Live demo of DALL·E 2, later today

JUNE 23, 2022

I recently got access to OpenAI's DALL·E 21 text-to-image beta. In short, this AI system can generate images from text prompts, create semantic local edits by selecting an image region and altering your prompt text, or generate variations of uploaded or generated images.

I've been playing with it and will be doing a live demo today in Live 79 and sharing some of my experiments and thoughts on this tool, which is quite impressive.

OpenAI's DALL·E 2 variations of a hand sketch of an Amazon box by Nono Martínez Alonso.

DALL-E 2 Demo Videos

  1. DALL·E 1 was released in early 2021 and DALL·E 2 in 2022. The latter produces more realistic images. Imagen, Google's version of DALL·E, was recently released but hasn't been opened to the public due to concerns about potentially harmful uses of the AI system, e.g., porn, harassment, etc.