Turn about and all that!
3 Likes
I just watched this video today. Itâs definitely a beautiful result. The criticism is: it could be better. The lighting and rendering are good⌠but (ironic) the charactersâ expressions and actions are completely artificial, even with a cool camera angle saving the situation in some scenes.
The old problem of hands (extra or missing fingers) is present, as is the problem of the consistency of the characters: in one frame the girl has her hair parted in the middle, in the other, she has bangs covering her foreheadâŚ
In certain frames the author placed the male character next to the girl. To me, it seemed that the character was too small in relation to her: perhaps these frames should be rethought, moving the camera a little further away, so as to gain more space and be able to place the characters in the correct proportion. I donât know if the author is an artist (I donât think soâŚ).
Anyone got thoughts on "Vibe coding"? Really interesting? horrifying? just a stupid meme?
2 Likes
How to shoot yourself in the foot, now solved with âvibeâ!
Michelist
3 Likes
I kind of understand where the hype came from, but like everything AI-related, it will rapidly turn into something much larger and potentially very dangerous.
(Note: I never heard of this concept before writing this message)
On one hand: much like any AI tool, I think this âvibeâ practice may be a great way to brainstorm new concept ideas by mixing existant themes into brand new ones, and assessing if an idea is worth it for the market in a small window of time, for example by doing a small prototype before assembling a team or getting into the real project.
In other words: people using it to boost their thoughts and honed skills (not replace them), might be an acceptable way of doing this âvibeâ practice.
However: we all know some people wonât just stop at the brainstorming phase and will carry their entire projects with AI as amateurs (probably for profit?), which can lead to serious security risks. There will be bugs that wonât be solved, doorways for attacks, sensitive info leaks⌠And the fact that people will literally run an unknown code (not checked by code professionnals or compliance experts, without their knowledge) onto their devices is indeed very scary. Especially at a time where cyberthreats are getting more frequent and elaborated.
Not only that, but Iâm sure some kind of copyrighted code could also mix into the âvibe codingâ productions, much like it already does with image-generation AI tools, so this also leads to another ethical/legal debateâŚ
In other words: people using a powerful tech with absolutely no knowledge of what theyâre doing⌠it is indeed scary. We can guess where this leads, THAT is not good.
Hoping to see more feedback and perspectives on this! Cheers.
1 Like
âVibe codingâ is definitely a funny meme, just like âtech broâ, and other brainrot terms 
To me, the best case scenario is that it can encourage non-coders to try their hand at âprogrammingâ, and actually get something amusing out of it. If they get interested, maybe in time it will turn into a desire to understand whatâs going on and really learn programming.
âBack in the dayâ you could code your own âgameâ in text mode by copying the full listing in BASIC from a magazine page. Nowadays, perhaps you ask LLM to generate the game for you, and I hope it wonât be where people lose interest.
Now, speaking in practical application terms, hell no, I think this is completely unusable. In a professional setting you need to understand what the code does, and there are dozen other people who work on the same code base. You need to be sure that what youâre submitting makes sense and wonât become a technical debt in, like, two weeks from now. Also, the concern of copyright is paramount, so anything âAI-generatedâ needs to be checked for copyright violations.
But donât get me wrong, code generation absolutely has its place and thereâs a big push in the field to âmake it workâ and increase productivity as a result. Generating the whole app makes no sense, but generating single test cases, snippets, or just suggestions to work off of is what people are actively using today.
(âŚchanging topicsâŚ)
Itâs getting a bit long, but on an unrelated note I wanted to mention art as well. Welp, I browsed X and Pixiv a bit today and wow, has the needle moved forward. Especially for me, a person dead set on anime style, itâs astonishing how good gen AI has become for this. If youâre in a niche of âa popular character striking a sexy pose, highly renderedâ, then just⌠do anything else. Thereâs no point âcompetingâ. The ML models are so good that 99% of non-artists just wonât know and wonât care, at all. If youâre still drawing and painting âanimeâ in 2025, to me the reasons should be the following:
- youâre doing a highly uncommon or experimental style (think impasto anime )
- you have original characters, great composition, and storytelling
- itâs a story-first work, like maybe a manga, animation, etc.
- youâre just really, truly, enjoying the process and not just the end result.
- or youâre using a traditional medium rather than digital
I may be overly negative now, but Iâd be surprised if AI gen wonât become mandatory in professional field soon (as in fully commercial stuff). Just think: would you pay the pro illustrator to paint an artwork in 10-20 hours, or would you rather ask them to correct an AI picture in 1 hour? Yeah, Iâm afraid that is or will be our reality soon. Same as with coding, increase the productivity (only that with source code I donât really care, but art is a whole different kettle of fish for me).
Personally, Iâm doing this for fun and a sense of accomplishment, but I need to rethink my goals and the types of artworks I draw in the futureâŚ
2 Likes
Solid points, butâŚ
(need to be said : The anime world is a very specific one, had always a lot of industrial production aspects, this makes it more vulnerable, besides having been scrapped to no endâŚand I believe that for all this, the things you said apply in many ways)
This is what I thought until not long ago. The thing is that I have been detecting (long enough to be sure it is not a âglitchâ or a âsoon to die trendâ) a much more strong -and more interestingly: increasing by the minute, the progression is important! - back slash against creative AI content (illustrations, video, music) than I ever expected. The articles and tutorials hyping AI often end up without traffic, now (except the ones with already communities formed of the so called âAI brosâ, but I am speaking of broad difusion, main public). I thought initially that such reaction would go nowhere, as at the beginning, it was mostly artists complaining, and while the internet multiplied crazily the number of artists in the last decades (compared to my times in the 80s and 90s), weâre a minority in the planet. Happens to be the case that the disgust, saturation or whatever it is, are massively coming also from the non affected, too!! Some due to the evident ethical reasons, others just not in love with some specifics of the outputs (and just saturation of certain aesthetics, repetition of âvibesâ). But the main reasons might be two: Many fear what AI would do to their non-creative jobs as well (the toy becomes a threat-> the joy disappears), at some point in the future. Among those, many young programmers (I check this daily, in several large communities), BTW. Yes, I know it wonât kill programming jobs, (or not many, for now) but many feel that way. And once again, perception is king, since always in our societies.
Besides that, the copyrights issue (outside of the anime world, or even the entire Instagram sub-world; I am referring to a much wider range) is quite a big one. Specially for publishing (so many distributors are not willing to use it) or doing anything commercial with it, protecting your own product that has that âartâ that can be used by anyone else, or even companies not trusting of secrets getting leaked, producing wrongful details/images, etc, etc, etc. Even if the US regulations are going backwards in that sense (in some cases), it is not going that way in the rest of the world. And I suspect things will get more sensible in the US, at some point, too, in this specific matter. There are many large firms, powerful multinationals, powers that be with strong interests in IPs, etc. Those are going to be very active, and have too much money to lose (and to use it in legal actions, pressure, etc).
Yet another factor (even more important reason, and related with the first aspect) : Humans look for the human behind the art piece, song, comic, etc. One piece known to be âgeneratedâ loses 100x times its perceived value (if it keeps any: âdoh, someone wrote a sentence and did hit a buttonâ), almost automatically. Hence why the whole problem would be resolved by enforcing any generated AI image to have a watermark or even small note saying âgenerated by AIâ. But the services, companies and individuals exploiting this, obviously do not want to act this fairly: they know the fact that non-human-made has way less value! The intention is to fool the audience!.
This is all about psychological matters. One of the several reasons why artists should now get a lot closer to their communities, build links with it, âbuild communityâ, get personal. And why Youtube content creators, even just âtalking headsâ (as people can tell still very easily when itâs a real person and when notâŚsooner or later, they know) will keep thriving with no threat at all, as for these is where the thing is very hard to fake effectively by the AI (at least after the first videos, no live authentic videos, etc), towards a community. And folks want the human, weâre just built so (music concerts would not exist, otherwise would suffice just the freaking vinyl, disc, and now, just downloadable file or audio streamâŚand people pay tons of money to be âthereâ even if can only see the person from a far distance). People also want the relatable being. Not a washing machine or etc automatically producing art or music ;D
But we will need to change our ways and strategies, that is for sure.
I am both a traditional and digital painter. And I initially thought I would have to go back to traditional, to âsurviveâ. But more and more Iâm increasingly changing my opinion. Most likely I will become very âhybridâ (or go back, as there was a time when I produced about the same amount of traditional than digital art). And the reasons now are not that much more for âsurvivalâ⌠but because this whole thing made me value a lot more my so much loved oil paints, watercolors and brushes.
Last but not least. I believe that fully hand made and digitally drawn comic (including manga, even if it is not my cup of coffee) has an extremely long life ahead.
And those making fully rendered, realistic painted art, illustrations⌠I would totally keep doing it, as if nothing had happened, if one has a job besides art. You do what you love, AI has not changed a centimeter of that, it cannot do anything to stop people to keep enjoying it (as creators and as audience). Even more, if it is about relevance, I am 100% sure that actual artists producing their really painted art will have much better public reception, by a lot (in terms of traffic, too). Obviously proof of stages, sketches, etc might be needed, but careful with that⌠AI is scrapping that, too, as these people and companies are getting as far as faking speed-paint videos, sketches, etc (if that is not proof of the intent to fool the audience, and not competing âcleanlyâ, I donât know what it is).
1 Like
I think it might be useful for people to check out EU AI Act (the full text is here though the site gives the gist). The EU AI Act is mainly about what EU calls âunacceptable riskâ and âhigh riskâ uses of AI (i.e. automated social scoring systems are banned, systems that manage workers are considered high risk -AFAIK some wanted to categorize those as unacceptable too but not everyone agreed and it is something to reassess in a few years- and thus need to be approved for use within EU, etc).
Generative AI isnât considered unacceptable or high risk, but there are still requirements it has to follow, including transparency about the training data (copyrighted data can be used for training unless the copyright owners opt out in a machine readable manner, with an exception in place for scientific research) and a requirement for transparency about AI generated works so that if a work is generated using AI it must be labelled as such so that users can know about it.
There are some other stuff in there (e.g. âopen sourceâ AI has much more lax requirements, though what is considered âopen sourceâ is a bit stricter than what some AI companies claim - e.g. Facebook claims their LLM to be open source but EU disagrees), but for the most part the two important aspects for âcommon peopleâ here are:
- Copyrighted content can be used to train AI unless you mark it in a machine readable way (so that the scraping bots will know to ignore it). The exact mechanism isnât defined yet (theyâre still working on it but there are a couple of existing mechanisms and most likely will adapt one). Personally iâd expect places where artists upload images (not sure where that is nowadays) to use that mechanism automatically. E.g. this forum could auto-tag images as âdo not train for AIâ.
- If you generate a text/image/audio/whatever with AI you should mention it that it was generated by AI so people know about it. Since the purpose of this is for the users to know, some machine-readable tag isnât enough, it has to be mentioned somewhere in the proximity to the generated content. The EU AI Act page has an example of that where the image caption used for the article mentions the image was generated using Adobeâs AI.
There are a few things around copyright that the EU AI Act doesnât address. AFAICT there are multiple âcopyright concernsâ around AI use:
- Bot scrapping (i.e. the bots crawling the net for stuff to train on)
- If and how copyrighted materials affect the training results (hyperparameters/weights)
- Copyrightability of the training results (aka models, though really hyperparameters/weights)
- Copyrightability of the generation parameters (aka prompts)
- Copyrightability of the generated content
The EU AI Act is clear only on the first one (scraping) since it piggybacks on existing laws about data mining, search engines, etc. Essentially it defines an opt-out mechanism that, due to its machine-readable requirement allows for automation.
For the second (how copyrighted materials affect the training results) is less clear, though from my understanding (and some interview i read some time ago on Reddit) the copyright doesnât âtransferâ from the training data to the training results.
For the third (copyrightability of the training results) the act mentions nothing, but AFAICT it all depends on if the training process is considered machine-produced or not (like in US, the output of a program is not considered copyrightable in EU, so if the training process is not human supervised, the training results -hyperparameters/weights/etc- are also not copyrightable). A caveat though that even if the results are deemed not copyrightable, the containers (files) can still be. To understand this consider that how under some jurisdictions (including, AFAIK, the US) bitmap fonts are not copyrightable but bitmap files are - this means you canât share a -say- .FON file (Windows bitmap font file format) but you can extract the bitmap data from the file and share the results as your own file.
The fourth isnât addressed either though i expect it to be treated as user interactivity since the âpromptâ is essentially the UI into the model, thus treating that as copyrightable would be the same as expecting to copyright the actions involved clicking buttons and menus in a GUI program or entering text in a command line - in other words, not really copyrightable. I do mention it though since iâve read some people online who think they can âownâ prompts.
The fifth (copyrightability of generated content) also falls under the âmachine generatedâ umbrella and isnât copyrightable - nor covered by the EU AI Act. However one misconception i see often is that if one uses AI to generate part of a work then this ânon-copyrightabilityâ extends to the entire work, which is not the case. The reason computer output is not copyrightable is because the computer is not a person but as soon as some human effort becomes part of the work, then the human gets copyright. In other words if you generate an AI image of a cat, then the cat is not copyrightable. However if you open Krita and draw a moustache on the cat then that combined image is copyrightable. Think of AI generation in this case as a filter or noise generator - you wonât expect someone to âownâ random noise, but youâd expect someone to own an image that incorporates random noise as part of its creation.
(BTW i mentioned the above in a forum with some very âAI-proâ people and someone pointed out that this would make 3D renders noncopyrightable, which obviously isnât the case â i think it should be obvious to people who do art like here, but just to be clear in the case of 3D renders the human effort is into modelling, setting up the scene, lighting, etc)
The EU is working on a new act (or whatever it is called) explicitly about copyright with AI to clear most of the above.
Of course that is only about EU but TBH considering its history with various other tech topics so far, i doubt the rest of the world will have more strict rules than EU around the topic.
2 Likes
The EU is a big chunk of the world, so this is all progress. Thanks for sharing. I wish training on copyrighted works w/o individual consent was illegal (the holy grail), but Iâll take any protections we can get!
3 Likes
I know the conversation around AI art is very controversial and I donât really want to get into the weeds with this post.
But I did want to see what people thought of AI video generation from an image.
Personally I think this particular aspect of AI is very cool. I love the thought of being able to draw something I really like and then watching it come to life. It feels almost like something from fantasy.
1 Like
I thought this was interesting:
Now the âbig gunsâ are going after them.
5 Likes
Favourite line from the article:
âMidjourney is the quintessential copyright free-rider and a bottomless pit of plagiarism,â
6 Likes
Honestly, I feel like AI art does not have the emotion in it that original art has. The expressions and emotions in it are what makes it beautiful.. Anyway just my thought⌠
2 Likes
Is there light at the end of the tunnel or is it just authors waving a torch?
(The subheading has spelling mistakes. Itâs $1.5 billion.)
I have no legal qualifications or experience but Iâd make the following points:
- Published books usually have prominent and detailed statements of copyright in them.
- Book publishers are often wealthy even if not many authors are.
Also, note that the article states that it was the use of definitely âpiratedâ books that was judged to be illegal.
The use of purchased books was apparently judged to be ok 
An extract from the article:
The company nonetheless won part of the case, on grounds that scanning books is fair use and using them to create âtransformative worksâ â the output of an LLM that doesnât necessarily include excerpts from the books â was also OK. But the decision also found Anthropic broke the law by knowingly ingesting pirated books.
I think a book is âpiratedâ if you didnât pay for it and then used its contents in breach of its copyright statement. (I am not a lawyer.)
Also, I believe, any award for damages is related to estimated loss of current and future probable earnings from the work by the creator of the work and/or owner of the copyright.
2 Likes
Is AI a threat or a serious threat to the artist community ? 
Directly or indirectly?
To find it out by yourself, it would be sufficient to read up this topic, and perhaps also its predecessor topic.
But in short, even some artists from our community are now replicated by AI because their works were illegally scraped and used to train these AI pixel generators. Other users from this community lost their jobs, because they were replaced by AI.
I donât know how you would call it if someone sells pictures in your style and says: âLook, here I have a picture from @luttuputtu for you, what are you willing to pay?!â, or how you would like it to be fired and donât know how to pay the bills for your living anymore? But I guess, you wouldnât find it great and funny, I guess further, the word âserious threadâ may be amongst those that you will then have in your head and on your tongue! Am I wrong?
Michelist
2 Likes
So, software engineers and artists both are on the same pace!
Still, I canât figure out why we are going so deep into AI because if we donât use it in the first place, the level of training itâs getting could be affected but thatâs a hypothetical situation now! It canât happen. 
We? Whoâs we? A bunch of billionaires and tech bros are mostly pushing it onto the rest of the world. Itâs not âusâ or âweâ, itâs them. Itâs the same thing as with NFTs or Cryptocurrencies.
2 Likes
Yes, rich people have found a way to become even richer. It doesnât matter to them that others have to suffer because of their greed.
And all the blind people who believe in them, are only so long blind as long as they are not affected by the side effects, so lost job or income because now everyone can âpaintâ (calculate!) pictures that look like those they made their living with. Some found it funny, did it allow them to pretend to be an artist, something the most never were or wanted to be, almost always because they are too lazy to practice and learn how to paint.
This technic has brought so much evil into the world of art, it is almost as if you have to witness how your child is abused in front of you, and you are unable to help.
If you havenât already done so, you should definitely read this and the other topic I linked to. I think they will open your eyes to what this technology has done to the artistic community. It began with the biggest art theft in the history of the world. They have stolen far more than all thieves, including the fascist regime in Hitlerâs Germany in the last century, could ever have stolen combined. They scrubbed any museum, art gallery, artist community, etc., they found and fed it into their AI without paying anyone.
Only stupid people can be happy that computers do their thinking for them. But we like to be dumb as a brick!
Michelist
2 Likes
@luttuputtu Right now it poses an ethical dilemma because of the art theft these AI generators rely on in order to produce images. Thatâs why no AI art is allowed to be posted in this forum.
In terms of the threat, I personally believe some of the art jobs that have been eliminated by AI will return due to the lack of copyright on the generated images and also because of the difficulty AI prompters experience in complying with edit requests from project teams.
2 Likes