AI総合研究所

SHARE

X(twiiter)にポストFacebookに投稿はてなブックマークに登録URLをコピー

Ghibli-Style AI Art with ChatGPT|Free Tool to Try the Latest Ghibli AI Trend

この記事のポイント

  • GPT-4o enables high-quality image generation using an autoregressive model
  • You can create Ghibli-style images directly in ChatGPT with simple instructions
  • The free plan allows up to three image generations per day
  • Common errors can be avoided by adjusting prompts and image size
  • Be mindful of copyright and ethical considerations when generating images
Shoma Sakamoto

監修者プロフィール

Shoma Sakamoto

XでフォローフォローするMicrosoftMVP

Founder of LinkX Japan and Microsoft AI Partner.Holds a Master’s degree in Technology and Innovation Management from Tokyo Institute of Technology, with research focused on natural language processing and financial engineering. Formerly engaged in AI and blockchain research at NHK Science & Technology Research Laboratories. Presented at academic conferences, published in international journals, and received the Outstanding Paper Award at the Japan Society for Management Information (JASMIN) National Conference. Previously founded and managed IT and Web3 ventures in Singapore before establishing LinkX Japan.

In March 2025, OpenAI released its latest model, GPT-4o, which introduced not only advanced text generation but also powerful image generation capabilities—capturing widespread attention.
One of the most talked-about features is the ability to generate Ghibli-style illustrations.
With ChatGPT, you can simply upload a photo and give a short prompt to create dreamy, animated images reminiscent of Studio Ghibli films.
In this article, we’ll walk you through the actual process we tried, share impressions of the output, highlight common errors to avoid, and explain key copyright and ethical considerations

What Is Ghibli-Style Image Generation with ChatGPT? A Guide to GPT-4o's AI Image Features

In March 2025, OpenAI launched GPT-4o, introducing groundbreaking image generation capabilities alongside its powerful text functions. Among the most talked-about features is the ability to create “Ghibli-style” illustrations—a trend often referred to as Ghiblification.

This “Ghibli-style” refers to visuals inspired by Studio Ghibli's signature look: soft color palettes, warm landscapes, and anime-like character designs. Simply by uploading an image and typing "Make it Ghibli-style," ChatGPT uses its trained model to apply this aesthetic and transform your photo into a charming, animated-style illustration.

This powerful style transfer is made possible by a new image generation engine integrated into GPT-4o.

👉 Learn more about how GPT-4o generates images:
What is GPT-4o’s Image Generation? Features, Pricing, and Prompt Tips


Key Features of ChatGPT’s Image Generation

Here are the main characteristics that make ChatGPT’s image generation stand out:

  • Autoregressive Model
    Produces more coherent compositions and consistent color tones compared to diffusion-based models.

  • Advanced Text-to-Image Alignment
    Simple prompts like "Make this Ghibli-style" are accurately interpreted and applied.

  • All-in-One Chat Interface
    Upload, instruct, and generate—all within the ChatGPT UI.

  • Available to Free Users
    Image generation is available even on the free plan (with daily limits).


How to Generate Ghibli-Style Images with GPT-4o

We tested GPT-4o to create an actual Ghibli-style illustration. Here’s a step-by-step guide:

🔧 Step-by-Step Guide

  1. Open ChatGPT (GPT-4o Model)
    Visit chat.openai.com and select the GPT-4o model. Image generation is available on both desktop and mobile, even for free users (limits apply).

  2. Upload an Image
    Click the “+” icon at the bottom of the screen and upload your photo—portrait, landscape, etc. JPEG or PNG under 2MB is recommended.

prompt_example
prompt_example

  1. Type a Prompt
    Enter a prompt like “Please make this image Ghibli-style.” Be as specific as you'd like.

  2. Wait for AI to Generate
    GPT-4o will process the image and return the result in 1–3 minutes.

Result
Result example

  1. Check and Save the Result
    You’ll see the generated image right in the chat. You can save it for personal use.

How to Edit the Generated Image

You can tweak the AI-generated image further:

  1. Click the Area You Want to Edit

    Edit

  2. Give a New Prompt for That Area
    For example: “Please make the stomach pink.”

White belly prompt
White belly prompt

ChatGPT will apply your edits based on your feedback.


Better Quality with the New o3 Model

OpenAI’s latest model, o3, significantly improves image generation quality. Previously, text within generated images often appeared distorted—but o3 makes them cleaner and more natural.

o3 sample

You can even ask where a photo was taken, and o3 will try to guess:

o3 location guess

👉 Want to learn more?
What Is OpenAI o3? Features, Pricing, and Comparison with o4-mini


Other Styles: AI “Personification” & Bikkuriman Stickers

Beyond Ghibli-style, ChatGPT can generate other fun image styles:

  • Bikkuriman Sticker Style – Retro anime sticker look
  • Personification (Anthropomorphization) – Turn pets or objects into people

Bikkuriman-Style Example

Input

Prompt: “Make it like a shiny, rare Bikkuriman card with an angelic prince vibe.”

Output

👉 Tutorial: How to Generate Bikkuriman-Style Images Using ChatGPT

Personification Example

Dog to boy

You can even personify a dog into a boy—or a middle-aged man, like this:

Older character

👉 More: How to Personify Animals with ChatGPT (Free Guide)


Troubleshooting: Common Errors and Fixes

Error Cause Solution
Upload fails File too large Resize your image
Style doesn’t apply Prompt too vague Use clear style directives
Generation stalls Server overload Try during off-peak hours
Can I use it commercially? Copyright concerns Use only for personal, non-commercial purposes

Image generation is also subject to usage limits (e.g., 3 images/day on the free plan).

limit_message
limit_message

👉 Details: ChatGPT Usage Limits and How to Avoid Them


Prompt Tips for Ghibli-Style Image Generation

  • Be Specific:
    Instead of “Make it Ghibli-style,” try “Ghibli-style forest background with warm tones.”

  • Use Reference Styles:
    “Ghibli-style” + “like Totoro’s forest” helps guide the AI better.

  • Resize Images if Needed:
    Try 800x600px for smooth performance.

  • Give Feedback:
    Comments like “Make it brighter” help fine-tune the output.

  • Experiment!
    Different prompts = different results. Try “Ghibli-style night scene” or “Ghibli-style character portrait.”


While Ghibli-style generation is fun, be cautious of copyright risks:

  • Avoid mimicking exact characters or settings
  • Use for personal enjoyment only
  • Remember that Studio Ghibli’s creators, including Hayao Miyazaki, have expressed caution toward AI art

🔗 OpenAI Image Generation System Card


In Japan, copyright law around AI-generated content is still developing.

🧠 According to the Agency for Cultural Affairs:
If a human enters prompts with creative intent, they may be recognized as the author of the AI-generated work.

📜 In April 2025, a government official stated:

“Mimicking a style alone doesn’t constitute copyright infringement. However, if an AI output closely resembles specific characters or settings, legal risks remain.”

This was in response to a question about Ghibli-style images in the Japanese parliament.


10 Use Cases for Ghibli-Style Image Generation

This feature has gone viral on social media! Popular use cases include:

  1. Sam Altman used a Ghibli-style image as his profile pic on X
    Sam Altman Ghibli-style

  2. Users are creating Ghibli-style animations

    https://x.com/roaneatan5/status/1910511267066110291

  3. Celebrities are being illustrated in Ghibli-style on Instagram
    Barbie Ghibli

  4. Musicians using it in promotional posters
    Concert flyer

  5. Transforming personal photos into fantasy scenes

  6. Sharing scenic Ghibli-style photos on social media

  7. Stylizing hand-drawn art

  8. Ghibli-fying original characters

  9. Creating keepsakes from family or friends' photos

  10. Turning kids’ drawings into Ghibli-style illustrations


Final Thoughts

GPT-4o’s image generation is impressive, and Ghibli-style conversions feel delightfully authentic. Just be mindful of copyright and ethical considerations—especially with specific characters or IP. AI art is evolving fast, and it’s never been easier (or more fun) to explore your creativity with it.

AI活用のノウハウ集「AI総合研究所」サービスご紹介資料

「AI総合研究所 サービス紹介資料」は、AI導入のノウハウがないというお客様にも使いやすい最先端のAI導入ノウハウを知れる資料です。

資料ダウンロード
AI総合研究所サービス紹介資料
監修者
Shoma Sakamoto

Shoma Sakamoto

Founder of LinkX Japan and Microsoft AI Partner.Holds a Master’s degree in Technology and Innovation Management from Tokyo Institute of Technology, with research focused on natural language processing and financial engineering. Formerly engaged in AI and blockchain research at NHK Science & Technology Research Laboratories. Presented at academic conferences, published in international journals, and received the Outstanding Paper Award at the Japan Society for Management Information (JASMIN) National Conference. Previously founded and managed IT and Web3 ventures in Singapore before establishing LinkX Japan.

AI導入の最初の窓口。

お悩み・課題に合わせて活用方法をご案内いたします。
お気軽にお問合せください。

AI総合研究所 Bottom banner

ご相談
お問い合わせは
こちら!