The Internet has become a Ghibli-inspired studio.
Japanese animation studio praisers are using the new updates to ChatGpt to create portraits and memes inspired by the distinctive art style popularized by filmmaker and studio co-creator Hayao Miyazaki.
This trend appeared within a day of Openai introducing new native image generation capabilities in the latest version of ChatGPT, GPT-4O on Tuesday. Users paying premium tier payments starting at $20 a month can upload any image to the system and “convert to Studio Ghibli Style” into the tool. Users can also use the tool using SORA, CHATGPT’s Text-to-Video AI model.
Some of the most shared images posted on social media on Wednesday recreated political events in US history, including President Donald Trump’s meeting with Ukrainian President Voldy Mie Zelensky and the assassination of President John F. Kennedy. Others have changed popular memes, including a recent one that mentioned fitness influencer Ashton Hall’s morning routine and a distracted boyfriend meme from a decade ago.
However, many people seem to have come out of the Ghibli movie, using it to make their favorite photos with their family, friends and pets.
Openai CEO Sam Altman also joined the trend and changed his X profile photo to the same studio Ghibli style. On Thursday, he shared that “our GPU is melting” due to the demand for these images.
“We’re going to temporarily introduce some rate limits while we’re working to make it more efficient. It won’t work!” he wrote to X, adding that the free tier for ChatGpt will soon be three generations a day. He also pointed out that this model rejects generations that should be allowed. This is an issue the company is working on fixing.
ChatGpt has released its feature amid continuing concerns from artists about the invasion of generative artificial intelligence technologies, and amid continuing concerns about all artistic mediums, including images, voices, and music.
At the end of 2023, the widely distributed list revealed the names of thousands of artists who were used without consent to train popular AI art generator Midjourney. Last year, more than 11,000 creatives signed an open letter denounced the use of human art to train AI without permission.
And last month, thousands of artists likewise signed an open letter to Christie’s New York Gallery to abolish AI art auctions.
“Many of the artwork that we are planning to auction was created using AI models known to be trained in copyrighted work without a license,” the letter states. “These models and the companies behind them will misuse human artists and use their works without permission to build commercial AI products that compete with them.”
While many social media users were excited to use the new AI tool, others lamented that its use would reduce Miyazaki’s work.
Some called the reimagined art an AI slop. This is a term used to describe images that are often dominant, unusually generated, which dominates social media feeds. Others called the images “soulless.”
“What I saw today is the very definition of AI slop,” X-user wrote. “Flood feeds that knock off the richness of glybri. Art has reduced the ‘content’. Unique designs are meme-generated. Zero’s creativity. ”
Openai, like many of its competitors, does not reveal the specific data used to train the model. However, the accuracy of Studio Ghibli’s impressions recreated user questions about whether they had shattered studio pieces without permission.
An Openai spokesman said in a statement Wednesday that the company’s goal is to “give users as much creative freedom as possible.”
“We continue to prevent generations in the style of individual living artists, but allow for a wider range of studio styles. This generates and shares original fan work that is truly fun and inspired,” the spokesman wrote in an email. “We’re constantly learning from real-world use and feedback and will continue to refine our policies as we go.”
Studio Ghibli did not immediately respond to requests for comment.
Openai has been sued by several news outlets, authors and visual artists claiming copyright infringement, but this month it says to the US government that AI companies need to strengthen their global American lead to advancement in technology, making it easier for them to learn from copyrighted materials.
The company claims that modern image generation tools are excellent at closely tracking user prompts, especially those of users, and accurately rendering text using the context of the chat.
“These features help you create the images you imagine accurately, communicate more effectively through visuals, and move image generation into practical tools with accuracy and power,” the company announced in its blog on Tuesday.
The legendary animator Miyazaki was previously known as AI-generated art.
“Nobody who made it thinks of pain. It’s very uncomfortable,” he said in a 2016 AI demonstration, according to a clip from the moment that circulated through social media on Wednesday. “You can make something scary if you want, but I have nothing to do with that.”
