Ever since OpenAI launched ChatGPT 4o image generation featureAfterwards, everyone is probably playing with it, seeing what other styles of images can be generated. Actually, besides ChatGPT, Grok can also generate Ghibli-style images, but the results were really poor before, so I didn’t introduce it. But recently, perhaps feeling threatened, Grok improved its image generation capabilities. After testing, the Ghibli-style images Grok generates are much better than before, and regular members have no limits on the number of generations. If you want to know how Grok’s image generation looks now, keep reading!
Grok’s image generation just got upgraded! Ghibli-style images have taken a major leap in quality, and there’s no usage limit.
Everyone’s attention lately has probably been on what you can do with ChatGPT’s image generation feature, but actually, not long after ChatGPT 4o rolled out image generation, Musk reposted a post about Grok being able to generate anime-style images too (though the results were pretty terrible). At the time, our editor also tested it out, and the results really were awful, plus they were stingy with the generation limits—only three images a day, which made it completely impossible to share anything.
— Elon Musk (@elonmusk) April 2, 2025
However, I recently re-tested it and found that Grok now generates better results across various styles. So far in testing, there are no limits on the number of images you can generate. That’s why I thought it was worth sharing. Using Grok to generate images is simple: you can click Grok on X or open the Grok mobile app directly.
Then click “Link” to upload the image you want to restyle, type “draw me in Ghibli style,” hit send, and Grok will generate the image for you. Grok creates two images at a time, so you can pick the one you like better. Or click “Edit” to have Grok modify the parts of the image you’re not happy with.
Everyone can take a look at the same photo and see the difference between the one generated on 4/1 and the one generated today. Although there are still some flaws, it’s already much better compared to before.
Although the character generation results are still somewhat lacking, the pet generation results are actually pretty good. Everyone can take a look at these Ghibli-style cat photos generated with Grok.
Here’s a comparison of the images generated by Grok and ChatGPT using the same cat photo and the same prompt. Which one do you all think is better?
Of course, Grok’s Ghibli-style image generation isn’t always perfect either. For example, with more complex photos or photos with more than two people, the results aren’t good. For instance, even if it’s just one person taking a photo with an elephant, Grok will suddenly add a second person. Or with a two-person group photo, Grok’s results are also poor and it can’t capture the subjects’ features. However, since Grok currently has no limit on image generation attempts, you can just keep having it redo them until you’re satisfied.
Although Grok has improved at generating Ghibli-style images, its results for other styles are still not good enough. Here are some images generated with Grok in the styles of Doraemon, Snoopy, and Animal Crossing. If you want to generate images in these styles, I’d recommend using ChatGPT instead for better results. Below are links to ChatGPT-generated Animal Crossing, Pixar, and Snoopy characters, along with the full prompts included in the article. If you’re interested, feel free to click and read:
Source: KOCPC Chinese