DALL·E launched in 2021 was able to generate simple images from text, and DALL·E 2 has taken the technology much further. It allows to create complex images with several objects and even plots as well as adjust the provided imagery by adding or removing objects. The quality has increased, too: DALL·E 2 generates up to four times higher quality images than the previous one. The result is provided in the size of 1024×1024 pixels instead of 256×256 pixels.
This is all about technical features, but what about the visual part? DALL·E 2 performs like a human artist, so it can freely mix styles and concepts and match multiple aesthetics. It can equally imitate the famous paintings and create advanced experimental art comparable to the works of the prominent creators like Banksy or Beeple.
Currently, access to DALL·E 2 is open only to a group of users through the waiting list. It’s unclear when this AI system will be available for public use or become open-source. However, the abundance of posts and shots on Twitter seems enough to assess the magic created by computer vision technology (that’s what DALL·E 2 is based on).