Outpainting: August Kamp
Create new images that extend the edges of a given image
OpenAI's text-to-image AI model DALL-E now has a new "outpointing" function that lets the system make new images that extend the edges of any given picture. In the picture above, you can see how DALL-E "imagines" what's outside the frame of Johannes Vermeer's "Girl with a Pearl Earring" with the help of a person. Even though the portrait only gives a small amount of information, the system is still able to match Vermeer's style by making the same shadows and highlights as the original.
Outpainting is a new function of DALL-E
Outpainting is a function that can be used to add to the original content, but many DALL-E users have been playing with it to see what is outside the frame of famous images. (Scroll all the way down to the bottom for my favorite example...)
From a larger perspective, outpainting doesn't really add to the basic functions of text-to-image AI systems, but it does show how OpenAI will likely position itself in the growing market for these systems: by making ease of use a key selling point.
DALL-E stands out from systems like Midjourney
Many text-to-image AI models can do the same important thing as outpainting, but, like DALL-E before this update, it took a lot of manual tweaking. Making outpainting as easy as possible will help DALL-E stand out from systems like Midjourney and Stable Diffusion that are smaller but do similar things.
More than a million people can use it right now
DALL-E is now available through a beta program, and more than a million people can use it right now. Each beta user gets 50 free image generations in the first month, then 15 more uses each month after that. Then, for $15, they can buy 115 more image generations.
Nicolas Sarazin
| ArtMajeur Magazine