"ChatGPT Images 2.0" Arrives, with Major Gains in Multilingual Text Rendering and Composition Accuracy
OpenAI has updated its ChatGPT-integrated image generation model to “ChatGPT Images 2.0.” As the latest iteration of the image generation feature descended from DALL-E, it brings a particularly large improvement in the accuracy of text rendering across multiple languages.
Details
- Accuracy and composition control: Improved fidelity to specified details and better consistency in complex compositions
- Multilingual text rendering: Now supports accurate text generation across a wide range of writing systems, including Japanese, Arabic, Korean, Devanagari, Cyrillic, Bengali, Greek, and Chinese
- Range of styles: Improved fidelity across diverse styles, including photorealistic, illustration, comic, pixel art, and editorial design
- Reflecting current knowledge: Strengthened ability to translate up-to-date knowledge and contextual understanding into clear visuals
- Aspect ratios: Supports variable sizes for generation, from banners to tall mobile-oriented images
How to try it
Available via “Try in ChatGPT” from the image generation feature within ChatGPT. No additional plan subscription or special setup is required (pricing details were not specified).