DALL-E 2

OpenAI's generative model transforms natural language prompts into high-resolution 1024x1024 images with precise semantic alignment.

DALL-E 2 utilizes a diffusion process to generate original visuals from text inputs. This OpenAI system excels at Outpainting (extending an image beyond its original borders) and Inpainting (replacing specific elements within a frame). Since its April 2022 debut, it has enabled creators to produce complex compositions: think "a 1920s jazz band playing on Mars" or "a bowl of soup that is a portal to another dimension." The model leverages a CLIP-based architecture to maintain high visual consistency across diverse artistic styles.

https://openai.com/dall-e-2