of cyan looked blueish, and no one bothered to add the more precise cyan tag because "blue" was good enough), that search will fail. So if you really want to find cyan things, you need to search blue anyway and get all the blues next to cyan, making it no better than now (if not worse actually, since some shades of cyan will be more green, so searching blue will get you some cyan posts that look green, which can't be fixed because it is cyan and the cyan tag would be correct).

Unless you regard cyan as blue, in which case those greenish cyan colors are blue. But again, this is no different from how things are currently being done. You can find plenty of greenish cyan colors under the blue_* tags as-is - post #3399496, for example. All I’d be doing is changing it from an alias to an implication, but blue_* would continue to contain all those greenish cyan colors as it currently does. The difference is that you’d actually be able to find some cyan things, even if the tagging of that color wouldn’t be comprehensive. Perhaps it wouldn’t be ideal, but it certainly wouldn’t be worse in any way than what we have now.

If they're kept as separate tags and enough people are able to keep cyan separate, then it's all good. But if you can't, if too many cyans end up being tagged blue (or green) to make it not useful to search, then an implication should not be used because that will create a worse problem; the alias should be reinstated.

In what way would it be worse than what we have currently? Greenish cyan is already being tagged as blue for the most part. Again, post #3399496 for a recently posted example. This is far closer to green than it is to true blue, but it is definitely still cyan, which means it’s blue if you regard cyan as blue. Leaving it as an alias doesn’t solve the problem you’re seeing with the implication idea. Therefore, simply leaving it as an alias would be the worst possible option, and re-aliasing it after putting in a bunch of work to separate the two would be worse still.

I say we unalias cyan, try to keep it separate from blue - and if it eventually becomes clear that we’ve failed in keeping them separate, only then have it imply blue - as a way of salvaging the work put in to separate them, if nothing else. That would still leave cyan as a taggable and searchable subset of blue, as opposed to an alias which will continue to see green-cyan being tagged as blue regardless.

hurr_durr said:
I just realized that even if cyan doesn't implicate blue, we may need an exception for blue_eyes. We almost always call cyan eyes "blue", so there could be a special case for eye color. Maybe "deep blue eyes" and "cyan eyes" could be used, and both could fall under "blue eyes". I don't think this would be needed for other attributes, just eyes. Kind of like how we describe hair colors in special ways, like "blonde" for yellow, and "red" and "auburn" for red-orange and reddish brown.

I mean, this isn’t actually unique to eyes. Cyan in general is considered blue. It’s just that true blue eyes are very rare whereas cyan eyes are fairly common, so the fact of cyan being called blue is more prominent in that case. But if cyan was a natural and common hair color, people would also call it blue.

hurr_durr said:
I think generally colors shouldn't be put into hierarchies in tags. Firstly, because it's going to make tagging colors more complicated. Secondly, because what colors we categorize as the same is very dependent on language and culture. IMO, rather than basing the tags on the weird subtleties of the English language, we should base it on what we actually see and use.

Our eyes use RGB color, and so do computer displays. In RGB color, the three primary colors are Red, Green, and Blue. The secondary colors are Cyan, Magenta, and Yellow. The fact is, on a fundamental, mechanical, measurable level, cyan is a much more distinct color than pink or tan. (Pink/tan/brown are distinct enough as concepts, I don't think we need to remove them or anything, I just think that cyan is no less valid of a color.)

Traditionally, the word "blue" spans a wide range of hues from indigo to cyan. But in digital art, we talk about color knowing that we're using an RGB color space, so we don't tend to use "blue" in the traditional sense. We use it to refer to the "ultramarine" color that strongly activates our blue cones, and which we use as one of the primary colors in RGB displays.

I think the most straightforward solution at the moment is to allow cyan to be a valid color tag, instead of aliasing it to blue. Variations of cyan such as teal, aqua, or turquoise can be aliased to cyan. That would be most consistent with how color is used in digital art and graphics, which is especially relevant given that this is an art website.

This, ultimately, is what I think and what I would most agree with. However, I’ll refer you to my above suggestion that a cyan -> blue implication be created only if we can’t keep the two colors separate - or if the average tagger consistently fails to make the distinction often enough that searching for blue gives results full of cyan despite our best efforts.

I don’t, however, think a color hierarchy would make anything complicated. The site is already full of tag hierarchies of all kinds. X is a type of Y, so X implies Y. It’s not a difficult concept at all. I still think it would be useful for things like, say, lavender. Lavender is a type of purple, and everyone knows that. But it’s still recognizable on its own. X is a type of Y. I think that could be a useful idea, even if we do continue to regard cyan as not-blue and have no cyan -> blue implication (really, I’d prefer that if we can actually make it work). That’s rather outside the scope of this thread, however.

scaliespe said:
Again, this is true, but only if you regard cyan as being separate from blue. Which I do, but which the average English speaker probably does not, and which the site currently does not. I’m not suggesting that greenish cyan shouldn’t be tagged as cyan - only that, if we regard cyan as simply a type of blue and that we structure our tagging system to reflect that, then even greenish shades of cyan are blue.

That sounds ludicrous. If it looks green, it shouldn't be tagged blue. Having greenish shades of cyan tagged blue because "cyan is blue" would be like having greenish yellow tagged as red. Cyan is as much blue as yellow is red, so if greenish shades of cyan should be tagged blue, greenish shades of yellow should be tagged red with that same logic. That makes no sense to do. If it looks greenish, it shouldn't be tagged blue (or red) just because it has a bit of it to make it appear more cyan (or yellow).

scaliespe said:
In fact, this is how it’s being done currently. Whenever I see a greenish cyan color, such as a shade like turquoise, it’s always tagged as blue, even though green would be technically more accurate.

If something looks green but is tagged blue, the blue tag should be replaced with green. If a greenish cyan is tagged blue, you would replace it with the cyan tag, which would incorrectly tag the greenish shade as blue, and that can't be fixed without removing the valid cyan tag.

scaliespe said:
In what way would it be worse than what we have currently?

Because things that look more green than blue will be tagged blue which can't be fixed. So if you then search for blue wanting blue, you'll get more greenish results because of cyan. If you search for blue now and find green things, those are incorrect tags that can be fixed. However with an implication of cyan to blue, if you search blue and find some greenish cyan things, that can only be fixed by using green instead of cyan despite it being cyan, making cyan a pointless tag to have if it can't be used on cyan things.

scaliespe said:
Greenish cyan is already being tagged as blue for the most part. Again, post #3399496 for a recently posted example.

Those are incorrect and should be tagged green since they're more green than blue, which is easy to fix. They're fixed now, but if you had cyan implicating blue and they were tagged cyan, that would require leaving them with the incorrect blue tags or removing the otherwise-valid cyan tags to be able to remove blue.

scaliespe said:
I say we unalias cyan, try to keep it separate from blue - and if it eventually becomes clear that we’ve failed in keeping them separate, only then have it imply blue - as a way of salvaging the work put in to separate them, if nothing else. That would still leave cyan as a taggable and searchable subset of blue, as opposed to an alias which will continue to see green-cyan being tagged as blue regardless.

It wouldn't be very searchable if you can't be sure you'll find cyan things under cyan, either because someone tagged it as blue instead, or because cyan couldn't be used due to it more greenish. There wouldn't be much worth salvaging, especially if it meant forcing some green things to be incorrectly tagged blue.

watsit said:
That sounds ludicrous. If it looks green, it shouldn't be tagged blue. Having greenish shades of cyan tagged blue because "cyan is blue" would be like having greenish yellow tagged as red. Cyan is as much blue as yellow is red, so if greenish shades of cyan should be tagged blue, greenish shades of yellow should be tagged red with that same logic. That makes no sense to do. If it looks greenish, it shouldn't be tagged blue (or red) just because it has a bit of it to make it appear more cyan (or yellow).

If something looks green but is tagged blue, the blue tag should be replaced with green. If a greenish cyan is tagged blue, you would replace it with the cyan tag, which would incorrectly tag the greenish shade as blue, and that can't be fixed without removing the valid cyan tag.

Because things that look more green than blue will be tagged blue which can't be fixed. So if you then search for blue wanting blue, you'll get more greenish results because of cyan. If you search for blue now and find green things, those are incorrect tags that can be fixed. However with an implication of cyan to blue, if you search blue and find some greenish cyan things, that can only be fixed by using green instead of cyan despite it being cyan, making cyan a pointless tag to have if it can't be used on cyan things.

You’re still assuming that cyan is not blue. I understand that it’s technically not blue, but it is blue in the layman’s concept of color. The logic is as follows:

Cyan is blue. Therefore, even greenish shades of cyan are blue. Just the same as yellowish-green is green. Your red-yellow comparison is perfectly valid in a scientific or color theory sense, but that simply isn’t how most people in the English-speaking world see those colors.

Example: despite being a tertiary color in reality, most people regard orange as a primary color. There is no such thing as “reddish-yellow,” there is only orange. There can be reddish orange or yellowish orange, but no reddish yellow. By comparison, cyan is secondary in this non-technical definition of color. There is blue of various shades, which includes cyan, and there is green. That’s why greenish-cyan and bluish-cyan do not exist in this view - there is only greenish-blue or bluish-green. If cyan is blue, then greenish-cyan is actually greenish-blue. Even Wikipedia considers it this way (they should know better, but that’s beside the point) - the article for turquoise, which is a greenish-cyan color, is called “greenish blue” in the article.

Note that also in this view is the fact that “true blue” or RGB blue is not the “center” of blue in the sense that it is in a computer display. Since cyan is blue in the layman’s sense, the center point of blue would actually be about halfway between true blue and cyan (called “azure” by Wikipedia), while true blue would be called something like navy blue or ultramarine, seen as a darker shade of blue.

The crux of the matter is that we’re left to deal with this technically incorrect concept of color. Most of this site’s users are probably two things: they speak English as a primary language (or their primary language is one that similarly does not distinguish between blue and cyan - as far as I know, only Russian and Italian hold that distinction); and, they know little to nothing about color theory (it’s not generally taught in primary education, and the average person has no reason to learn it). Therefore, most users will not regard cyan as a distinct color from blue. Which means most users will see that greenish cyan color in post #3399496 and call it blue. This is going to happen regardless of anything we do, due to the problem of the layman’s idea of color that I explained above. Even if cyan is a separate tag from blue, I assume most people will continue to tag posts such as #3399496 as blue, because that’s what most people think blue is.

Those are incorrect and should be tagged green since they're more green than blue, which is easy to fix. They're fixed now, but if you had cyan implicating blue and they were tagged cyan, that would require leaving them with the incorrect blue tags or removing the otherwise-valid cyan tags to be able to remove blue.

Actually, I don’t think that post was mistagged at all. Cyan is currently aliased to blue, meaning that E621 currently considers cyan to be equivalent to blue for tagging purposes. That means that a color between cyan and green that is closer to cyan than green is also closer to blue than green, since cyan = blue. I would tag that as blue myself, if I was the one uploading it. In fact, I’d not be surprised if someone ends up changing it back to blue. Most people will probably look at that and say that it’s not green. And in fact, it isn’t green, but cyan - the point of this thread, after all - but in the absence of a cyan tag, we have to regard it as blue.

It wouldn't be very searchable if you can't be sure you'll find cyan things under cyan, either because someone tagged it as blue instead, or because cyan couldn't be used due to it more greenish. There wouldn't be much worth salvaging, especially if it meant forcing some green things to be incorrectly tagged blue.

Nothing would be incorrectly tagged as blue if cyan is considered blue. Either it’s closer to green than cyan, in which case it’ll simply be tagged as green; or, it’s closer to cyan than green, therefore tagged as blue with cyan being considered blue. My point is that it couldn’t possibly be any worse than the current situation because this is already how it’s being done. Look through the blue_* tags for a bit and you’ll see what I mean. Greenish cyan colors are consistently being tagged as blue despite the fact that they’re technically closer to green than “true blue;” but again, this is the layman’s idea of color that we’re rubbing up against - it simply does not agree with how these colors technically work, and that’s the problem we’re facing with these tags. Therefore, since greenish cyan is already being tagged as blue, having it as an implication wouldn’t cause blue to be mistagged any more than it is now. It would be the same thing, but with the benefit that we’d actually be able to find some things that are cyan, rather than all the shades of blue and cyan, right up to the very boundary of green, being lumped together under the blue tags.

A few more recent examples of greenish-cyan being tagged as blue:

post #3406517
post #3406501
post #3406445
post #3406387
post #3406377
post #3406320
post #3406336
post #3406352
post #3406020
post #3406066
post #3406003
post #3405965

All uploaded in the last 6 hours from the time of posting. But looking through green_*, I see very few things that can be reasonably called cyan instead.

…regardless, I assume that we’re all on board with removing the cyan -> blue aliases for now. At this point we’re just arguing about the hypothetical future situation that cyan/blue mistagging gets out of control after the tags have been re-established, which is probably years away, if we ever actually come to that point (because, who knows - maybe we actually will be able to keep cyan and blue separate this time around).

scaliespe said:
You’re still assuming that cyan is not blue. I understand that it’s technically not blue, but it is blue in the layman’s concept of color.

We're not talking about layman concepts of color, we're talking about what it looks like. If it looks like a greenish cyan, it doesn't look blue, and if it doesn't look blue, blue should not be tagged on such posts.

scaliespe said:
Actually, I don’t think that post was mistagged at all. Cyan is currently aliased to blue, meaning that E621 currently considers cyan to be equivalent to blue for tagging purposes.

No, it means that e621 didn't want to have a distinct cyan tag, and most people using cyan would likely mean the more blue-looking shades. Unlike implications, an alias is allowed to result in a manageable number of mistags since they can be fixed after the fact. The only other alternative for getting rid of a tag would be to alias it to invalid_tag, which is something to avoid when possible. It doesn't mean cyan is always equivalent to blue for tagging purposes.

scaliespe said:
And in fact, it isn’t green, but cyan - the point of this thread, after all - but in the absence of a cyan tag, we have to regard it as blue.

In the absence of the cyan tag, we have to use a different valid tag that's closest. Most of the time that should hopefully be blue, but in that particular case, since it's closer to green than blue and we can't use cyan, green is the more appropriate alternative instead of blue there.

scaliespe said:
Nothing would be incorrectly tagged as blue if cyan is considered blue.

But cyan isn't always considered blue. Common uses of it may appear more blueish, such that aliasing it to blue results in fewer mistags than aliasing it to green would, but that doesn't mean all uses of it always look more blue instead of green.

scaliespe said:
My point is that it couldn’t possibly be any worse than the current situation because this is already how it’s being done. Look through the blue_* tags for a bit and you’ll see what I mean.

Primary difference being, any instances of greenish cyan being tagged blue because of the alias can be fixed. The blue_* tags could and should be replaced with green_* when they look more green than blue. That cannot be done with an implication, without requiring users to remove the valid cyan tags, meaning it can only be used on the blueish half of the spectrum or we have to live with mistags, which would certainly be worse than the current situation.

scaliespe said:
A few more recent examples of greenish-cyan being tagged as blue:

Some of those do come across as more blueish to me. But some are just straight-up green, I wouldn't even consider tagging ones like these cyan:
post #3406501 post #3406387

watsit said:
We're not talking about layman concepts of color, we're talking about what it looks like. If it looks like a greenish cyan, it doesn't look blue, and if it doesn't look blue, blue should not be tagged on such posts.

Unfortunately, we can’t simply disregard the layman’s concept of color because that’s what most users are going to abide by. What I’m proposing with the cyan -> blue implication is an adherence to the layman’s concept of blue in the event that we find ourselves unable to go against it. In that notion of blue, all shades of cyan are blue, so no use of the cyan tag would imply blue incorrectly.

No, it means that e621 didn't want to have a distinct cyan tag, and most people using cyan would likely mean the more blue-looking shades. Unlike implications, an alias is allowed to result in a manageable number of mistags since they can be fixed after the fact. The only other alternative for getting rid of a tag would be to alias it to invalid_tag, which is something to avoid when possible. It doesn't mean cyan is always equivalent to blue for tagging purposes.

Perhaps not in theory, but it is in effect. The layman’s idea of cyan as blue currently rules the usage of the blue_* tags as far as I can tell. You can say greenish cyan is not blue all you want, but people will continue to tag it as blue regardless.

Primary difference being, any instances of greenish cyan being tagged blue because of the alias can be fixed. The blue_* tags could and should be replaced with green_* when they look more green than blue. That cannot be done with an implication, without requiring users to remove the valid cyan tags, meaning it can only be used on the blueish half of the spectrum or we have to live with mistags, which would certainly be worse than the current situation.

What you’re suggesting is that the greener shades of cyan should be tagged as green instead - but as I said in my last post (I edited it after posting, you may have missed it - sorry), I found 12 posts with greenish cyan being tagged as blue uploaded just within the last 6 hours. In the same time period, I found only one, maybe two greenish cyans tagged as green instead. If you want to say that only the bluer half of the cyan range should be tagged as blue, you’re going to be fighting an uphill battle against everyone abiding by the common conception of blue, which includes all cyan. At that point, it would just be easier to have separate cyan tags, as I fear that even trying to fix those “mistags” of greenish cyan as blue is going to be met with resistance, people looking at those colors and saying “that’s not green” and changing it back to blue.

watsit said:
Some of those do come across as more blueish to me. But some are just straight-up green, I wouldn't even consider tagging ones like these cyan:
post #3406501 post #3406387

the one on the left is extremely cyan, average hue of the hair and fur are 178º and the background is literally #00FFFF. the one on the left is a bit closer to green, 164º is on the borderline but it still seems very cyan to me.

scaliespe said:
Unfortunately, we can’t simply disregard the layman’s concept of color because that’s what most users are going to abide by.

What people are going to abide by is, "does it look blue?" or "does it look green?". They're not going to get into color theory and human perception to consider what it technically is. They're just going to look at it and use what it looks most like to them, and that will mean, sometimes cyan things will look more green than blue.

scaliespe said:
What you’re suggesting is that the greener shades of cyan should be tagged as green instead - but as I said in my last post (I edited it after posting, you may have missed it - sorry), I found 12 posts with greenish cyan being tagged as blue uploaded just within the last 6 hours.

Mistags happen, yeah. And in my own edit, I also remarked that I would agree some of them may be more blue (or close enough that I can call it a legitimate question that can go either way), and some were so green I wouldn't have even considered cyan to be correct, which just goes to show how far into green some people consider cyan to stretch. Yet colors that are so obviously on the green end of the spectrum, you would want tagged blue because it's like 15% blue and just barely close enough to cyan for some people.

scaliespe said:
If you want to say that only the bluer half of the cyan range should be tagged as blue, you’re going to be fighting an uphill battle against everyone abiding by the common conception of blue, which includes all cyan.

Is it people "abiding by the common conception of blue", or is it people tagging cyan when they see cyan and not noticing or caring that it got replaced with blue?

darryus said:
the one on the left is extremely cyan, average hue of the hair and fur are 178º and the background is literally #00FFFF. the one on the left is a bit closer to green, 164º is on the borderline but it still seems very cyan to me.

Which just goes to show how greenishly cyan can be perceived. If they were just tagged cyan I might've left it alone, but blue isn't even in the cards to me (and that's not me as some color expert using some eyedrop tool to get a precise reading, I'm just looking at it and my brain is going "green").

why are we even arguing about this at this point? this is all just pointless hypotheticals, we don't have any data on how people would tag cyan_* on stuff, or how it'd effect searching or anything because most of the uses are currently aliased away to the adjacent colors. until cyan tags are actually added properly (if they ever are) we can't possibly know what the outcome is going to be, and when (or if) that data actually starts coming in we can discuss it then. at this point we're just going in circles arguing about nothing.

from where I'm sitting the best thing would be cyan is given the same tags as all of the existing colors, and is treated more or less identical as the existing color tags. if everything works out fine which it probably would we go from there, and if it turns out terribly we'll deal with it. but right now we're not even talking about cyan anymore we're arguing about huge site-wide overhauls about how tagging colors works. we should be taking things one step at a time.

watsit said:
What people are going to abide by is, "does it look blue?" or "does it look green?". They're not going to get into color theory and human perception to consider what it technically is. They're just going to look at it and use what it looks most like to them, and that will mean, sometimes cyan things will look more green than blue.

Mistags happen, yeah. And in my own edit, I also remarked that I would agree some of them may be more blue (or close enough that I can call it a legitimate question that can go either way), and some were so green I wouldn't have even considered cyan to be correct, which just goes to show how far into green some people consider cyan to stretch. Yet colors that are so obviously on the green end of the spectrum, you would want tagged blue because it's like 15% blue and just barely close enough to cyan for some people.

There’s a weird thing about “mistags” when it comes to color - as colors are somewhat nebulously defined and our perception of them is affected by shading and adjacent colors, I’m not sure these can really be considered mistags. If we had separate cyan tags, like we once did, then tagging something that’s clearly very cyan as blue would be a mistag - a mistag caused by the user’s conflation of cyan with blue. They don’t think it’s not cyan, they simply don’t really know what cyan is, or they don’t know of the existence of cyan tags. But in cases like these where the visible color is between two other colors - in this case, green and cyan - it’s probably more accurate to call it a difference in opinion rather than an outright mistag. Like determining the difference between huge_* and hyper_*, there’s a definite gray area and different people are going to give different answers. Just the fact that you can look through the first page of blue_* at any given time and likely find a few greenish cyans in there goes to show how far into green territory can be considered “blue” by some people, and how common that is. I personally would’ve tagged at least one or two of my examples as green instead of blue or cyan, but I think they could all at least arguably be considered blue due to how much cyan is visible in them, and how broadly the term “blue” is usually applied. Of course, we’ll still have to deal with this when cyan_* is unaliased, but it’ll hopefully reduce the size of that gray area between blue and green by allowing most of them to be thrown under cyan instead.

Is it people "abiding by the common conception of blue", or is it people tagging cyan when they see cyan and not noticing or caring that it got replaced with blue?

Probably some of both, to be honest.

Which just goes to show how greenishly cyan can be perceived. If they were just tagged cyan I might've left it alone, but blue isn't even in the cards to me (and that's not me as some color expert using some eyedrop tool to get a precise reading, I'm just looking at it and my brain is going "green").

I’m just going to agree and point to how very green some of the shades of cyan can look on this Wikipedia article.

darryus said:
why are we even arguing about this at this point? this is all just pointless hypotheticals, we don't have any data on how people would tag cyan_* on stuff, or how it'd effect searching or anything because most of the uses are currently aliased away to the adjacent colors. until cyan tags are actually added properly (if they ever are) we can't possibly know what the outcome is going to be, and when (or if) that data actually starts coming in we can discuss it then. at this point we're just going in circles arguing about nothing.

from where I'm sitting the best thing would be cyan is given the same tags as all of the existing colors, and is treated more or less identical as the existing color tags. if everything works out fine which it probably would we go from there, and if it turns out terribly we'll deal with it. but right now we're not even talking about cyan anymore we're arguing about huge site-wide overhauls about how tagging colors works. we should be taking things one step at a time.

Well, yeah. I suppose I’ll start with a BUR to unalias them when I get the chance, since I think we can all agree on that.

I love that this discussion is being had. One of my tagging fundaments is the list of colors on the tag group:colors page. I would personally rather have cyan instead of teal and beige instead of tan, but it's still okay as it is, except for certain annoying aliasings of teal to green or blue. I often seen something tagged green that for me actually looks so much more like blue that I'd rather tag it blue than even teal/cyan.

Color sight and psychology is interesting. We also have people with color blindness, which would suggest that even physiologically we see things quite differently often.

Tan is a big challenging colour for me. I'm used to think of the caucasian skin color to be "pink", but here the rule is tan+light. I can understand sun-tanned caucasians, as well as tanned leather... There's variation. But it is there somewhere between pink, brown and yellow. Brown being somewhere between red and yellow and pink being light/pale red.

watsit said:
Taggers also play a role here too. It's all well and good to say "if you want cyan, search cyan", but if the tagger simply tagged it "blue" (because that shade of cyan looked blueish, and no one bothered to add the more precise cyan tag because "blue" was good enough), that search will fail. So if you really want to find cyan things, you need to search blue anyway and get all the blues next to cyan, making it no better than now (if not worse actually, since some shades of cyan will be more green, so searching blue will get you some cyan posts that look green, which can't be fixed because it is cyan and the cyan tag would be correct).

As said, people see things differently, both physiologically and psychologically. On the physiological part there are good examples the different assortions of colorblindness. On psychological part we have the famous dress that appeared as golden for some and blue for others. When I tag pictures that have a generally tinted colour palette – I guess one could say pictures with blue_theme for example, and some part of a character i obviously light blue there, it is difficult to say, if it should be tagged as white or blue. The brain says that the character's body there is white, but the lighting in the scene gives it a blue tint. The eyes and color pipette tool says it is blue.

Nevertheless, this is mostly a different issue, as regardless of how many colors we have in the official palette, there will always be these borderline cases. The valid question here is, how many, and which colors we should have in our palette, so that there would be enough distinction, but not too many to make searching next to impossible and also increase the amount of disputes between which colours do individual cases fall.

My opinion would be: black, blue, brown, green, grey, orange, pink, purple/violet/lilac, red, tan/beige, teal/cyan, white, yellow

(The purple/violet/lilac is an interesting case, as it is a color that doesn't actually exist. In the color spectrum red is in one end and blue in the other opposite. They don't loop back, so there is no color between red and blue. Or alternatively, ALL of the other colors are between red and blue.)

A blue disambiguation would be one hell of a job. Most of them are going to be indigo/violet, right? And rest cyan/turquoise, right? Too bad we can't have a color wheel come up when tagging a color tag.

Not to mention magenta, which is either pink, purple, or even a color at all by being a combination of red and blue light.