OpenAi is an organization that specializes in creating useful programs that use artificial intelligence to make our lives easier. They recently announced a new and improved version of the DALL-E program. Aptly named DALL-E2, it is a program that can generate raw images with nothing more than textual input.
This program is used to create concept art in a matter of minutes. Anything you can think of, DALL-E2 can create a photorealistic never-before-seen image before your eyes. It can understand language and create these images in two separate parts of its program.
Clip is the part of the program that understands language and is what DALL-E2 uses to get an idea of what it needs to make. It understands concepts such as riding, next to, with, and can pass that information along to the second half of the program
This part of DALL-E2 is what makes the images. It scours the internet for images of the input, and then using the instruction generated by Clip, it can assemble the information given into an image.
OpenAi has created this software for specialized research and business purposes, so there are things it cannot do by design. DALL-E2 will not generate any adult content, any images with weapons, and will not output any celebrity faces.
The fully developed program currently has an extensive waiting list for companies and organizations interested in the software to experiment with.
However, it has a less powerful lite version that anyone can play with. It is far from perfect but is still a lot of fun to play with, and even generated the thumbnail you saw when you clicked on this article. To experiment with it yourself, click here.
To see everything OpenAi has to about DALL-E2, and their other projects go here.
To see the code for Dall-E mini from GitHub, click here.