Everything is Ghibli
carly.substack.com
carly.substack.com
There were a million Doom clones, none of which were as good as Doom. The same will be true of AI art copycats.
This, however, is not the first time ghibli has had competition https://en.wikipedia.org/wiki/Mary_and_the_Witch's_Flower in fact they have a whole studio dedicated to copying ghibli: https://en.wikipedia.org/wiki/Studio_Ponoc
Yes they used to work at ghibli, but so too did john romero work at id, and yet daikatana was not a quake-killer.
This doesn't devalue ghibli at all, I think
(In fact, I think AI will always have the fundamental problem that most people have no taste or sense or introspection, they don't know why good things are good, and can't see that crap things are crap, so they are predestined to only be able to produce garbage. Nod to Ted Sturgeon.)
Enabling mass-production of Ghibli style without permission or monetary compensation is theft.
It's promotion for OpenAI's product, without any of the appropriate licensing. 3D printing companies don't provide Lego schematics to sell their products. There's also the small matter of their ex-employee turned copyright whistleblower, who ended up dead:
Japanese artists exist.
Would the world be better off if Picasso's heirs owned Cubism and any artist wanting to produce works in that style had to buy a license?
“When we opened the new studio, Studio Ponoc, I went to report this to Mr. Miyazaki,” he went on, “and he gave his blessing and said, ‘You really need to have the conviction to go create a new film studio and the conviction to show children worthwhile films. And every film you make, you’ll have to realize that has to be a film that is worthy to show to children’.[1]
But if your model produces outputs that too closely match their inputs and a company can show it that is a copyright violation and you can be sued for it.
It is not "competition" and "copying", it is the fact that Ghibli almost closed for good several times, so some employees created their own studio.
>On August 3, 2014, Toshio Suzuki announced that Studio Ghibli would take a "brief pause" to re-evaluate and restructure in the wake of Miyazaki's retirement. He stated some concerns about where the company would go in the future. This led to speculation that Studio Ghibli will never produce another feature film again. On November 7, 2014, Miyazaki stated, "That was not my intention, though. All I did was announce that I would be retiring and not making any more features."[40] Lead producer Yoshiaki Nishimura among several other staffers from Ghibli, such as director Hiromasa Yonebayashi, left to found Studio Ponoc in April 2015, working on the film Mary and the Witch's Flower.
True, but this was also during a time when it was incredibly hard to make a video game. It wasn't like anyone could spin up a Doom clone in 2 minutes. Competition vs. commoditization at massive scale are different things.
But for illustrators and graphic artists? What's now keeping me from downloading an illustrator's portfolio and telling ChatGPT to make me something in that style, but for my company?
Is there still value in that illustrator pioneering her unique style? It used to be a client magnet, now a good portfolio might take your potential clients away.
I guess the other side of this is that the more a style/idea/media spreads, the more comesw back to the original creator, but I'm doubtful this is true in AI.
AI is just another tool for artist. AI by itself never generate "art". It cannot by definition.
Sorry to take this on a tangent, but the problem with Doom clones isn't that they aren't as good as Doom, it is that Doom already exists and is known to the audience. If you've had your mind blown by Doom, playing a 10% better version of Doom isn't going to blow your mind again, it is going to merely be a fun experience. Many people won't even bother to try that 10% better Doom clone, since all they'll see is a clone of something they already tried.
That is, clones rarely work, but evolutions do. Stardew Valley on the surface can be considered a Harvest Moon clone, but it iterated on the formula, leading to a lot of attempts at casual farm games from many different competitors. Minecraft was an Infiniminer clone (or inspired by?) and iterated on the idea. Fortnite was a PUBG clone which was a DayZ clone.
It’s always the already-popular artist’s identity giving clout to the model, not the other way around.
And if the artist’s style is not well known, their identity is obfuscated because the model/LORA/“finetune”-peddler can get away with it. And they’re all peddlers, if it’s not OpenAI it’s grifters with Patreons to fund their “hard work” of tagging people’s work and throwing it at rental GPU compute.
Sure, but there were some "Doomlikes" I would still rate as better than doom; like Build Engine games Duke Nukem and Blood/Blood 2 and other IdTech based games like Hexen.
And even in Ghibly studio case it's not quite clear, if they won't be affected long term.
Is this an exaggeration? Or do some people literally have no introspection?
[1] Mary and the Witch’s Flower https://imdb.com/title/tt6336356/
[2] https://otakuusamagazine.com/hayao-miyazaki-says-he-wont-see...
The trouble with things like climbing is there are only so many mountains to go around. We already can't walk in many places because of cars. I don't look forward to the day that similar vehicles can go up mountains. The existence of mass-produced clothing doesn't affect your ability to do your own knitting, though.
> The trouble with things like climbing is there are only so many mountains to go around.
This is taking the metaphor far too far. Nobody is literally taking mountains away from people.
There's a big gap between "pretty much nobody" and the reality, which is somewhere between one third and two thirds of everybody. You might want to reflect on exactly how your perception diverged so radically from reality.
Do people in cities suffer from not being able to grow their own food and make their own clothes? I don't know for sure, but official statistics claim that, even today, they commit suicide at much higher rates despite having much less material scarcity. Robinsonades have been a popular genre of fiction for centuries, suggesting that people long for that kind of autonomy. Today, we also have zombie apocalypse fiction, RPGs, and preppers.
From another angle, sports consist entirely of skills and challenges which aren't needed and never have been, suggesting that they don't get forgotten. Hobbies also consist of skills and challenges which aren't needed.
It's not about climbing.
It’s not a moral judgement, that’s just how humans are wired. The lows make the highs higher.
I can imagine AI art having a similar effect (creating a glut of images/logos/whatever that devalues ones made with care) but am hopeful that we'll get better at filtering the cream of the crop. In 5 years tons of things will have AI logos that would have been made by a graphic designer (or simply not made) in the past. That sucks for graphic designers who are out of a job, is good for people who get cheaper logos, and TBD for overall society who now has lots more "custom" logos etc to wade through.
Obviously having roads is a great boon to a park's accessibility, and the ability of people with different mobility needs to appreciate nature. But it also made me thoughtful to imagine the feeling of wonder at seeing bridalveil fall after hiking for days into a roadless yosemite valley; how much more special and impressive it would seem after that journey?
This metaphorical tangent is pretty far removed from the original discussion, but how do you weigh the accessibility of a thing against how that accessibility changes its nature?
This is such an old man “I used to walk uphill both ways” take.
Not everybody has the TIME COST to pursue being an expert in art or code or whatever. But if they have an amazing idea and can now use AI to produce the idea then that is a beautiful thing!
For example: Having an idea for a cartoon used to be a dead end. It would die in your head because most people cannot stop their life and dedicate a substantial amount of time, effort, and sacrifice to produce the single cartoon idea.
This kind of mentality would ban Star Trek replicators, should they be invented one day. "In my day, you had to actually make things, we didn't get to replicate them, so we shouldn't, even if it's possible!"
What's the point in having an idea for a cartoon in your head if an LLM can just write an infinite amount of cartoon ideas in a heartbeat, and probably a better one than you came up with.
But drawning a cartoon isn't very challenging. Most of my peers could draw someone from South Park in a junior school.
The hardship in making cartoons is the amount of choices you need to make and the amount of knowledge how those choices would impact a viewer. If you delegate all of that, the cartoon wouldn't be simply blunt, it would be self-contradicting. And we already had a way of making cartoons, that allow your writing to shine through bland animation – since flash, actually. It might actually be even faster then using generative AI
It is enshittification.
Tangentially, what does enshittification mean now? Quoting Wiktionary, at one point it meant "The phenomenon of online platforms gradually degrading the quality of their services, often by promoting advertisements and sponsored content, in order to increase profits" (coined by Doctorow), but now people seem to use it to mean... things becoming shit?
You could argue this is the very sort of activity they were criticising when they posted! We are all vulnerable.
> Second: the fact that a neologism is sometimes decoupled from its theoretical underpinnings and is used colloquially is a feature, not a bug. Many people apply the term "enshittification" very loosely indeed, to mean "something that is bad," without bothering to learn – or apply – the theoretical framework. This is good. This is what it means for a term to enter the lexicon: it takes on a life of its own. If 10,000,000 people use "enshittification" loosely and inspire 10% of their number to look up the longer, more theoretical work I've done on it, that is one million normies who have been sucked into a discourse that used to live exclusively in the world of the most wonkish and obscure practitioners. The only way to maintain a precise, theoretically grounded use of a term is to confine its usage to a small group of largely irrelevant insiders. Policing the use of "enshittification" is worse than a self-limiting move – it would be a self-inflicted wound.
We have no right to tell people they have to learn to climb to get to top of the Everest.
I can't draw but I want to create my Art using AI. What I now see is a bunch of people who associate their self worth with a rare talent and don't want others to join the party. I want to resolve the issues around copyright for training, but once this is out of the way I want to draw exclusively through AI because it's the only way I can do it. And I LIKE IT.
I'm a skilled pianist. The funny thing is that I heard similar criticisms about computer music a couple decades ago. "No playing skill needed". Despite knowing how to play, I'd rather do computer music nowadays anyway. Please stop telling me what I can and can't do!
Computer music, as it existed a couple decades ago, still played exactly what you asked it to, and it wasn't filling areas where you underspecified the music with a statistical model of trillions of existing songs. And that's the difference, for me: the ability to underspecify, and have the details be filled in and added in a way that to the audience will be perceived as intentful, but which is not.
Whether drawing (writing etc.) through AI counts as drawing (as making art) is a debate we have to resolve in the upcoming future.
As soon as we get more control over AI output, those arguments will finally die their well deserved death
To those that AI art offend: don't think of people as artists, but simply as art directors dealing with stubborn artists that won't ever work to spec
Dismissing the argument that we are losing something in this "democratization of creativity" by fighting a strawman that says you are not allowed to participate instead is a bit lazy
My, my, you really took the worst example to defend your point. The Everest is now an overcrowded dumping ground full of cadavers, shit and trash, with idiots putting not only themselves but their sherpas and other mountaineers in danger due to their arrogance, lack of ability and shittiness.
>What I now see is a bunch of people who associate their self worth with a rare talent and don't want others to join the party.
What I see is a bunch of people creating digital doubles of existing artists without their consent and using it to make money.
>Please stop telling me what I can and can't do!
Oh the irony...
You're just not shallow in the parts of reality that you care about.
Don't feel superior for you are not.
Unlike many, I am not going to follow along into caricatural post-modernism where nothing is good, nothing is bad anymore.
Explain to me why nihilism is factually incorrect?
Good and bad is all relative to the perceiver.
Code is a way of creating something that itself may or may not be good. But the actual code - I agree, it can be objectively bad.
The Nihilism concept, as I mean to use it here, is more about meaning, values, aesthetics. Concepts like Logic & Math, not so much.
is it a net negative to society if the average person could produce so much art that it becomes post-scarce?
AI generated images are only an extension of what e.g. photography has experienced in the last decades. We’ve had film cameras, then digital cameras, then smartphones, each of these commoditized image creation by a then-unthinkable factor.
It’s an ongoing process, even if this leap seems especially big.
I suspect the wannabees exist in the narrow window when technology has expanded enough for non-professionals, but hasn't seen wide enough adoption that the man on the street will recognize the pretentious self-aggrandizement.
You must not be looking very hard. There are many youtubers or influencers making indie films or shows.
NigaHiga, Annoying Orange and Shane Dawson all made movies. Freddie Wong started out as a Youtuber and created Video Game High School.
Go to Amazon and drop a few grand on mics, lights, cameras and lenses. The result is production quality beating any 90s talk show, which would have taken a whole team to do.
No, I was talking about photography - and people replacing a digital camera with a smartphone. For most this substitution works very well; and the whole digital camera industry has shrunk significantly[1].
The photography community has been discussing wannabe photographers ever since my uncle bought a dslr and started taking photos at family weddings.
[1]: https://petapixel.com/2024/08/22/the-rise-and-crash-of-the-c...
Today's musicians have far greater access to lessons, recording equipment, inspirational material than 100 years ago.
Mountain biking (80s single speed with no gears, suspension, etc.) versus modern e-bikes with radial tires and hydraulic brakes.
Who cares? Value your own experience as you do. The less we all think about prestige, the more it will go away.
>The time is coming fast when the only living thing around a motion picture house will be the person who sells you your ticket. Everything else will be mechanical. Canned drama, canned music, canned vaudeville. We think the public will tire of mechanical music and will want the real thing. We are not against scientific development of any kind, but it must not come at the expense of art. We are not opposing industrial progress. We are not even opposing mechanical music except where it is used as a profiteering instrument for artistic debasement.
https://www.smithsonianmag.com/history/musicians-wage-war-ag...
I don't think that rises to the level of "mass media has made the world worse", but it certainly has made it different and lost a few things.
That's exagerrated by movies. There were folk songs, but instruments - and players - have always been rare, and played by those who had enough time to learn to play - the poorest of the poor, the crippled and the blinds...
Note that at that time, either there was already a copyright to protect musical partitions, or if learning by ear the teachers had total discretion on who they taught and for how much. There were still barriers to knowledge.
And outside of Europe, any ceremonial/religious music was only taught to the initiated and played on rare occasions (there are interesting stories of Americans, etno-musicologists or even just musicians looking for new musical construction in the vein of Steve Reich, in the 60s, going to Africa to learn rythms with e.g. Nigerians, staying there for years, fully integrated - or so they thought - and bam! for a rare ceremony their hosts suddenly play something they never showed them before and refuse to discuss it afterwards)
Take a look at Super Auto Pets, a pretty successful and fun auto-battler game. It literally uses a free emoji pack for its core art. It doesn't really matter that they didn't hire an artist for those (though I think they did hire an artist after finding success) since a free emoji pack was enough for the creative product they wanted to create. If they had AI generated emoji instead, it wouldn't have really mattered much for the final result (creatively at least, I assume audiences would respond poorly due to GenAI's reputation). At the same time, the ability to create their product without paying a lot of money for artists was critical to make it in the first place.
This is what it means to me to "democratize creativity", to allow creatives to realize their creative ambitions in an area they are proficient at (e.g. video games) without requiring a lot of creative skill in adjacent areas that aren't critical to the experience they are trying to make.
If the art isn't critical to the game, then use simple art. It doesn't matter if the simple art is or isn't AI generated, what matters (in my opinion) is that it doesn't lead us into looking for meaning that isn't there.
However, complex art is needed to fit with genre tropes to attract the expected audience. It's like Apple shoving AI into their products needlessly—not a core part of the experience, but needed so Wall Street doesn't throw a hissy.
> looking for meaning when none is there
Or you generate your own meaning. Art is analysis for the audience.
No, people want to see their ideas come to life, previously this required effort in mastering a skill, now... it takes less and/or different amount of work.
It's like pupils using an LLM to do their homework. They get the grade, but Idiocracy is awaiting.
AI image generation is a gateway drug. You quickly grow resistance, and you will face withdrawal, too.
New tools and techniques that lower the barrier to entry always get over used, over hyped and bring in masses of people making everything from impressive new things to absolute cheap garbage. And then people learn the limitations, the tools get better and a lot of people stop using the tools because they realize there are still skills that go into the process and they aren’t interested in learning those skills. The tools find their place, and new masters rise into the new world. And then a new tool is created and we start the process all over again.
Comparing systems like DALL-E 3 and Midjourney to modern illustration software is a fool's errand. Even with advanced graphic design tools like Photoshop you still have to employ the same artistic skills that have been handed down for generations: 3-point perspective, shading, basic poses, and sketching, etc.
If you can't draw, the fanciest graphic manipulation software isn't going be of much help to you.
As opposed to some doofus who has the ability to type in the words "3D", "trending on artstation", "hyper-realistic,", and "4K" and then proceed to churn out thousands of images in a single day. Stable diffusion is more like the equivalent of having your own personal artist on permanent retainer.
I agree though that schools are even more in trouble now. But my point still stands: we have NO CHOICE. We must adapt, or Idiocracy awaits ;)
Schools definitely have to adapt, but there's so much inertia there, and in many countries, schools have a perverse stimulus to make as many pupils graduate as they can. It'll take another lost generation, I'm afraid. And the makers and exploiters of LLMs can take the blame for that.
I bet a LOT of people would be an artist if there was money in it. Most don’t attempt because of the poor risk/reward of that profession.
That is the whole point of copyright, yes.
> no, you cannot copyright the Studio Ghibli art style itself. Copyright law doesn’t protect styles, techniques, or general aesthetics—like the hand-drawn, watercolor-inspired look with soft colors, detailed natural backgrounds, and whimsical vibes that Studio Ghibli is known for. It only protects specific, original works, like an individual film frame or character design from Spirited Away or My Neighbor Totoro.
And your answer... Doesn't seem to cover my question, I think. So the point of copyright is that the value of the style doesn't degrade?
It's just like an API. The very definition can be changed upstream by patches. People seem to miss that.
But here’s the real meat of the issue I guess, how far do we wanna go? At what point SHOULD something be un-copyrightable? What describes a certain “style”? My style is black&white, can I copyright that?
What would your diff be? :)
If we valued and supported artists without just seeing them as laborers, we could have a remix culture where no one really owns anything and no rights are reserved.
In my view, the people who want to be dropped off at the apex of the mountain the most always dreamed of this outcome for a long time, even if only a little bit. They dreamed of the day they would be freed from the toil of having to study for years and years to produce art that satisfied their tastes, because they would not lower their tastes to make the process less stressful, or had other priorities so could not devote time to practice.
Now the market has innovated and their dream has come true. To people not serious about coming manual artists, there is nothing wrong with this picture. The market need is being fulfilled.
We have to ask where this market need to produce art at such a low level of effort comes from.
In your view, these people have always been cheapskates up in their minds, we just didn't realize it until their preferences were revealed by AI becoming available. If it hadn't, they just wouldn't be artists of any kind and we'd never understand they had any artistic ambitions at all.
At the end of the day, for whatever reason, these people want art they can call their own - it's just a matter of how much effort is required to realize their ambitions.
If these people are to reject AI, yet still care enough about creating art of some kind for a sustained period of time, you basically have to convince them that learning a hard artistic skill is more important than "doing the dishes"/whatever else occupies the rest of their time instead. Maybe a grueling 9-5 work schedule, for example. That is simply how the nature of practice/10000 hours-type advice works out.
If people for some reason just don't want to put in the time, but still want to produce quality art, then they'll choose AI. These two desires are no longer contradicting. They would have been 10 years ago, when you could just retort with "you're going to have to put in the effort, there's no other way." It's clear that that virtue of work-ethic being one's only path to results has been obliterated by AI, and to the new converts it sounds like gatekeeping in hindsight.
Cultivating new interest in learning a skill when it doesn't already exist is way harder than it seems. That gets into mental well-being and existentialist issues that many people in today's society find difficult to reflect about deeply.
I think capitalism has really done a number on people.
I mean, everyone knows it takes a programmer to spec an app. The "different" skill needed for vibe coding is regular old coding skillset. Only difference with image AI is that it doesn't do "vibe drawing" well.
AI art experiment is over. It's been long over, like camera based self driving was by the time some large orgs started embracing the technology. And it's taking longer for some to understand that it's over, just like that time.
And, in some ways he was right... but also not. What eventually happened is that art education that would be required for any aristocrat or an aspiring aristocrat became optional. Most intellectual elite today would be unable to draw much beyond a stick figure diagram. Art, at least where it concerns drawing or painting became a narrowly specialized field. And so, most people today don't understand and don't appreciate art.
But, art didn't die. Instead, artists started asking themselves more questions about the nature and philosophy of art, figuring out which aspects are essential, experimenting more. In retrospect, it's sad that fewer people today have decent access or understanding of art, but while more common before, the understanding was often very superficial anyways.
People want to run marathons on shoes with padded soles and be called an athlete.
Anyway plenty of people can't even finish 1K, so 42.2K in padded shoes is still impressive.
If you wanna draw for the feeling of creative satisfaction that’s great. I just need an image on a page in service of something else.
No shade on stock photographers / illustrators, but that's where I see image generation end up at. And you could already get custom illustrations made for cheap on sites like fiverr. The real long term question will be whether an AI generated image can compete with services like that. (my guess: probably, AI images are higher resolution/quality and generated faster, but I don't know the real total cost of them nor that of cheap illustrators or stock images)
AI is democratizing _execution_, not creativity. The advent of the smartphone camera era allowed for the average person to take vastly more photos than ever before, and yet the photos I personally find noteworthy are either remarkable in their context or aesthetic quality (e.g. [0]) or personally meaningful.
Alternatively put, "culture" values a proof-of-work that is creativity + execution. Taking a photo of a lunar eclipse setting over a crater has global value, because of the effort in planning and taking the shot (read [0], it's great). Me shitposting a meme for my game night groupchat neatly summarizing what happened in our last session of Pandemic: Legacy has local value, because of the creativity in adapting a meme format to our particular context and the execution of the edit, but nobody besides us would find it as funny.
(Interesting aside — I personally think that the photo in [0] is more impressive than someone going all that way to _paint_ the same scene en plein air, because of the split-second precision of the execution. Even though photography may have devalued a lot of the technical execution of capturing a scene, it has opened up new ways in which said execution can be valuable.)
So AI decouples creativity and execution in a way that is, like every innovation before, unlike every innovation before. The Ghibli aesthetic is the soft pastels of the color palette [1] and the comfortable, nostalgic character design [2] but also the restful rhythm of "ma" [3] and the detail lovingly lavished to background characters [4]. GPT-4o devalues some of those things, but it another sense it just re-weights our cultural proof-of-work valuations back towards human effort.
[0] https://lrtimelapse.com/news/total-lunar-eclipse-over-teide-...
[1] https://designmadeinjapan.com/magazine/graphic-design/the-ri...
[2] https://stylecircle.org/2020/09/the-studio-ghibli-closet/
[3] https://screencraft.org/blog/hayao-miyazaki-says-ma-is-an-es...
[4] https://www.rogerebert.com/reviews/great-movie-spirited-away...
Sorry, but most ideas aren't great. It's the painful process of refinement, hard work, and iteration that results in something great.
Many. And I love it when those people code something up with Scratch, Matlab, or even Excel. Even though I personally would dislike the aesthetics of the result, I’m much happier to see other people enjoying computers the same way I enjoy them (i.e., by using them to solve interesting problems) than if I insisted everyone do so using my preferred “real” programming languages.
Regardless of the tool used?
Creative vision is not being realized, it’s being stunted.
I don't think this is very true, I don't often see anyone bragging about their skills and demanding their outputs get put in a gallery and judged on equal merits as the old masters.
I'm not much of an artist and whenever I use an image generator to generate something, I don't do it to show off my artistic talent or whatever - if I was to do it before AI I would've commissioned an artist for it (which I probably wouldn't have done because it was too expensive, so I would forgo it) - the work the artist does would actually have even less of my own input than the AI's, since I'm giving them less description to go off - it's all the artist's, based on their own experiences.
There's nothing wrong with that.
What if a distinct style is not owned by one artist, even if they are most associated with it, but in reality, in art historical reality, is most commonly explored by a group of more and less famous creators, all as part of a movement, at a time? What influences did Ghibli draw on?
Buddhism says all things are essentially empty of their own existence but are merely conditions resulting from other causes, and so on, in a chain unending. Like this.
The pointillism[0] of Impressionism that ended up spread to the masses by countless including Bob Ross is one example of the memetic evolution and transcendent influence of "high art". Greatness can filter down but it doesn't mean it's not great.
An iPhone is still a masterpiece even if you hold it in the ghetto. Does the ghetto pollute it or does it lift the ghetto? Maybe there's cross pollination, but I think it's infected with its greatness.
If it was me I would feel horrible that what I gave to the public and dedicated my life to was contorted in this manner.
[1] https://www.indiewire.com/features/general/hayao-miyazaki-ar...
Reacting to an animation where a gross critter "learned to walk using AI" instead of being animated by a person 8+ years ago, and ended up using its head as a leg
https://www.youtube.com/watch?v=ngZ0K3lWKRc
It has nothing to do with the current image generation topic beyond the "AI" label being stuck on both of them
Which is not to say I expect he's thrilled about ChatGPT cloning the art style on a mass scale, but that quote that everyone keeps reposting doesn't have anything to do with it
He was pretty clearly talking about AI, at least to me.
I mean, the quote might very well reflect his actual views about generative AI, but that's definitely not what he was talking about.
That complaint is just as applicable to current Gen AI models. He wasn’t simply reacting with his gut to a gross looking video but to the concept of a thing with no concept of pain creating and animating artwork of living things. He understood the technology was about Gen AI, as “deep learning” is written on the whiteboard. He deserves some credit.
Watch the video - the purpose of the demo, as the creators explained it, was to train a creature to move quickly. Since the AI model didn't simulate pain it used its head like a foot, and since the result was creepy they thought it could be used for a zombie game. That's what they presented to Miyazaki, and that's what he commented on. Then Suzuki asked where they eventually wanted to end up, and a different presenter said the thing about machines that can draw.
> That complaint is just as applicable to current Gen AI models
If you like, but that's not what Miyazaki applied it to.
Watch the video
the purpose of the demo, as the creators explained it, was to train a creature to move quickly
- "Here, Mr. Miyazaki -- we have made a fast-moving zombie!"
- "I don't like the zombie you have made!"
1. The person who says that wasn't describing the purpose of the demo
2. He says it after Miyazaki's comments, ergo Miyazaki was not commenting on it
> - "Here, Mr. Miyazaki -- we have made a fast-moving zombie!"
Please don't sarcastically put words in the mouth of the person you're replying to - it's rude and it's never useful. All my previous comment did was summarize what was said, in the order it was said. I didn't suggest it was anything more or less than the words in the transcript.
Art is humane. It tells humans how to be humans. A thought about an ill person in pain is worthy of being told as a story. Not only that animation automation thing is of no use to someone trying to express those thoughts, its authors — just like many, many others — have no idea what humans do with their lives, and which tools artists may need to show it. They've made a toy, and were told that it's just useless wanking, together with the whole genres of pointless amusement that introduced such images into pop culture.
“An insult to life itself” is not just a phrase. There is life, and there are people who deliberately ignore it, and enjoy the sights painted on cardboards.
"After seeing a brief demo of a grotesque zombie-esque creature, Miyazaki pauses and says that it reminds him of a friend of his with a disability so severe he can’t even high five. “Thinking of him, I can’t watch this stuff and find [it] interesting. Whoever creates this stuff has no idea what pain is whatsoever. I am utterly disgusted. If you really want to make creepy stuff, you can go ahead and do it. I would never wish to incorporate this technology into my work at all. I strongly feel that this is an insult to life itself.”
He's disgusted by the creature, not the computer based technique. While he's on record as disapproving of CGI, Earwig and the Witch, directed by his son, used CGI so his disapproval isn't absolute.
I think it's clear that he is specifically responding to the the overall soullessness of the technique - to animate without a human understanding of what is being animated. But as others have pointed out this is well before modern AI image gen and I have been corrected in that aspect.
Presenters: "This is a presentation of an artificial intelligence model which learned certain movements [...] It's moving by using its head. It doesn't feel any pain, and has no concept of protecting its head. It uses its head like a leg. This movement is so creepy, and could be applied to zombie video games. An artificial intelligence could present us grotesque movements which we humans can't imagine."
The screen shows some Silent Hill looking vaguely humanoid, crawling blob. As the presenters say, it's pretty creepy looking.
Miyazaki: "I am utterly disgusted [...] I would never wish to incorporate this technology into my work at all"
IMHO saying Miyazaki outright hates AI is putting words into his mouth. All the clip shows is that a dude that doesn't make zombie horror films doesn't need a zombie horror generator thank you very much.
So yeah, he clearly rejects the product pitch. But judging from Kiki's Delivery Service and My Neighbor Totoro I don't see why you'd pitch him that product.
"Would you?"
"Yes"
Awkward silence
From this I don't think it's difficult to extrapolate his feelings about modern AI image gen. But you are correct in that this is not a direct assessment. Appreciate the correction, thanks.
It's in the article
the quote in context - https://youtu.be/ngZ0K3lWKRc?si=gw-_z17n_XWfqzcQ
The USSR had some absolutely incredible art used on their propaganda posters. But if you use that for anything outside of ironic Russian themes (e.g. a goofy game set in that era, some silly snack that claims to be Russian-inspired), people will think you're an unironic communist and you instantly turn away loads of people.
When a dominant political force uses AI "art" for everything they do and the style becomes apparent, anything that looks similar to that instantly disgusts loads of people. You can argue "well that's their fault for being disgusted", but pattern recognition and being conditioned to associate certain images with "bad" goes far deeper than the conscious and it's a base instinct in animals.
People know the "NFT style", and when people see any sort of image like that, they instantly think "annoying crypto scammer."
There's now the "AI image" style, which people are becoming more aware of. And people associate that with scammers, very confused old people, weird porn, and the right.
Most of the ghiblified stuff is from AI/cryptobros and right wing movements (with heavy overlap on those groups). People possess pattern recognition and will become aware of the pattern each time they see these images.
and that the model surely used his films as data.
I mean, I have no doubts. But proving it seems tough!
What is OpenAI in all this, if not a greedy, sloppy, soulless outsider stealing their Art and effort for financial gain without ever asking for permission?
I pointed out a reason why litigation could be difficult. I'm quite sure nothing I wrote could have been seen as defending OpenAI. Just that I felt litigation would be tricky.
But you don't want any of that. You want to have a familiar pop cultural label (“Miyazaki”) that produces a familiar reaction (“Oooh!”). Purely decorative, symbolic objects. Stories, ideas, hard work? Eh, don't bother me with that nonsense.
There is nothing new or “cutting edge” in ignorance. And AI companies know perfectly well that they work for exactly that audience. Despite all the talk, they don't create the next genius artist, they want to be a next “enhancement filter” in TVs, something that no one uses, but everyone has to add to impress the public. That's just parasitism on lack of ability to discern.
How much of it is _his_ style and _his_ art?
How many people work on the frames and animation?
On twitter maybe but on places like localllama google Gemini got way more airtime than this drama
Capturing mindshare matters no matter how you do it. I would bet my life savings that more people will join OpenAI to work on their models as a result of the Ghibli moment than will join Google as a result of this particular (incredibly impressive!) Gemini iteration.
Because we’re humans! And we’re all pretty easily impressed, even those of us who actually build these dang things.
No, frontier model researchers are not learning about the concept of AI from Twitter.
But it’s one more piece of visibility, one more sense of “hey this company is actually doing stuff with their models,” that absolutely contributes to where these insanely in-demand people choose to do their work.
> Leading edge chip designers at TSMC didn't end up there because they ended up playing graphic intensive video games
What a baffling and wrong example! I mean, I guess I don’t know about TSMC specifically, but tons and tons of technologists have an origin story that rhymes with “I was really into computer games, then got curious how they worked”
This is my whole point: engineers are real people, and real people are fascinated by the uses of AI!
Leading edge chip designers at TSMC didn't end up there because they ended up playing graphic intensive video games
Kind of a large assumption. I find it quite plausible that enjoying video games as a kid might lead to someone studying engineering as opposed to medicine or finance.
I feel like you're overestimating that. I work every day on developing LLM-based products, use them a lot while developing those, and have been doing so since November 2022 (beta release of GPT 3.5). So I should be a prime representative of the demographic you're talking about.
When a new frontier model is released, I run a series of tests, play around with it, and decide whether or not to use it based on that. Anything else would be ridiculously stupid, leaving either cost savings, product quality improvements, or development velocity increases on the table. We're not talking about databases or cloud VMs where there's a big switching cost. It's changing a model string and provider, so frictionless that it's a no-brainer.
We actually have some data that's on my side. Take a look at OpenRouter's model usage statistics. Note how OpenAI mmodels make up a tiny share on there, despite having more general mindshare on social media than all other model providers combined. Their data also shows people are very quick to switch when improvements to price/performance are released.
It's a bit like how vacuum cleaner specialists never use Dysons despite them having the most mindshare and overwhelming marketing budget. Regardless of vacuum cleaner specialists also just being normal people. Even if this one doesn't hold in your region, there's countless of these examples.
Yes, but Dyson still crushes the market because it has the mindshare with non-specialists. That’s OPs point, uninformed people (the vast majority) will be impacted more by the “ghibli-core” memetic moment, and have a greater impact on future events than specialists who appreciate the Gemini 2.5 release
the fact that vast majority of people are consumers matter
Which probably is just a small integer multiple of two weeks, but hey!
https://www.deeplearning.ai/the-batch/japan-ai-data-laws-exp...
Otherwise: Meet my new and surprisingly performant new model, IdentityFunctionGPT! Oh, the weights? Sorry, trade secret.
However, if I make a drawing in the style of Ghibli, I own the copyright because I made it. I don't know if in Japan, art style is copyrightable.
I don't know if copyright applies to generated content (of any sort) though. It probably does.
* Unsurprisingly a famous exception involves Nintendo
https://en.wikipedia.org/wiki/Pok%C3%A9mon_doujinshi_inciden...
But I do feel bad for partaking in part for Louis CK's Twitter-Jesus rant.
I fail to see how taking a distinctive artistic style that was incredibly difficult to produce and shitting out massive amounts of it everywhere as a super low quality commodity would pass the test of fair use.
"artistic style" is outright not copyrightable. Fair use doesn't play into it, any more than fair use doesn't play into whether a photo taken by a monkey can be redistributed[1].
[1] https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...
Also, Disney have deep pockets and Etsy would not want to argue it out with them if they got a complaint.
Not under US copyright law, AFAICT. He's won cases where he sued over things like false endorsement - i.e. claiming that listeners would believe it was actually him, not just that the style was similar.
(Apparently he did win a similar case in Spain under their copyright law, but from skimming articles it sounds like the issue there also was impersonation, not just stylistic similarity.)
- When does generative AI qualify for fair use?:
https://suchir.net/fair_use.html
- AI is currently just glorified compression:
AI can take a whole style wholesale with no effort and no contribution. It is greivous theft in the moral sense if not the legal sense.
>usher in an internet age where everybody can access the entire corpus of evolved knowledge
We have had that for decades and everything is going to shit. People now just reject facts.
right, because before copyright laws nobody ever created anything...
People consuming and copying your work means you are making a cultural impact
You think the only reason to create anything is to profit?
If this take is serious, I can confidently tell you: for many professional artists, it's incorrect.
I'm sure you didn't look at my profile and don't know my background, but I'm a professional bluegrass musician.
Believe me when I tell you: few of us believe that 'intellectual property' is designed to help us make a living. And we're happy to see it go. Copy my music all you want - that's half the point.
Creativity, especially in traditional arts, does not benefit from state intervention or shoehorning property concepts into our work.
Marvin Gaye estate tried to argue that style is copyrightable, and sued some artists as their songs use similar chord progressions as Marvin Gaye songs; they won one case (which I won't google) but they lost with Sheeran.
This is the tricky part; is it immoral? I think so, but that's a personal opinion. Is it illegal? Not by Japanese copyright laws.
The takeaway? If you’re launching a consumer product, technical superiority doesn’t automatically translate to cultural impact. In the battle for attention, a memorable vibe often beats a better benchmark.The biggest difference to what was previously available is the accuracy - especially the text. This opens up a plethora of possibilities.
It's chocked full of memes, historical and political photos, and general pop culture.
Could it be that all the "AI" evangelists are in their own bubble and desperately trying to make money off each other?
Like when mobile free to play gambling apps ("games" they said) had ads, but only for other mobile free to play gambling apps.
The White House twitter account: https://x.com/whitehouse/status/1905332049021415862
But seeing a beautiful, whimsical image of my baby daughter in the Ghibli style was pure joy and brought tears to my eyes.
I have no idea how I could have done this otherwise, and I hope it brings happiness to Miyazaki to know that it brought joy to someone.
This is highly personal though, and if it works for you, then you have one of the best reasons for using this sort of image generation -- to create memories and experiences with your family.
With Ghibli there is the fact that the original Italian word is pronounced with a hard G. In fact, that’s why it has an “h”. We can treat this similar to Roma vs. Rome.
it does make me uneasy if OpenAI charges more than at-cost to help people do so though. it's the profiteering off others' work that is gross.
a quote from "The Communist Manifesto" about the nature of capitalism:
"Constant revolutionising of production, uninterrupted disturbance of all social conditions, everlasting uncertainty and agitation distinguish the bourgeois epoch from all earlier ones. All fixed, fast-frozen relations, with their train of ancient and venerable prejudices and opinions, are swept away, all new-formed ones become antiquated before they can ossify. All that is solid melts into air, all that is holy is profaned, and man is at last compelled to face with sober senses his real conditions of life, and his relations with his kind."
I agree. But people aren't making art in a style they love, people are telling a machine to make an inferior copy of something resembling art in a style they claim to love.
I say "claim to" because I find it hard to believe someone could be a fan of Ghibli and treat the studio's work in a way that Miyazaki would despise. This isn't love so much as exploitation.
If people value the Ghibli style so much, they could learn the art of drawing and animation, and actually make art instead of insulting the studio's talent and legacy by avoiding the craft and seeking an endorphine high from AI slop. It doesn't have to be perfect or even good to be satisfying, but it will be yours.
My family and I absolutely love Ghibli films. I ran some family photos through 4o and my wife was thrilled. It made her day.
You can call it slop, you can say I'm not truly a fan, but I made my wife happy, and I can't see the harm.
How is it "insulting their legacy"? You think people writing Harry Potter fan fiction are insulting J.K. Rowling?
No one cares about your family photos. This isn't about you.
My wife cares. That was the point of my comment.
You're like the person who comes into threads about gun violence and says they don't see what the problem is, since they only shoot watermelons in their backyard.
Like, OK. You refuse to recognize the bigger issues at play, noted.
And so the problem isn't with people who find joy in something they saw, but in people who make money from derivative work without explicit permission
If the technology stays, the commercial incentives to produce art with such a depth and craftsmanship will cease to happen (unless perhaps new styles would be generated to feed an LLM as part of an art pipeline? Maybe? Wishful thinking most likely). You won't be able to stand up a company like Studio Ghibli in the new market. You'll have a harder time even standing up a career as an artist in general, let along getting enough to come together to form a studio. This will be the last time art creation like this happens. You may love Ghibli, but this will kill the process the allowed Ghibli to form.
---
Spread across all such art styles, art itself no longer signals any quality of the minds or depth of effort behind its creation. We will have to look elsewhere for things to love and that have meaning.
So art ended when someone created a photo camera?
Here's the oft repeated analogies that never have a proper response to hopefully help guide you out of your self created mental labyrinth, but trigger warning! (for people like you): photographers, printing press, classical music.
Did you get ChatGPT to write this copypasta for you?
i think they shouldn't be able to sell anything at all. violation of copyright isn't a excuse to sell stuff at-cost, let alone feed Nvidia with trillions of USD by releasing open-source (if they ever release) technology
It's not making, creating. It's consuming.
Wrt. the Communist Manifesto: it's precisely the bourgeoisie and the church who are responsible for the greatest art, and its development. As you say, the workers love kitsch.
And let's be honest: communist art is generally rather poor. The state, or the avant-garde of the proletariat for that matter, does not have good taste, but does tend to towards nepotism.
This type of quality is close to enabling a storyboarder like Miyazaki do an entire animation on their own. I have yet to hear him say he would not use such technology, despite him being rather opinionated.
They are like the Mafia.
“I don’t care, pay me”
> I strongly feel that this is an insult to life itself
I emphatically appreciate Miyazaki. Even if you take him as hyperbolic, or perhaps biased due to differences in intention between the engineer team and artistry he represents, it's very positive to have strongly opinionated voices in the room that can clearly articulate why. The anecdote about pain and stiffness is succinct but exactly what is necessary to communicate specifically what is missing from the animations at hand.
However, I'm curious what he would have thought about someone playing QWOP. I wonder if he would be more fully bought into such a concept (well outside his own work of course) if such technology might emphasize appreciation for life rather than trying to mimic/bastardize/reproduce it.
Unfortunately I'm sure this sensationalised fake-quote will spread much more quickly than any context / checking.
Hayao Miyazaki is just calling them out in a diplomatic way.
The "insult to life" is specifically about the tech, from my reading. Granted, if he had been shown something less grotesque, maybe he would've responded differently? I doubt it, the guy doesn't like CGI either.
This just looks like some painful RL or evolutionary algorithms, at best.
A common meme is to put him side to side with Junji Ito. Hayao Miyazaki as the grumpy old man whose work is cute and happy, and Junji Ito as the nice and cheerful guy whose work is the thing of nightmares.
Junji Ito's take on AI, is that he worries for his profession, but he thinks he still has something AIs can't replicate, at least for now. As expected, while both of them oppose AI, they do it in a way that couldn't be more different.
Hayao Miyazaki disliking AI is just Hayao Miyazaki being himself. Had he been enthusiastic about it, it would have been very noteworthy.
It can probably be modeled using Bayesian statistics or something.
Just think of using this type of movement or actions in media specially a film. That would be entirely unworkable and doomed idea. Disgust from most viewers would be similar.