Dall-E 3 will be able to process details and text in images better than its predecessor. The new version will be offered to paying ChatGPT users from October.

OpenAI has officially announced Dall-E 3, the third generation of the popular image generator. For those who don’t know Dall-E, it is an artificial tool that only requires a description in words to create an image. The current version Dall-E 2 was one of the founders of the current generative AI boom.

The devil is in the details

Dall-E 3 is not yet available to the general public, but OpenAI is already giving us a first look. The new image generator will be able to display details in an image much better. Anyone who has experimented with Dall-E 2 knows that the tool often had difficulty displaying body parts such as hands and facial expressions. The first set of examples OpenAI is releasing includes people who actually look like people.

In addition, Dall-E 3 is significantly better at word processing. With this, OpenAI refers to two things. Dall-E 3 will better understand your prompts so you don’t have to take a course fast engineering There is more to do to achieve the image you have in mind. The image generator can also process text fragments in an image. When you specify in the prompt that you want a specific object to say something, that text appears in a speech bubble, like a comic or cartoon.

Copyright ©

Tools like Dall-E 3 are not spared from controversy. Artists, photographers and graphic designers have little need for artificial designers who can create an image in seconds. Copyright in particular is a sensitive issue when it comes to AI-generated visual content.

Recently, there have been lawsuits in which people are suing AI for allegedly illegally using their intellectual property to produce content. OpenAI itself can also comment on this: The New York Times is considering a lawsuit and Game of Thrones author George RR Martin also wants to take the company to court for stealing excerpts from his well-known book series.

An American court also recently ruled that copyright cannot be asserted on AI art. OpenAI says it has taken the necessary precautions to avoid annoying copyright lawsuits. You can’t ask Dall-E 3 to create a picture in the style of a (living) artist.

The company now also expressly points out in the terms of use that all images created with the image generator are intended for “public use”; So you can do whatever you want with it. There are also restrictions on creating “harmful content” or including well-known public figures in your images to combat deepfakes and misinformation.

Integration with ChatGPT

While previous Dall-E versions were offered as a separate tool, OpenAI will fully integrate the third generation into ChatGPT. This also gives the chatbot new opportunities to generate images. Starting in October, Dall-E 3 will be available to paying Plus and Enterprise customers and later the image generator will also be offered to developers via a paid API. It is currently unclear whether there will be a free version of Dall-E 3.