Posted under Tag Alias and Implication Suggestions
The tag implication #51808 flat_colors -> colored has been rejected.
Reason: If there are flat colors, there are colors in the post, therefore the post should be tagged as colored. However, there are many posts tagged as flat colored, but not colored, see flat_colors -colored
Overall self-explanatory
EDIT: The tag implication flat_colors -> colored (forum #376021) has been rejected by @slyroon.
Updated by auto moderator
watsit said:
While this is fine for how most people seem to think colored should be used, there has been some discussion on invalidating colored since the vast majority of images are colored, making it largely unnecessary.
colored is just -greyscale -black_and_white right?
And given that colored is severely undertagged, it's not really useful
snpthecat said:
colored is just -greyscale -black_and_white right?And given that colored is severely undertagged, it's not really useful
Which would be fixed if more images were properly tagged, yes.
I personally dislike non-colored explicit images, but can’t do much because colored isn’t used enough. I want to fix that, and I think this alias is a good, surefire way to do it.
I want to add the colored tag to multiple posts at once but I do not know how to.
Colored wiki:
This tag is used for images and animations that feature color. Usually given to child posts of a sketch or line art, and indicates that it is a colored version of its parent.
dubsthefox the beginning of the wiki you quoted supports my point.
"This tag is used for images and animations that feature color."
"Colored" is a bit too vague to refer to only colored versions of a sketch. Almost every drawing starts from a sketch and not every sketch is posted here.
Maybe there should be a more specific "colorless parent post" tag for this specific case?
dubsthefox said:
Colored wiki:This tag is used for images and animations that feature color. Usually given to child posts of a sketch or line art, and indicates that it is a colored version of its parent.
The first line fits with this use. If an image or animation features color, the tag applies. If an image has flat colors, it features color.
The underlined portion comes from the old wiki description before the decision to change it, and is redundant with the updated description (if a post is a colored version of a parent or child post that's only a sketch or line art, it has color, so it doesn't matter if there's an uncolored parent or child post).
I feel like we have to just embrace the tag's new definition because of the sheer amount of misuse, but honestly, I'm not really a fan of a tag that's so underutilized. -monochrome provides much better results.
It's probably quite trivial to make a bot that detects whether images are colored or not and add the tag, but we'd be looking at applying the tag to millions of images. It currently has 44k posts.
faucet said:
It's probably quite trivial to make a bot that detects whether images are colored or not and add the tag, but we'd be looking at applying the tag to millions of images. It currently has 44k posts.
I have seen multiple users in older forum threads (like the stickied ones) that were marked "blocked" because of tag abuse, specifically because they mis-tagged posts using an auto-tagging system/bot.
I do not know if such a tool is still available, but I would really like to use it to tag colored posts.
I can try to individually tag every single post, but it is slow and a pain.
faucet said:
It's probably quite trivial to make a bot that detects whether images are colored or not and add the tag
I'm not sure if it would be that trivial, given all the potential edge cases and tricks of visual perception. Particularly with JPG and other lossy formats, they can introduce a minuscule amount of coloring on what appears to be a greyscale image, imperceptible at a normal view, but more apparent when zoomed in. PNG can also have issues around alpha transparency, since transparent pixels still technically have a color that could bleed into neighboring opaque pixels when scaled/resized in the browser, or have stray nearly-but-not-fully-invisible pixels with a tiny smidge of color that a viewer can't see. Antialiasing techniques, particularly for text, can also introduce subtle coloring on edges, which is intended to align with the individual red, green, and blue lights that make up a pixel on your display (so-called subpixel rendering ); to the eye at normal range, it can make black and white edges look smoother and sharper, but zoom in and measure it, and you'll see subtly colored pixels around the edges. Automated detection could easily detect color on what viewers see as completely grayscale.
watsit said:
I'm not sure if it would be that trivial, given all the potential edge cases and tricks of visual perception. Particularly with JPG and other lossy formats, they can introduce a minuscule amount of coloring on what appears to be a greyscale image, imperceptible at a normal view, but more apparent when zoomed in. PNG can also have issues around alpha transparency, since transparent pixels still technically have a color […] Automated detection could easily detect color on what viewers see as completely grayscale.
I think the best ways to go are either:
or
I’ll make a BUR so that every [insert color]_body tag implies colored. That way, if people agree on the correct meaning of "colored", it will be a lot easier to use and become more common.
The bulk update request #5523 has been rejected.
create implication red_body (156250) -> colored (79933)
create implication brown_body (403819) -> colored (79933)
create implication orange_body (197052) -> colored (79933)
create implication tan_body (191826) -> colored (79933)
create implication yellow_body (226701) -> colored (79933)
create implication green_body (153124) -> colored (79933)
create implication blue_body (344009) -> colored (79933)
create implication purple_body (155481) -> colored (79933)
create implication pink_body (132733) -> colored (79933)
Reason: All of these tags contain a color. Logically, if a post has a colored body in it, the post has color in it and should be tagged as colored.
The [insert color]_body tags encapsulate all [insert color]_skin, [insert color]_fur, [insert color]_feathers and [insert color]_scales posts, so they are the most effective to catch posts containing a colored character, and therefore colored post.
EDIT: The bulk update request #5523 (forum #376069) has been rejected by @DimoretPinel.
Updated by auto moderator
Should colored apply to monochrome images with a single color tone, that uses a one color in place of black or gray?
post #4274012 post #4273832 post #4272954
watsit said:
Should colored apply to monochrome images with a single color tone, that uses a one color in place of black or gray?
post #4274012 post #4273832 post #4272954
I think colored should apply to monochrome images! Monochrome does not mean greyscale.
Your second example,
post #4273832
should definitely be considered colored despite being monochrome. Everything is orange, not black and white or otherwise greyscale.
The other two posts are definitely colored line art. That should count as colored, but I’m not sure many people would agree with me here.
dimoretpinel said:
Which would be fixed if more images were properly tagged, yes.I personally dislike non-colored explicit images, but can’t do much because colored isn’t used enough. I want to fix that, and I think this alias is a good, surefire way to do it.
I want to add the colored tag to multiple posts at once but I do not know how to.
most posts are coloured, no? and we don't tag defaults? I already gave a simple solution to help with that problem.
Okay, let's say you implied EVERY *<colour>* tag to colored
Let's see the result
-black_and_white -greyscale -*red* -*orange* -*yellow* -*green* -*blue* -*purple* -*pink* -*brown* -*tan* -*color* -*tone*
these are the posts that escape the grasp of your implications, yet are coloured.
Updated
I stand by with colored's original wiki definition. At some point, color_edit gets implied to it.
From what I see, the former should be for official coloured versions of monochrome sketches (that exist as parents) while the latter should be for unofficial third-party colour edits.
The new definition that "expands" it to all coloured artworks is unnecessary and is grounds for invalidation, since most of the artworks that exist here are coloured already.
Searching -monochrome should be sufficient for finding coloured artworks.
thegreatwolfgang said:
I stand by with colored's original wiki definition. At some point, color_edit gets implied to it.
From what I see, the former should be for official coloured versions of monochrome sketches (that exist as parents) while the latter should be for unofficial third-party colour edits.The new definition that "expands" it to all coloured artworks is unnecessary and is grounds for invalidation, since most of the artworks that exist here are coloured already.
Searching -monochrome should be sufficient for finding coloured artworks.
It would be better for colored to be aliased to colored_version, to avoid mistags.
also monochrome includes more than black white (and/or grey)
snpthecat said:
most posts are coloured, no? and we don't tag defaults? I already gave a simple solution to help with that problem.Okay, let's say you implied EVERY *<colour>* tag to colored
Let's see the result
-black_and_white -greyscale -*red* -*orange* -*yellow* -*green* -*blue* -*purple* -*pink* -*brown* -*tan* -*color* -*tone*
these are the posts that escape the grasp of your implications
Something strange is happening with the color tags
*pink* -*pink* rightfully has 0 posts
pink_horn -*pink* has more than 0
snpthecat said:
Okay, let's say you implied EVERY *<colour>* tag to colored
Let's see the result
-black_and_white -greyscale -*red* -*orange* -*yellow* -*green* -*blue* -*purple* -*pink* -*brown* -*tan* -*color* -*tone*
these are the posts that escape the grasp of your implications, yet are coloured.
A lot of posts there are mistagged to begin with. There are characters with brown/blue/green bodies that aren’t tagged as such on the first page of the search,
(for example post #4276240 does not contain any color tags despite green_skin and yellow_eyes being applicable)
post #4276240
as well as untagged greyscale posts.
(for example post #4276232 only being marked monochrome and not greyscale)
post #4276232
I would like to fix it, but correctly tagging every one of those posts on my own would be too long…
Updated
thegreatwolfgang said:
I stand by with colored's original wiki definition. At some point, color_edit gets implied to it.
From what I see, the former should be for official coloured versions of monochrome sketches (that exist as parents) while the latter should be for unofficial third-party colour edits.
Why not just stick with colored_version for this particular definition? There is also color_edit and colored_sketch.
dimoretpinel said:
Why not just stick with colored_version for this particular definition? There is also color_edit and colored_sketch.
Read my second line:
thegreatwolfgang said:
The new definition that "expands" it to all coloured artworks is unnecessary and is grounds for invalidation, since most of the artworks that exist here are coloured already.
Searching -monochrome should be sufficient for finding coloured artworks.
By default, the vast majority of artworks that exist on the site are coloured already, so we do not tag that.
We would and already do tag the opposite case, i.e., non-coloured/monochrome artworks.
dimoretpinel said:
as well as untagged greyscale posts.
(for example post #4276232 only being marked monochrome and not greyscale)
post #4276232
Some images can be automatically determined to be greyscale based on the header (encoded as greyscale rather than RGB). It would be interesting to know how many posts are 'correctly' encoded in this sense. I don't think the server currently tries to do anything with this fact at upload time, but I think it could.
However, ironically this particular post doesn't contain exclusively grey pixels anyway, so it doesn't match the definition of greyscale in the wiki, and monochrome is AFAICS therefore correct. I noticed this by checking the histogram -- 285 unique colors.
('colors'. Most of them are greys)
dimoretpinel said:
A lot of posts there are mistagged to begin with. There are characters with brown/blue/green bodies that aren’t tagged as such on the first page of the search,
(for example post #4276240 does not contain any color tags despite green_skin and yellow_eyes being applicable)
post #4276240
as well as untagged greyscale posts.
(for example post #4276232 only being marked monochrome and not greyscale)
post #4276232I would like to fix it, but correctly tagging every one of those posts on my own would be too long…
That's my point, so many posts are not tagged with any color tags at all, that implications to colored will catch only a fraction of all colored posts.
While of course there are many black_and_white/greyscale posts which aren't tagged properly, there are significantly fewer posts which should have the tag, by virtue of colored being default (and people are more likely to notice black_and_white/greyscale and tag it on a post).
dimoretpinel said:
dubsthefox the beginning of the wiki you quoted supports my point.
"Colored" is a bit too vague to refer to only colored versions of a sketch. Almost every drawing starts from a sketch and not every sketch is posted here.Maybe there should be a more specific "colorless parent post" tag for this specific case?
This part was added in June 2023. Before this change, it was very clear for 12 years that it is used for posts with an uncolored child/parent post.
thegreatwolfgang said:
I stand by with colored's original wiki definition.
The original wiki definition is improper, as it goes against the core tagging principles of the site and violates TWYS. We don't tag posts based on the contents of other posts, the presence or lack of color in another post shouldn't influence how a given post is tagged. No other tag is like that; line_art isn't for a line art version of a parent, sketch isn't for a sketch version of a parent, lineless isn't for a lineless version of a parent, etc. We don't tag sequence for a single panel image that's part of a sequence with its parent.
Parent posts can be added or deleted at any time, making a tag like that always in flux and unreliable. Say, someone posts a wip sketch or line art piece, then the finished colored version is posted and tagged colored. Then some days or months later, the artist issues a takedown on the unfinished version leaving only the colored version. What use would it be to remove the colored tag due simply to no longer having an uncolored parent? Or say someone posts a finished colored piece, then some weeks later decides to post a grayscale version, how does it help to go back and add colored to the old post? Nothing about the post changed, so there shouldn't be a reason for the tags on it to be changed.
I think the convo should be moved to topic #40278 because we’re mostly arguing about the correct usage of colored rather than the revelancy of the implication.
It would be better to untangle the earphone wires this debate is before moving on with implications.
The bulk update request #5523 (forum #376069) has been rejected by @DimoretPinel.
watsit said:
The original wiki definition is improper, as it goes against the core tagging principles of the site and violates TWYS. We don't tag posts based on the contents of other posts, the presence or lack of color in another post shouldn't influence how a given post is tagged. No other tag is like that; line_art isn't for a line art version of a parent, sketch isn't for a sketch version of a parent, lineless isn't for a lineless version of a parent, etc. We don't tag sequence for a single panel image that's part of a sequence with its parent.Parent posts can be added or deleted at any time, making a tag like that always in flux and unreliable. Say, someone posts a wip sketch or line art piece, then the finished colored version is posted and tagged colored. Then some days or months later, the artist issues a takedown on the unfinished version leaving only the colored version. What use would it be to remove the colored tag due simply to no longer having an uncolored parent? Or say someone posts a finished colored piece, then some weeks later decides to post a grayscale version, how does it help to go back and add colored to the old post? Nothing about the post changed, so there shouldn't be a reason for the tags on it to be changed.
If it violates TWYS because it uses content from other posts, how is it any different from color_edit or even edits (e.g., cropped, uncensored) in general?
thegreatwolfgang said:
If it violates TWYS because it uses content from other posts, how is it any different from color_edit or even edits (e.g., cropped, uncensored) in general?
Edits are a special case since they're adding things that the original artist didn't put there (or removing things they did), so it's helpful to know what was changed from the piece as made by the artist. Also, the tags apply even if the original isn't here; a color edit is still tagged color_edit regardless of the uncolored original being here or not (there's generally no reason it shouldn't be, but it's technically possible).
Friendly reminder that meta category tags are not strictly limited to TWYS - meta tags must only be “factually correct” to be valid.
scaliespe said:
Friendly reminder that meta category tags are not strictly limited to TWYS - meta tags must only be “factually correct” to be valid.
in practice that's really only true for the date tags and a few others. most meta tags still can be emphatically proven, it's just their scope is different from general tags.
The tag implication flat_colors -> colored (forum #376021) has been rejected by @slyroon.
