Pictures are literally no longer limited by their frames thanks to the new ability of DALL-E, an amazing artificial intelligence (AI) capable of generating images from text descriptions. Simply explain to DALL-E through a few named instructions prompts what you want to see in an image so that it delivers results that cause awe.
Artificial intelligences like DALL-E generate images by following text instructions that they interpret and recreate
But if that ability of DALL-E was already impressive enough, with «Outpointing» (Overcoming) the barriers that until now imposed the frames to the paintings are knocked down. It supposes being able to extend beyond those limits what appears in an image, generating images of what lies beyond what the canvas shows.
To do this, it is enough to supply this AI with a frame so that it takes it as a base. From there and through the aforementioned prompts it describes what appears on each side (or in the lower or upper parts) of the frame so that DALL-E follows these instructions, generates new images and also, and to make this enlargement more coherent, mimic both the painterly style and the color palette or the lights and shadows present in the painting from which it starts.
DALL E: Introducing Outpainting
Extend creativity and tell a bigger story with DALL-E images of any size
blog: https://t.co/dBRzCXWDYF pic.twitter.com/9sOyDMPEb7
— AK (@_akhaliq) August 31, 2022
Other AIs that generate images from text instructions, such as Midjourney or Stable Diffusion, are able to achieve similar effects from reference images, but not as easily as DALL-E.
Currently accessible in beta version, more than a million users have already registered. During the first month, 50 images can be generated for free, subsequently moving to a maximum of 15 monthly images. A total of 115 AI-generated images can be obtained by paying $15.
