In late March 2025, OpenAI rolled out native image generation capabilities in ChatGPT's GPT-4o model. Within hours, someone uploaded a family photo and asked ChatGPT to transform it into the style of a Studio Ghibli film. The internet lost its collective mind. The resulting images, soft watercolor palettes, whimsical backgrounds, those impossibly warm Ghibli eyes, looked stunning. People flooded the platform with their own photos, their pets, their food, their houses, their memes, all "Ghiblified."
The trend was so massive that it crashed ChatGPT's servers, gained OpenAI a million new users within a single hour, and pushed the platform past 150 million total users. Sam Altman, OpenAI's CEO, jokingly posted on X asking users to "chill" because his team needed sleep. OpenAI temporarily restricted free users to three image generations per day to stabilize the servers.
A year later, the Ghibli aesthetic remains one of the most popular AI art styles. And the ethical questions it raised have only gotten louder. Let me walk you through how the trend works, the best prompts to use, what Miyazaki and Studio Ghibli actually said about it, and whether we should feel weird about doing it.
The spark was genuinely organic. When OpenAI launched GPT-4o's image generation on March 25, 2025, a Seattle software engineer named as the first viral poster uploaded a family photo and asked it to be redrawn in Ghibli style. The result was immediately recognizable: that distinctive Studio Ghibli warmth, the slightly rounded features, the dreamy backgrounds that look like watercolor paintings come to life.
The trend spread with the kind of velocity you rarely see. It hit every social platform simultaneously. People were not just generating random Ghibli scenes. They were "Ghiblifying" themselves, their kids, their pets, their wedding photos, famous memes, historical figures, politicians. Sam Altman and other OpenAI employees jumped on the trend and posted their own Ghiblified images, which accelerated it further.
What made it go viral specifically was the combination of nostalgia and personalization. Studio Ghibli films are beloved worldwide, and people have a deep emotional connection to that visual style. Being able to see yourself rendered in that aesthetic hit a very specific emotional nerve. It felt personal in a way that most AI art trends do not.
The simplest method, and the one that started the whole trend, uses ChatGPT directly. You need a ChatGPT Plus subscription ($20/month) to access the image generation features. Free users get limited generations.
The basic process is almost comically simple. Upload a photo to ChatGPT and say something like "Restyle this image in the style of a Studio Ghibli film." ChatGPT handles the rest. But if you want better, more specific results, you need to be more deliberate with your prompts.
ChatGPT is the easiest route, but Midjourney and Flux-based models offer more creative control for experienced users.
Midjourney is particularly effective with its Niji model (version 5 and above), which is specifically designed for anime-style art. The key technique in Midjourney is using the --sref (style reference) parameter with a screenshot from an actual Ghibli film, then combining it with your text prompt. This gives the model a concrete visual reference rather than relying entirely on the words "Studio Ghibli." You can also adjust the style weight with --sw to control how strongly the Ghibli aesthetic dominates versus your text prompt.
Flux and Stable Diffusion users have a different approach. The community has created dedicated LoRA models specifically trained on Ghibli film frames. These are downloadable from Civitai and can be loaded into ComfyUI or Forge alongside your base model. A Ghibli LoRA plus a quality base model like Juggernaut XL or Pony Diffusion v6 produces results that are often more faithful to the actual Ghibli aesthetic than ChatGPT's interpretation, because the LoRA was trained specifically on those film frames.
This is where things get complicated, and where the fun of making Ghibli art collides with some uncomfortable realities.
"I strongly feel that this is an insult to life itself."
That quote went massively viral during the March 2025 Ghibli trend, and for good reason. Miyazaki, now 85, has been one of the most vocal critics of AI in the creative arts. During a 2016 demonstration of an AI animation system, he watched an AI-generated creature drag itself across the floor and was visibly disturbed, calling himself "utterly disgusted" by the demonstration. His philosophy of animation is deeply rooted in the belief that art must come from the human spirit, from an animator's personal connection to life and movement.
Studio Ghibli has not issued a formal legal challenge against OpenAI specifically over the Ghibli trend, but the studio's position is clear through broader action. In November 2025, Studio Ghibli joined a coalition of Japanese publishers demanding that OpenAI stop training its models on their copyrighted work. Japan's Content Overseas Distribution Association (CODA) has stated that under Japan's copyright system, "prior permission is generally required for the use of copyrighted works," suggesting that the Ghiblification trend could constitute copyright infringement under Japanese law.
The legal landscape is still evolving. No court has definitively ruled on whether AI-generated images "in the style of" a specific artist or studio constitute infringement. But the trend has become one of the most prominent examples in the ongoing debate about AI and creative rights.
If you love the Ghibli aesthetic, you will probably also enjoy these related styles that produce beautiful results with AI image generators.
Makoto Shinkai style (Your Name, Weathering with You). Known for hyper-detailed backgrounds, especially skies. Use prompts referencing "Makoto Shinkai style, dramatic clouds, golden hour light rays, ultra-detailed cityscape backgrounds, vibrant blue and orange color palette." The signature look is photorealistic backgrounds with slightly simplified anime characters in front of them.
Satoshi Kon style (Perfect Blue, Paprika). A darker, more psychologically complex aesthetic with realistic proportions and surreal visual elements. Try "Satoshi Kon animation style, realistic character proportions, psychological atmosphere, surreal visual transitions, muted color palette with sudden vivid accents."
Watercolor anime is a broader style category that captures the softer, hand-painted feel of older anime without referencing a specific studio. Prompts like "soft watercolor anime illustration, visible paper texture, gentle color bleeding between areas, pastel palette, delicate linework" produce gorgeous results that feel handmade without invoking a specific artist's copyrighted style.
Lo-fi anime aesthetic has become its own distinct category, inspired by the lo-fi hip hop study girl and similar nostalgic, cozy anime imagery. Try "lo-fi anime aesthetic, warm interior lighting, cozy room filled with plants and books, character studying at a desk, rain on the window, soft ambient glow, muted retro colors."
I am not going to pretend this is simple, because it is not. On one hand, artists have always been influenced by other artists. Art students study the masters. Painters reference their predecessors. "In the style of" has been part of creative practice for centuries. AI generating an image "in the style of Studio Ghibli" is, in some ways, doing what humans have always done, just at scale and at speed.
On the other hand, there is a meaningful difference between a human artist spending years absorbing Ghibli's influence and developing their own interpretation of it versus an AI model that was trained on Ghibli's actual copyrighted frames and can now mass-produce approximations of that style in seconds. The human artist transforms the influence through their own experience. The AI extracts and replicates patterns from specific training data.
Miyazaki's emotional reaction, calling it "an insult to life itself," resonates because it touches on something real. When every person on the internet can generate a passable Ghibli scene in 10 seconds, it does something to the perceived value of the thousands of hours that Ghibli's artists spent hand-painting each frame. Whether that "something" is devaluation, democratization, or both depends entirely on your perspective.
My personal take? Use the tools. Experiment. Create art that makes you happy. But also pay attention to how the legal and ethical landscape evolves. Support the original artists and studios whose work made these AI capabilities possible. Watch the actual Ghibli films. Buy the actual merchandise. And be honest about what AI is doing when it replicates a style: it is not creating from the spirit, as Miyazaki demands. It is pattern matching. Those are very different things, even when the output looks similar.
The Ghibli trend is not going away. The questions it raised are not going away either. How we navigate both will say a lot about what kind of creative future we are building.