Commons:Deletion requests/Files uploaded by 冷床系

This deletion discussion is now closed. Please do not make any edits to this archive. You can read the deletion policy or ask a question at the Village pump. If the circumstances surrounding this file have changed in a notable manner, you may re-nominate this file or ask for it to be undeleted.

Files uploaded by 冷床系 (talk · contribs)

I doubt these videos are really free, and they are probably out of scope. Seeing the descriptions, they are mostly advertising with a lot of external links.

Yann (talk) 18:53, 15 September 2022 (UTC)[reply]

Keep I do not think those videos should be deleted.

All videos I upload are videos published on Youtube as CC BY 3.0. All videos are created by the contributors themselves and are not license laundering videos. In addition, once CC BY is granted, it cannot be revoked.
About 90% of the videos listed here are from two Youtube channels, "PINK-latte TV" and "Mr&Miss Campus KANDAI 2022". Mr&Miss Campus KANDAI 2022 is about the Japanese beauty pageant. There are some images of Japanese beauty pageant winners on Commons, but there are no videos that actually show what these people are doing, so I consider this video to be of educational value. PINK-latte TV is about web modeling, a new profession that was first created in the 21st century. There is also a video with junior high school student models explaining makeup and outfit coordination. Since there were no videos showing modeling, especially among Japanese, so I consider this video to be of educational value in showing the content of the profession.
As for external links, all advertising-like external links (bit.ly, lin.ee, amzn.to, onlineshop.jp, link.to, page.link) have been removed. If you are concerned about the descriptions, you may leave it blank.

--冷床系 (talk) 00:12, 16 September 2022 (UTC)[reply]

Talking the first in the list as an example File:【雑談】●●●してるときってナニ考えてるの…？【こまる-あおぎり高校】.webm, there is no link to the source (mandatory to verify the license), only to the channel, and there are still external links which are IMO not appropriate here. And there is no category, and the {{Information}} template is broken.
File:Speedpaint- At home (clip studio paint).webm and File:Huion Kamvas Pro 24 (4K) Review - Will this replace my Cintiq...-.webm look like advertising, are full of non appropriate external links, lack useful categories, and probably derivative works.
File:【Daily vlog】~Mr&Miss Campus KANDAI 2021~.webm: Non appropriate external links. Is the music really free?
File:CES 2022 - New Wireless VR Experience Powered by Private 5G and Edge Computing.webm: Advertising, no category, and derivative work.
Finally there are 420 videos with the same issues. We don't need to import all videos with a free license from YT. Yann (talk) 10:03, 16 September 2022 (UTC)[reply]
- I fixed the template issue and put categories on some of the videos.
- File:Speedpaint- At home (clip studio paint).webm and File:Huion Kamvas Pro 24 (4K) Review - Will this replace my Cintiq...-.webm; I recognize that the videos is not an advertisement, but a product description and a making-of video of Sunako's illustrations. The creator of this video is a person named Sunako, who is also the illustrator. The long blond elf character in the video is Sunako's original character.As you can see on Sunako's pixiv illustration page, the illustration is tagged "オリジナル"(original) which means a primary creation, not a derivative work.
- File:【Mr&Miss Campus KANDAI 2021】: This contest is considered a gateway to success for Japanese announcers and is a high-profile event. Normally, when using non-free background music for such events, some kind of contract is made with JASRAC. If they use non-free background music without a contract, JASRAC will immediately send a request to Youtube to delete it. It is obvious that we do not have access to the contract made between the governing body of the contest and JASRAC, but the final result of the contract can be seen in the license granted by "Mr. & Miss Campus KANDAI 2022" on Youtube. If you have clear evidence of infringement by "Mr&Miss Campus KANDAI 2022", the contributor of this video, I will upload the video with the audio removed instead.
- File:CES 2022 - New Wireless VR Experience Powered by Private 5G and Edge Computing.webm : I recognize this as a description of feature, not an advertisement. I do not think that HTC VIVE describing the HTC VIVE's functionality is a derivative work.
- Regarding the external links of various videos, you have described them as inappropriate. video2commons, which transfers videos from Youtube to Commons, will not work if there are inappropriate external links in the descriptions of the video unless they are removed. Are there any files that have been deleted because of the presence of Twitter, Instagram, or other external links of the video contributor on Youtube? video2commons does not restrict Twitter or Instagram external links, I don't see those external links as a problem.
- In Japan, girls rarely upload images and videos to Wikimedia Commons. Therefore, there is very little Creative Commons content representing the Japanese girl subculture and kawaii culture. From that point of view, I do not consider 420 videos to be a particularly large number. And there is no restriction ("If an article on the subject of the video exists, ...") like "one-article minimum" for free media files because files on Commons can also be used outside of Wikimedia projects for talking about Japanese girls subculture and Kawaii culture (that's why files on Commons are freely licensed). and there are no duplicates on the uploads, so keep all.
冷床系 (talk) 06:32, 17 September 2022 (UTC)[reply]

Delete Not even one of PINK-latte TV's files are used.(COM:NOTUSED) Commons is not backup storage of Youtube movie. --Netora (talk) 11:53, 17 September 2022 (UTC)[reply]
Since PINK-latte TV's files are "realistically useful for an educational purpose," I do not think that PINK-latte TV's files are eligible for deletion on COM:NOTUSED grounds.

Wikimedia Commons states: The expression "educational" is to be understood according to its broad meaning of "providing knowledge; instructional or informative".

The job of a web model is different from the old modeling job of wearing clothes and walking in front of people. A web model really has to do a lot of work, including TV-like projects and personal Youtuber-like projects, in order to maintain the number of views on Youtube. I don't think there are many 420 files to providide knowledge. 冷床系 (talk) 22:41, 17 September 2022 (UTC)[reply]
To prove that there are indeed useful and that they meet Commons standards, the following are necessary: all the files must have categories, the inappropriate external links must be removed, and at least some should be used in articles. Yann (talk) 08:54, 18 September 2022 (UTC)[reply]
”all the files must have categories, the inappropriate external links must be removed, and at least some should be used in articles.”

About this(↑). Is this your personal essay? Or is this a Commons policy? Commons:SCOPE states the following. "The expression "educational" is to be understood according to its broad meaning of "providing knowledge; instructional or informative." It does not mention categories or external links here. And there is no restriction ("If an article on the subject of the video exists, ...") like "one-article minimum" for free media files because files on Commons can also be used outside of Wikimedia projects for talking about Japanese girls subculture and Kawaii culture (that's why files on Commons are freely licensed). And please indicate if there is a Commons policy page that states that external links to a Youtube video contributor's Twitter or Instagram account are inappropriate. 冷床系 (talk) 03:32, 19 September 2022 (UTC)[reply]

Deleted: per nomination. . Jim . . . (Jameslwoodward) (talk to me) 15:14, 4 November 2022 (UTC)[reply]

This deletion discussion is now closed. Please do not make any edits to this archive. You can read the deletion policy or ask a question at the Village pump. If the circumstances surrounding this file have changed in a notable manner, you may re-nominate this file or ask for it to be undeleted.

Files uploaded by 冷床系 (talk · contribs)

Outside project COM:SCOPE, AI-generated images that are unused on Wiki projects.

Adeletron 3030 (talk) 17:13, 14 October 2022 (UTC)[reply]

Delete as per nom. Yann (talk) 19:33, 14 October 2022 (UTC)[reply]
Comment. File:Smiling high school girl with square glasses s-4012097278.webp is in use on the English Wikipedia, and File:Girl, impressionism, (faint light), looking at hand, s-2228255702.webp is in use on the Japanese Wikipedia. Maybe others are in use too - I haven't checked. --Zundark (talk) 19:35, 14 October 2022 (UTC)[reply]
@Zundark Sorry, thought I checked but apparently my eyesight isn’t great. Adeletron 3030 (talk) 20:03, 14 October 2022 (UTC)[reply]

Though the en-wiki NovelAI article is up for deletion, and more importantly for our purposes, the ja-wiki article says NovelAI images are copyright-free in Japan but in the US, they are considered derivative works of the original images they’re based off. Adeletron 3030 (talk) 20:14, 14 October 2022 (UTC)[reply]
Do you want to remove File:昭和４９年防衛庁記録.webm from the list, because the nomination rationale ("AI-generated") doesn't apply? whym (talk) 12:53, 11 November 2022 (UTC)[reply]
@Whym: I've struck it out in the list above and I'll remove the deletion template from the file page. Thanks. Adeletron 3030 (talk) 13:15, 11 November 2022 (UTC)[reply]

Keep a bit spammy with the sheer number of similar, generic images but as far as I know, A) AI images are ineligible for copyright, and B) there’s no reason high quality art is OOS because a robot made it. Dronebogus (talk) 13:33, 15 November 2022 (UTC)[reply]

Keep I do not think those images should be deleted.

I think these images satisfy COM:SCOPE because NovelAI can produce high quality images that satisfy a variety of prompts at the same prompt and it would take many images to demonstrate its ability to do so. And there is no restriction ("If an article on the subject of the video exists, ...") like "one-article minimum" for free media files.
File:昭和４９年防衛庁記録.webm is a video file about the Japanese Self-Defense Forces transferred from Youtube and is not related to NovelAI. File:Virgin-killing sweater s-1962640643.webp and File:High-waisted black skirt, white blouse, corset, full-length image s-3784166878.webp are used for 童貞を殺す服 on the Japanese Wikipedia page. File:Dakimakura, 2view s-2424069217.webp is used for 抱き枕 on the Japanese Wikipedia page. File:Nsfw, 1girl, legs folded, white panties, from above s-388429682.webp is used for まんぐり返し on the Japanese Wikipedia page.
The copyright description in the NovelAI article on the Japanese version of Wikipedia is based on the FAQ page on NovelAI's official website. That page does not state that NovelAI's copyright varies by location. Also, the United States Copyright Office indicated on February 14, 2022 that works created by AI are not copyrighted[1]. Please indicate if there are any legal documents or court cases where images generated by NovelAI were deemed to be derivative works of the original machine learned original image.--冷床系 (talk) 01:11, 15 October 2022 (UTC)[reply]

Keep I don't think it's a good idea to keep posting images of the same lineage endlessly, but I think it's okay to keep images that are likely to be used for educational purposes. Some of these images are available for educational purposes, as used by the Japanese Wikipedia.--Kainioaefa (talk) 10:59, 15 October 2022 (UTC)[reply]

Comment Regarding the pictures drawn with AI, if it is used to illustrate the functions of NovelAI, it is enough to keep only one picture. Although I am still a little worried that these paintings created with NovelAI may have unpredictable copyright issues, because it has been confirmed that the learning materials of this AI come from pirated illustration websites. (Or this's a mix of multiple original creators) --Cwek (talk) 07:06, 19 October 2022 (UTC)[reply]

Comment I do not think there is a special problem with copyright, and I will state my rationale for this.

Under Japanese law, the use of copyrighted works for the purpose of AI machine learning is basically permitted under "Article 30-4-2 of the Copyright Law" (著作権法第30条の4第2号). Regardless of the type of copyrighted work, its intended use, or whether or not it is used for commercial purposes, permission is not required.
If a painter writes "no unauthorized reproduction" or "no learning use for AI," it means nothing in law. Since it is not mutually agreed upon, but only written unilaterally, it has no effect as a contract.
From the viewpoint of governing law, if a "work created in Japan by a Japanese person is downloaded in the U.S. by an American person and uploaded to an American server", it is nearly impossible to judge that person under Japanese law. The acts punishable as foreign crimes are strictly specified in Japanese law (Articles 2, 3, 3-2, 4, and 4-2 of the Penal Code)（刑法2条、3条、3条の2、4条、4条の2）, and their provisions do not include copyright infringement.
In addition, there is a fair use clause when U.S. law is the governing law. Danbooru, a reprint site, seems to be very careful not to damage the interests of the person in question for fair use, such as running the site on a non-profit basis and always stating the URL of the reprint source.
Recall that Stable Diffusion, the same AI that automatically generates images, was also used to train a large number of images from Pinterest, which is known to reprint images without permission, but this has not been a particular problem.--冷床系 (talk) 04:20, 21 October 2022 (UTC)[reply]

Keep specifically the images that are in-use within various Wikimedia projects. Specifically referring to the point raised by Adeletron, until the United States passes a new law that applies additional restrictions on AI-generated works, we should not be playing "copyright police" on Commons purely on the basis of "what-ifs". Yes, we're still in wild-west territory, and legislation needs to catch up with technology, but we shouldn't be making such decisions until after the relevant legislation has been changed. Copyrights and ethics are two separate issues, and while I may argue that the use of AI diffusion models to generate art may not be fully ethical, that in itself is not a valid argument for deletion on Commons. For prior discussion on the topic, see Commons:Village pump/Archive/2022/10#AI-generated works. --benlisquare_{Talk•Contribs} 05:53, 6 November 2022 (UTC)[reply]
Delete per above. IMO they have COM:DW concern as well. --A1Cafel (talk) 16:54, 11 November 2022 (UTC)[reply]

It is the responsibility of the accuser to provide evidence of COM:DW violations. No evidence, no case. Can you provide a link to the original image(s) which were plagiarised by these uploads? --benlisquare_{Talk•Contribs} 07:59, 12 November 2022 (UTC)[reply]

AIs are kinda inherently DW but as mentioned ethics and copyright are two different things. There’s no ethical code against AI art on Wikimedia and also as mentioned there’s no copyright either. Dronebogus (talk) 13:38, 15 November 2022 (UTC)[reply]

My concern is that it sets a bad precedent that will ultimately harm the prospects of free-licence content, on Commons and elsewhere, if users can simply go "I suspect that this plagiarises something" as opposed to "this content plagarises X work by Y author, here is the URL for proof". Yes, it is definitely possible for a user to instruct an AI model to generate a derivative work (e.g. "generate a photograph of the Eiffel Tower at night with Spiderman on top"), but none of the cases in this nomination match that criteria; these are clearly "generic" anime women that are not protected under any pre-existing IP or art piece, and any slight resemblance to an existing art style would fall under de minimis within the context of United States copyright law.

Finally, regarding "AIs are kinda inherently DW": I have to correct you here - AI models have their algorithms trained on scraped image data, but the models themselves do not store this data, nor do they retrieve this data during the image generation process. As an example, the Stable Diffusion 1.5 model checkpoint file is 3.97GB in size; this AI model saw 6 billion pictures during the AI training process, and "condensed" that information into a set of algorithms stored in a single file that totals at roughly 4 billion bytes. Tell me, can you fit an entire picture in 66% of a single byte of information? It is impossible from a computer science standpoint. Training an AI model the instructions required to "draw a dog" or "draw a car" is no different from an art student observing a dog, and using this observation to draw a wolf. Now, is it ethical to train AI models using copyrighted imagery scraped from the public web? From my purely subjective opinion, it is definitely not ethical, as it does not allow creators to object or opt-out. But while it is not ethical, it's not a copyright violation, just like how me looking at a dog for inspiration isn't a copyright violation. --benlisquare_{Talk•Contribs} 00:29, 16 November 2022 (UTC)[reply]

I didn’t say it was a copyright violation p, but the correction is completely fair. Dronebogus (talk) 01:02, 16 November 2022 (UTC)[reply]

Comment For an example of an evidence-based COM:DW violation accusation for an AI-generated image, see Commons:Deletion requests/File:Nicocado Coinslot.png. A URL to the original image, ideally the name of the author if known, and optionally a summary of how image A is a derivative of image B. I would consider this to be the mere baseline for any accusation of COM:DW breaches. --benlisquare_{Talk•Contribs} 10:09, 3 December 2022 (UTC)[reply]

Keep It should be kept unless evidence of clear non-compliance is shown.--Bluebosh (talk) 01:37, 26 November 2022 (UTC)[reply]
Comment File:Smiling high school girl with square glasses s-4012097278.webp, File:A high school girl, rectangular glasses, cynically smiling, cat ears s-2194273305.webp, and File:Girl,_impressionism,_(faint_light),_looking_at_hand,_s-2228255702.webp are currently in-use on some Wikipedia articles as examples of NovelAI works and not "unused on Wiki projects".-Peacearth (talk) 16:38, 30 November 2022 (UTC)[reply]
Delete (except 3 in use) personal art, COM:WEBHOST, no educational value, out of scope. --P 1 9 9 ✉ 03:02, 1 December 2022 (UTC)[reply]
- FYI not only 3 images are in use. See earlier comments in this RfD. I added short texts for convenience. --emk (talk) 08:55, 1 December 2022 (UTC)[reply]
comment it’s been established that copyright and ethics are not valid reasons for deletion in regards to AI art. As for scope issues, none of the delete voters who complain about scope explain why a large number of quality images representing various anime-related concepts are out of scope in any detail, whereas at least one keep voter has. I think this nomination is indiscriminate and should be closed without prejudice to re-nomming redundant or less useful images. Dronebogus (talk) 12:41, 1 December 2022 (UTC)[reply]
You said "high quality" and "quality images" here a few times. How did you measure quality? Is it related to Commons:Quality images in any way? whym (talk) 07:26, 4 December 2022 (UTC)[reply]
No, but I think anyone with functional vision can tell they don’t look like a bad MS paint drawing. Dronebogus (talk) 14:04, 4 December 2022 (UTC)[reply]

Deleted: per nomination, except a few in use. It should be clear that AI erotic images, and more generally AI images without a clear educational use are not welcome on Commons. --Yann (talk) 10:48, 5 January 2023 (UTC)[reply]

Commons:Deletion requests/Files uploaded by 冷床系

Files uploaded by 冷床系 (talk · contribs)

Files uploaded by 冷床系 (talk · contribs)

Navigation menu

Search