OpenAI has introduced the DALL-E 3 image generator, which has learnt to draw fingers

OpenAI has released a new version of the DALL-E 3 image generator that understands significantly more nuances and details than previous systems.

Here's What We Know

According to the developers, DALL-E 3 is significantly better than DALL-E 2 at creating complex images For example, the new model can accurately depict a scene with specific objects and the relationships between them.

DALL-E 3 also outperforms DALL-E 2 when creating text within the image and human details such as hands.

A comparison of DALL-E 2 (left) and DALL-E 3 (right) results generated from the same query

"DALL-E 3 creates engaging images by default—no hacks or prompt engineering required" OpenAI stated.

DALL-E 3 is currently in a preliminary research version. The model will be available to ChatGPT Plus users and Enterprise customers in October via API.

Source: OpenAI