OpenAI announced the third release of its generative AI visual art platform, DALL-E; this version now allows users to use ChatGPT to create prompts and includes more security options. DALL-E 3 is also visually much more advanced than its previous version.

DALL-E 3 is out

Today’s cutting-edge, AI-powered rendering tools take prompts that are descriptions of images and transform them into works of art in a range of styles, from photorealistic to fantastical. However, preparing the right request or prompt can be difficult, so much so that the profession of “prompt engineering” emerged for this purpose.

DALL-E is an artificial intelligence that turns text prompts into visuals, just like Stable Diffusion or Midjourney. Even though DALL-E 2 has been in our lives for a long time, it misunderstood some things and fell behind in terms of quality compared to its competitors. OpenAI researchers say DALL-E 3 now understands context much better.

A new feature of DALL-E 3 is its integration with ChatGPT. It will now be possible to access DALL-E3 and create visuals directly via ChatGPT. In a demo, Aditya Ramesh, principal investigator and head of the DALL-E team, asked ChatGPT to help him come up with a logo for a ramen restaurant in the mountains. ChatGPT wrote a long prompt and DALL-E offered four logo options.

Thanks to ChatGPT, creating a request is thus eliminated. You say the visual you want to make, and ChatGPT translates it into a language that DALL-E can understand. ChatGPT integration is not the only innovation coming with DALL-E 3. OpenAI says DALL-E 3 also produces higher quality images that reflect prompts more accurately, especially when dealing with long prompts. It can now also deal better with human hands, faces or writings.

DALL-E 3 release date

This new version of DALL-E will be available first to ChatGPT Plus and ChatGPT Enterprise users in October, and then to research labs and the API service in the fall. OpenAI plans to stagger the release of DALL-E 3, but there’s no word yet on when the free public version will be released.

OpenAI claims to have focused a lot of work on creating robust security measures to prevent obscene or potentially hateful images from being created in DALL-E 3. DALL-E 3 will also not be able to recreate images of public figures – provided a name is specifically mentioned in the prompt. However, it is also stated that DALL-E 3 has been trained to refuse to produce images in the style of living artists. However, unlike DALL-E 2, it can imitate art in the style of certain artists when desired.

Currently, competitors such as Midjourney and Stable Diffusion are well ahead of DALL-E 2. Both tools can create photorealistic scenes. In addition, they can produce not only a single scene, but also versions of the scene, such as a continuation or extended version. So DALL-E 3 has a lot to prove.