The scores of Studio Ghibli-inspired images flooding the timelines of all social media users have been made possible by ChatGPT-4o (short for “Omni”), which OpenAI says is its latest and most advanced AI model.
The model has been put to use by millions of users online for transforming their photos into Studio Ghibli-inspired images, so much so that OpenAI CEO Sam Altman said that their GPUs were melting. The Japanese animation studio is known for its serene aesthetics and dreamy storytelling.
However, that is not the only advanced feature offered by the latest version of ChatGPT. To understand more about GPT 4o, we asked ChatGPT about some hidden/secret features of 4o image generation.
Also read: Sam Altman urges users to slow down as Ghibli trend takes over internet: ‘Our team needs sleep’
What ChatGPT said
ChatGPT, one of the most popular AI platform across the world, was quick to list out the lesser-known tricks and optimizations of its latest 4o image generation model-
Transparent Backgrounds
You can generate images with a transparent background by enabling the transparent background option. This is great for logos, stickers, and compositing.
Consistent Characters in a Series
Want to create a character and keep them consistent across multiple images? You can reference previously generated images to maintain visual similarity.
Hybrid Editing (Inpainting-Like Tweaks)
You can modify parts of an image rather than generating a whole new one. This allows for things like changing colors, adding accessories, or fixing small details.
Higher Detail with Specific Prompts
Using detailed descriptors like “4K, ultra-detailed, cinematic lighting, volumetric fog” can push the quality higher. Also, specifying materials (e.g., “velvet dress” vs. “silk dress”) refines textures.
Multiple Images in One Prompt
Instead of generating one image at a time, you can request multiple variations at once using the n parameter.
Art Style Transfer
You can guide the image toward a specific art style (e.g., “Studio Ghibli-style fantasy town” or “Picasso-inspired portrait”). It won’t be an exact match, but it strongly influences aesthetics.
Text Inclusion (Limited)
AI-generated images struggle with text, but using simple, bold words in prompts (like “a neon sign that says ‘OPEN'”) increases the chances of readable text.