[Ghibli-style images with AI] Introducing how to create free images with ChatGPT and copyright

The latest model GPT-4o released by OpenAI not only can generate text, but also has image generation functions, which has attracted widespread attention. The most eye-catching feature is the generation of "Ghibli-style illustrations". Simply upload the photo to ChatGPT and give simple instructions to generate a dreamy image like a work in Studio Ghibli. This article will explain in detail the actual operation process, the effect experience, as well as the errors and copyright issues that need to be paid attention to.

For details of the latest model " o3 " released by OpenAI, please click here ⬇️ OpenAI o3 (ChatGPT o3) detailed explanation: Features, pricing system and o4-mini introduction

The English version of Ghibli-style articles, please click here⬇️Use ChatGPT to create Ghibli-style AI art | Free tool to experience the latest Ghibli-style AI trends [GPT-4o]

Table of contents

How does ChatGPT's AI achieve Ghibli style? Features of AI image generation

Features of image generation function

How to make Ghibli-style images in practice (Ghibli)

? Steps

The o3 model improves the quality of image generation function

Generation other than Ghibli style (AI anthropomorphism, method of ビックリマンチョコ style)

Images in style

Anthropomorphic images

Errors when generating failure and solutions

Image generation prompt word skills

Copyright and ethical considerations

Copyright considerations for image generation (generating AI)

Ten application cases of Ghibli-style image generation

Summarize

How does ChatGPT (ChatGPT) AI achieve Ghibli style? Features of AI image generation

GPT-4o released in March 2025 has added image generation function, which can not only generate text. The most popular one is the "Ghibli Style (Ghibli-style)" image generation function .

"Ghibli style" refers to the reproduction of AI's unique worldviews such as soft colors, warm scenery, and animation-style character portrayal in Studio Ghibli's animation works. The user simply uploads the image and instructs “Generate Ghibli Style”, and the AI ​​will convert based on the style it learns to generate unique illustration-style images.

This style conversion function is realized thanks to the new image generation model equipped with GPT-4o.

For detailed information on GPT-4o image generation, please refer to the following article. [Related Articles] Detailed explanation of ChatGPT4o (GPT-4o) image generation: usage method, pricing and prompt words

Features of image generation function

The characteristics of ChatGPT image generation function are as follows:

• Using autoregressive model Compared with traditional diffusion models, composition and color are easier to keep consistent and the effect is more natural.
• High correlation between text and image Even instructions such as "00 Style" can achieve high-precision style conversion.
• Operation is done in ChatGPT Uploading, directives, and generation are all done on the UI interface of ChatGPT.
• Free users can also use it The free plan also provides daily image generation services (※Restricted).

! Previously, the free plan was limited to generating 3 images, but as of April 2025, the number of images available has increased.

How to make Ghibli-style images in practice (Ghibli)

We use GPT-4o's image generation function to actually try to generate "Ghibli-style" images. The following will introduce the specific steps from uploading an image to viewing the generated results step by step.

? Steps

1. Start ChatGPT (GPT-4o) to access the OpenAI official website and select the GPT-4o model. Free plan users can also use the image generation function, but there are generation restrictions. Both mobile phones and tablets are available.
2. Upload images Click the "+" icon at the bottom of the screen to upload the image you want to use (for example: character photos, landscape photos, etc.). It is recommended to use JPG or PNG format, the size should not exceed 2MB.

Image upload

3. Enter the "Generate Ghibli Style" command Enter clear style specification information such as "Convert this image to Ghibli style" in the text input field.
4. After waiting for the AI ​​to generate the image (about 1-3 minutes) to send the instruction, GPT-4o will process the image and display the conversion result. The generation process may take 1-3 minutes.

Ghibli style generated image

5. View the generated image The generated image will be displayed directly on the screen. You can download and save the images you like (※It is recommended that you only use for personal use).

Image editing method

The generated image can be further edited. For example, the hue and brightness can be adjusted, or accentuated specific parts, resulting in a more desirable effect.

First select the part you want to modify.

1. Click on the image to select

Image modification

2. Issue a chat command to the selected section. The instruction here is "Patch white part of the abdomen".

Instructions to whiten the abdomen

With clear instructions, AI can modify the corresponding parts.

The o3 model improves the quality of image generation function

OpenAI released the latest model "o3", which improves the quality of image generation. In the past, images often deformed when text generation, but the images generated by the o3 model are more natural.

O3 model image generation

In addition, the o3 model can also infer the shooting location based on landscape photos. Send the image to o3 and ask the location of the image

Do you know where this is? For answers and more about the o3 model, check out the following article!

【Related Articles】 Detailed explanation of OpenAI o3 (ChatGPT o3): Features, pricing system and o4-mini introduction

Generation other than Ghibli style (AI anthropomorphism, method of ビックリマンチョコ style)

In addition to Ghibli style image generation, it can also be converted to other styles. For example, the following style conversions can be performed:

• ビックリマンチョコ风格 It can be converted to the character style of ビックリマンチョコ.
• Anthropomorphism Animals or characters can be anthropomorphized.

Images in style

Next, we try to generate images in the style of ビックリマンチョコ.

Kumamoto's input image

The command is: "Generate the style of " ビックリマンチョコ. Glitter, like a rare card. It is best to be the Angel Prince type".

Output image

The generated images are nostalgic!

[Related Articles] [AI Free] Method for producing image generation in style - ChatGPT application details

Anthropomorphic images

Anthropomorphic images

In addition to Ghibli style, changing the prompt word can also achieve anthropomorphism. The picture above shows anthropomorphizing a dog into a boy.

It can also be anthropomorphized into an old man-like character.

Change to the old man's instructions

This interesting usage is also very practical. For details, please see the following article.

【Related Articles】 【AI Free】Detailed explanation of the method of anthropomorphization of ChatGPT (ChatGPT) for pets and animals

Errors when generating failure and solutions

Sometimes the generation may fail. There may be several reasons for this:

Error content	reason	Solution
Upload failed	Too large file	Resize image
Unable to generate Ghibli style	The instructions are unclear	Directly indicate "Ghibli style"
Stop midway processing	The server load is too high	Change the time period and try again (relatively stable from late night to early morning)
Can it be used commercially?	Copyright concerns exist	Avoid public and commercial use, only for personal use

It should be noted that there is a limit on the number of times. In particular, free plans can only be generated up to 3 times a day (the number of paid plans is more), and it is recommended to plan and use them reasonably.

Number of times limit display

For more information on the number of times limits, please refer to the following article. [Related articles] Detailed explanation of the number of ChatGPT limits: hourly usage limit and response methods

Image generation prompt word skills

The content of the prompt word will greatly affect the quality of the generated image. Here are some tips for generating Ghibli-style images:

• Give specific instructions Don’t just talk about “Ghibli style”, but talk about specific requirements such as “Ghibli style color, background is forest”, so that you can get a more ideal image.
• Consider image size Too large images may lead to excessive processing and prolong generation time. It is better to choose the right size (for example: 800x600px).
• Give a style reference In addition to "Ghibli style", specific references can also be given, such as "00 style color", so that AI can understand the style more accurately.
• Utilize feedback Give feedback such as "brighter" or "change color" to the generated image, and the AI ​​will learn and apply it to the next generation.
• Try multiple prompt words Even with the same image, trying different prompt words can get images of different styles and atmospheres. For example, you can try different themes such as "Ghibli-style night view" or "Ghibli-style characters".

Various attempts can be made to achieve diversified output. Whether it is a newbie or a veteran, it is worth a try.

Copyright and ethical considerations

Copyright and ethical issues are required when generating Ghibli-style images. Please note the following points:

• "Ghibli Style" is a style imitation and should be avoided in commercial use
• Don't over-imitate specific characters and works (even AI should pay attention to originality)
• It is necessary to consider director Hayao Miyazaki's cautious attitude towards AI creation

Please refer to the system description for details.

Copyright considerations for image generation (generating AI)

At present, Japanese law has not clearly stipulated the copyright ownership of AI-generated content.

According to the Department of Culture, if a user enters a prompt word with creative intention and is considered to have “used AI for the purpose of expressing thoughts and emotions”, the user may be identified as a copyright holder.

Excerpts from the Department of Culture Information Reference: Department of Culture: AI and Copyright

In addition, at the House Cabinet Committee meeting on April 16, 2025, the scientific strategy officer of Yuhiko Nakahara, the Ministry of Culture and Science of the Ministry of Culture and Science made the following explanation on "Ghibliization (Ghibli-style AI-generated image)":

"The final decision will be made by the judiciary, but if it is just a similar style and creativity, it does not constitute infringement. Copyright law does not protect styles and creativity that are inadequately expressed in creative ways. However, if the AI-generated content has similarity or dependence to existing works, it may constitute infringement."

This is the answer to the question of the Cadet Masato Imai’s question on whether Ghibli-style images infringe copyrights. The current law interprets it as “ only imitating styles is legal ” and “ it may be illegal if the specific roles and settings are similar .”

!It is particularly important to note that it contains specific works titles and worldview prompts such as "like chinchillas". Images generated based on such prompt words are very likely to violate Studio Ghibli's copyright and trademark rights , and unauthorized use, disclosure and commercial use are all legal risks .

Therefore, AI-created works may involve copyright, trademark rights, portrait rights and other rights , so careful processing is required when generating images that imitate specific styles and characters.

[Related Articles] What is the copyright of AI-generated works? Detailed explanation of precautions and actual cases

Ten application cases of Ghibli-style image generation

Ghibli-style image generation has also become a hot topic on SNS, and many users have shared their works.

• 1. OpenAI Sam Altman uses Ghibli-style image as X's profile photo

Sam Altman's Icon Image Reference: X's Post

• 2. Ghibli-style animation production
• 3. Entrepreneur using Ghibli-style images

There are more and more sharing on Instagram and X (old Twitter).

Instagram: Barbie's Ghibli-style image conversion (@fallin.love.barbie)

Comments like "I originally wanted to be plum, but..." appeared.

• 4. Convert creatives to Ghibli style

In addition, violinist Takashima さko also used Ghibli-style images and shared them on SNS. She made her own concert flyer-style image.

Convert creatives to Ghibli style reference: Instagram

• 5. Convert your own photos to Ghibli style for entertainment
• 6. Convert landscape photos to Ghibli style and share them on SNS
• 7. Adapting your own illustrations in Ghibli style
• 8. Convert your own characters to Ghibli style
• 9. Ghibli-style images are used as entertainment between family and friends and gifts
• 10. Convert children's paintings to Ghibli style and save them as memories

A variety of use cases have been reported. It's also fun to try out unique usages.

Summarize

GPT-4o's image generation function is very powerful, and the "Ghibli style" conversion is also extremely natural. It is highly recommended to use within the scope of entertainment and interests, but be sure to pay attention to copyright and ethical issues. The future development of AI generation technology has also attracted much attention.

Please note that all "[https://www.php.cn/link/39cec6d4d21b5dade7544dab6881423e]" above should be replaced with the actual https://www.php.cn/link/39cec6d4d21b5dade7544dab6881423e. I can't access or create a web link directly.

The above is the detailed content of [Ghibli-style images with AI] Introducing how to create free images with ChatGPT and copyright. For more information, please follow other related articles on the PHP Chinese website!