OpenAI launches ChatGPT image-generator | Campaign US
The latest iteration of image generation from OpenAI seeks to address many of the flaws in AI-powered visuals. OpenAI has launched 4o image generation within its flagship ChatGPT platform. Use cases highlighted by Open AI for 4o include creating 'workhorse' imagery including logos, diagrams, and illustrations, which point to several applications across the marketing communications industry.
Key Features of 4o Image Generation:
- More accurately rendered images: With 4o, OpenAI aims to create visually beautiful and useful imagery. This includes rendering text on images far more accurately, making it easier to create consistent logos.
- Better binding and adherence to prompts: 4o image generation has improved 'binding' capabilities, leading to better adherence to prompts in the final output based on text inputs.
- Being able to iterate on images: Users can now refine images through natural conversations, ensuring consistency in visuals, crucial for creating multiple iterations in ad campaigns.
- The ability to render more complex visuals: 4o can handle a higher number of objects in visuals, allowing for better control and tighter binding of objects to their traits and relations.
- A wider variety of styles: Trained on a vast array of images, 4o offers a variety of styles, including photorealism, for more convincing outputs.
Like all AI-powered models, OpenAI acknowledges that 4o image generation is not perfect. The company has also implemented limitations to prevent illegal imagery generation and included metadata to identify AI-generated images. 4o image generation is available in the ChatGPT suite of products and may soon be launched for Enterprise and Edu tiers. Other companies, such as Google with Imagen, have also highlighted logo-creation abilities in their image generation software.
This story first appeared on Campaign Asia-Pacific.
