10 Crazy AI-Generated Images made with GPT-4o’s New Power​

Say Bye to DALL-E and Hi to GPT-4o’s new NATIVE Image Generations superpowers! Now, if you ask ChatGPT to make an image, a new system will be working behind the scenes. The results are fantastic and bring a new enthusiasm to the AI fandoms online. So, I decided to share some of the wildest images made by GPT-4o today.

10 Wild Examples of Images made by GPT-4o

OpenAI has integrated its latest image-generation model, GPT-4o, directly into ChatGPT. This feature is available to both free and paying users, with free users having certain usage limits.

1) It Can Really WRITE

Many image models, even OpenAI’s ones, suffer when writing texts. But not anymore. Here’s one of the official examples shared by the company:

Prompt: A wide image taken with a phone of a glass whiteboard in a room overlooking the Bay Bridge. The field of view shows a woman writing, sporting a t-shirt with a large OpenAI logo. The handwriting looks natural and a bit messy, and we see the photographer’s reflection.

Output:

They also included the text in the prompt, and the model generated the image with the exact text, with no mistakes at all. The best part is how well it is understood to add a reflection on the board with correct opacity.

2) Creating Diagrams

Either you need a graphic designed or spend hours on photoshop to create some amazing diagrams. Now, since ChatGPT already knows about a lot of stuff, just prompt it with the name of the diagram or what the image needs to explain.

Prompt: An infographic explaining Newton’s prism experiment in great detail.

Output:

Like the previous example, there are no spelling mistakes. Then, the diagram is accurate with the right colors used. This is going to be a great tool for teachers as well as students to learn new concepts.

3) Edit Existing Images

Not just making new images, you can edit older ones too. Here’s an example of this feature of GPT-4o:

The native editing feature was available in DALL-E, and now it’s more refined!

4) Images with 20 Objects

If you ask any image generation to AI to add multiple objects specifically, it will struggle. Not with GPT-4o. It can handle 10-20 objects.

Prompt: Times Square in New York City in the afternoon, with no people, vehicles, or illuminated billboards.

Output:

Most users are impressed by this improvement making it unique compared to its peer.

5) Photo-realism on a new level

Midjourney and other top AI-image generation tools are now more capable of making realistic images of humans. GPT-4o took it to another level.

Prompt: Generate a candid, Polaroid-style photograph of four diverse friends in their early 20s at a gritty dive bar. The lighting features a very harsh, direct flash, creating sharp shadows and giving the photo a very overexposed, vintage instant-camera feel. Colors should be slightly muted, evoking nostalgic, early-2000s party vibes. The aesthetic is casually emo. No border or logos or signs. There’s an interesting-looking wall behind them with some light graffiti. The quality of the image should be very sharp and detailed (very little grain). The energy should be silly and chaotic. They’re either playfully grimacing, smiling, or pretending to look tough. One of them should have their friend in a silly, playful headlock. Their mouths are closed.

Output:

I can see you thinking Who can write such accurate prompts. Well, that’s going to be a new industry, and we will still need experts who can describe the exact image suitable for certain situations.

6) Add Multiple Images

You can add your images for editing, but you can also add one image and ask GPT-4o to merge them.

This was not something even OpenAI talked out. So, people are finding new ways to test the new capabilities.

7) Get Your Ghibli Version

While looking at X, some of the users are using GPT-4o to change their pictures into anime style.

Studio Ghibli is a Japanese animation studio known for beautifully crafted films like Spirited Away and My Neighbor Totoro. Their movies feature stunning visuals and themes of nature, adventure, and magic.

8) Create Memes

Because AI was not good with spelling, meme generation was not something it could handle. But AI is now growing up and it has learned to write English. See this example:

Now, any normal person just needs an idea of what images they need. AI will take care of the rest of the work.

9) Marketing Images

If we are in marketing, you still need to know at least one image editing tool, something easy like Canva, to transform the marketing materials into JPEGs. It will take some time to make a new poster or pamphlets.

No need to choose a font, align the texts, or change the sizes. Now, just copy the text from your marketing copies into the ChatGPT prompt box and add your product images, and it will give you a great post for Instagram.

10) Consistent Characters

One of the problems with the AI image generation model is that when you ask it to use a character for a new image again, it will get confused. Even after explicit prompting, it will change the character a little bit. But users have said online that GPT-4o is so far the good in this area.

This will be great for branding for companies as well as storyboarding for movie concept artists.

Takeaways

While this feature offers significant creative potential, OpenAI has implemented robust safeguards to prevent misuse. Generated images include digital markers indicating their AI origin, and there are restrictions on creating images involving real people, nudity, or graphic violence.

Then, there is always room for improvement. OpenAI has clarified that this new capability might fail sometimes, especially when generating text on its own and editing existing images.

Overall, this integration marks a significant advancement in ChatGPT’s capabilities, providing users with a powerful tool for creating detailed and accurate images through conversational prompts.