Ghibli Style Image Generator

Unless you've been living under a rock for the past week, you've likely heard that OpenAI has launched their new ChatGPT image generation model. The 4o image generation model was launched on March 25th, 2025, with the announcement posted on their website.

As the leading AI API provider, we at PiAPI already have our very own GPT-4o Image API.

And we have to say—the 4o image generation by ChatGPT looks very impressive, easily overshadowing most text-to-image and image-to-image AIs in the market.

In this blog, we'll be generating both text-to-image and image-to-image examples in the popular Studio Ghibli style. Many people have been using the new ChatGPT 4o image generation model as their own personal Ghibli-style generator, and it's been really fun to watch! So we decided to test our own GPT-4o Image API this way and see how it performs.

Examples

Trump Tariffs

For our first example, we imagine what the recent Trump Tariffs announcement might look like in the world of Studio Ghibli. The image below is used as input, along with an appropriate prompt.

And here it is after being "Ghibli-fied" by ChatGPT 4o:

Prompt:

can you recreate this image into studio ghibli style

We must say—it looks amazing. Even though it didn't get all the numbers correct (especially at the bottom), it demonstrates impressive text adherence, far better than most models on the market.

A Minecraft Movie

Next, we take Jack Black as Steve from the Minecraft movie and pass the image through our ChatGPT 4o Image API. What would a Minecraft movie look like if created by Studio Ghibli?

And here's the result:

Prompt:

can you recreate this image into studio ghibli style

This one also turned out great! The image quality, style consistency, and text adherence are all top-tier. It also doesn't struggle with hands or fingers, which is always a plus.

Another impressive result from ChatGPT 4o.

Comic Panel

For our final test, we create an AI-generated comic panel, something that has become increasingly popular on Reddit. This time we use a pure text-to-image prompt:

Can you create a comic panel with 4 scenes, make the entire comic in studio ghibli style:

the first scene (top left) is a guy looking smug and pointing at two guys talking (the two guys talking have the openAI logo on their shirts), the smug looking guy says that "AI art is not art"

the second scene (top right) is just the two guys with the openAI logo on their shirts looking back at the smug looking guy awkwardly

the third scene (bottom left) is one of the two guys giving the smug looking guys a thumbs up before returning to their conversation

the fourth scene (bottom right) is the smug looking guy now angry, shouting at the two guys who are now ignoring him, the smug guy is shouting "STOP HAVING FUN"

And here's the result:

This one turned out especially well! It nails:

• Human expressions
• The OpenAI logo
• Strong text adherence
• Ghibli-style composition

It truly feels like you could generate an entire manga using only our GPT-4o Image API. The new ChatGPT 4o image generation represents a huge leap forward.

Conclusion

It's clear that the new ChatGPT 4o Image Generation is outstanding. It can accurately recreate a wide variety of art styles—including the Ghibli style featured in these examples.

We're excited to see how OpenAI continues to improve ChatGPT image generation and the GPT model family in general!

If you want to try it yourself, you can use our GPT-4o Image API today.

And if you're interested, be sure to check out our other Generative AI APIs as well.