High Image & Video Gen · 1 min read
DALL·E 2: the quality leap in image generation
In one sentence OpenAI announces DALL·E 2, a diffusion-based text-to-image generator producing 1024×1024 images, four times its predecessor's resolution, and able to edit and vary existing images. At first it is open only to a small group of users.
Reading level
About fifteen months after DALL·E 1, OpenAI unveils DALL·E 2: same idea (text to image), but with quality that makes many images hard to tell apart from photos or professional illustration.
Besides creating images from scratch, it can edit regions of an existing image while respecting shadows, lighting and textures, and generate variations of a given image.
At announcement, access is limited to a small group of users, widened gradually.
Companies
OpenAI
Tools
DALL-E 2
Tags
OpenAIDALL-E 2DiffusionText-to-ImageunCLIP
Sources