OpenAI has released an addendum to the GPT-4o system card, detailing its new image generation capabilities. The 4o approach is said to be significantly more capable than previous models like DALL·E 3. It can produce photorealistic images and even transform input images into new ones. This suggests a major advancement in AI's ability to generate realistic visuals. According to OpenAI, the 4o system card addendum provides additional information on the model's architecture and capabilities. --- Why it matters: This matters because it represents a significant improvement in image generation capabilities, which could have far-reaching implications for applications like art, design, and even advertising. Engineers working with AI may be interested in understanding how this new approach can be leveraged to create more realistic and transformative visuals. Source: https://openai.com/index/gpt-4o-image-generation-system-card-addendum