DALL-E image generator is now open to everyone

  News
image_pdfimage_print
An artwork created with OpenAI's DALL-E image generator.
Enlarge / An artwork created with OpenAI’s DALL-E image generator.

If you’ve been itching to try OpenAI’s image synthesis tool but have been stymied by the lack of an invitation, now’s your chance. Today, OpenAI announced that it removed the waitlist for its DALL-E AI image generator service. That means anyone can sign up and use it.

DALL-E is a deep learning image synthesis model that has been trained on hundreds of millions of images pulled from the Internet. It uses a technique called latent diffusion to learn associations between words and images. As a result, DALL-E users can type in a text description—called a prompt—and see it rendered visually as a 1024×1024 pixel image in almost any artistic style.

Aside from image-to-text generation, DALL-E also includes a feature called “Outpainting” which allows you to upload an image and extend its borders using image synthesis. You can also merge several photos into one by generating a visual bridge between them, blending styles.

A DALL-E example of "An astronaut riding a horse."
Enlarge / A DALL-E example of “An astronaut riding a horse.”

OpenAI announced an earlier version of DALL-E in January 2021 with surprising capabilities but obvious limitations. In April this year, it premiered DALLE-2, which wowed a limited test audience of 200 researchers with its ability to generate near-photorealistic images and imitate the styles of artists. After gradually opening up to more users and instituting a pricing model in July, more than 1.5 million users worldwide are now using DALL-E (re-branded from DALL-E 2) to create more than 2 million images a day, according to OpenAI.

DALL-E is a commercial service, and it comes with some significant limitations. New users get 50 free credits, and you can purchase credits in groups of 115 for $15 apiece. Images generated with DALL-E are owned by OpenAI, but the company grants users exclusive rights to “reproduce and display” their generations as long as they comply with the content policy. Content violations will revoke your right to legally use generated images.

Previously, we’ve reported extensively about Stable Diffusion, which is an open source image model (and competitor) similar to DALL-E created by Stability AI. For the past month, Stable Diffusion has been available with few restrictions through open source channels and its own commercial website. Now that DALL-E is open to everyone, competition between image synthesis models will likely heat up.

https://arstechnica.com/?p=1885259