DALL·E 2 is a new AI system developed by OpenAI that can create realistic images and art from a text description in natural language. The system can combine concepts, attributes, and styles to create original, realistic images and art. DALL·E 2 can also expand images beyond what’s in the original canvas, creating expansive new compositions. Additionally, the system can make realistic edits to existing images from a natural language caption and take an image and create different variations of it inspired by the original.
DALL·E 2 has learned the relationship between images and the text used to describe them through a process called “diffusion”. DALL·E 2 generates more realistic and accurate images with 4x greater resolution compared to DALL·E 1. The system began as a research project and is currently available in beta.
The developers of DALL·E 2 have implemented safety mitigations to prevent harmful generations, curb misuse and phased deployment based on learning. The system has limited the ability to generate violent, hate or adult images, and implemented advanced techniques to prevent photorealistic generations of real individuals’ faces. The content policy does not allow users to generate violent, adult, or political content, among other categories.
The hope is that DALL·E 2 will empower people to express themselves creatively and also help understand how advanced AI systems see and understand our world, which is critical to the mission of creating AI that benefits humanity.