{"id":5890,"date":"2025-04-08T11:23:55","date_gmt":"2025-04-08T09:23:55","guid":{"rendered":"https:\/\/blog.keitaro.io\/en\/?p=5890"},"modified":"2025-09-16T16:59:05","modified_gmt":"2025-09-16T14:59:05","slug":"chatgpts-new-image-generator-the-good-and-the-bad","status":"publish","type":"post","link":"https:\/\/blog.keitaro.io\/en\/chatgpts-new-image-generator-the-good-and-the-bad\/","title":{"rendered":"ChatGPT\u2019s New Image Generator: The Good and The Bad"},"content":{"rendered":"\n

AI just got a lot more visual \u2014 and it\u2019s opening up wild new possibilities for advertisers. With ChatGPT\u2019s latest model, GPT-4o<\/a>, users can not only chat, code, and brainstorm… but also generate good-looking images in minutes<\/strong>.<\/p>\n\n\n\n

And if you’re in advertising or content creation, this is a big deal.<\/p>\n\n\n\n

Let\u2019s break down why everyone\u2019s talking about it \u2014 and why it\u2019s both a creative dream and something to keep a careful eye on.<\/p>\n\n\n\n

A Whole New Way to Create Visuals<\/strong><\/h2>\n\n\n\n

Until recently, image generation was cool but kind of\u2026 bad. With too many arms and fingers, weird texts and somewhat creepy details in the background.<\/p>\n\n\n\n

Now, with GPT-4o, it\u2019s smoother, way more detailed and realistic.<\/p>\n\n\n\n

Instead of just spitting out a random image, ChatGPT now builds visuals step-by-step<\/strong>. Just how a human would sketch \u2014 starting from the top-left corner and working across the canvas. And in the end, surprisingly, you get an image that actually looks good.<\/p>\n\n\n\n

\"\"
Prompt: \u201cCreate an image of a woman holding a Keitaro Tracker logo in her hands in a realistic style. She is pointing at it and smiles\u201d<\/em><\/figcaption><\/figure>\n\n\n\n

All you need to do is to write, \u201cCreate an ad image with [idea] ,<\/em>\u201d and it\u2019ll deliver. What\u2019s more, now ChatGPT can add a cohesive text<\/strong> on the image \u2014 something it couldn’t do before.<\/p>\n\n\n\n

\"\"
Prompt: \u201cCreate an ad for a BIC pen in a hair, saying “we’ve been holding your hair since 2000\u201d<\/em><\/figcaption><\/figure>\n\n\n\n

The new GRT-4o model can also mimic different art styles. It makes images in 3D, 2D and turns photos into drawings. <\/p>\n\n\n\n

You\u2019ve definitely seen it lately with everyone turning themselves into Ghibli characters. This trend was so popular, it led to OpenAI\u2019s servers crashing down<\/a>.<\/p>\n\n\n\n

\"\"
Prompt: \u201cCreate an image of a cat in Voxel style for a cat clothes ad \u201d<\/em><\/figcaption><\/figure>\n\n\n\n
\"\"
Prompt: \u201cCreate an image in a cartoonish style with a man sitting before the computer. He is happy and excited, because he can finally see his ad statistics thanks to Keitaro Tracker\u201d<\/em><\/figcaption><\/figure>\n\n\n\n

Why This Is So Powerful for Advertisers<\/strong><\/h2>\n\n\n\n

OpenAI\u2019s new GPT-4o image generation is an interesting tool for marketers. Here\u2019s why:<\/p>\n\n\n\n