OpenAI DALL-E generates fantastic images

OpenAI’s DALL-E 2 creates fantastical images of almost anything you can think of

The OpenAI consortium, founded by Elon Musk, and financially supported by Microsoft, unveiled their most ambitious project, the DALLE machine learning system, in January 2021. The multimodal AI could generate images (albeit cartoonish) based upon the attributes specified by the user.

The first DALLE (a portmanteau consisting of \”Dali\” as in an artist and \”WALLE\”, as in a Disney animated character) was able to generate images, combine multiple images in a collage and provide different angles of perspective. It could also infer certain elements of an image, such as shadow effects, from a written description.

The OpenAI team stated in 2021 that \”Unlike 3D rendering engines, which must have inputs specified in full detail and unambiguously, DALL*E can often \”fill in the gaps\” when the caption implies the image must include a specific detail that isn’t explicitly stated.\”