Leading AI startup OpenAI has recently previewed the third version of Dall-E, their widely recognized text-to-image tool.
Dall-E 3 offers an essential upgrade, as it empowers users to use ChatGPT –the company's state-of-the-art AI chatbot– for creating image prompts.
It has also enhanced the tool's interpretative capabilities, enabling it to understand nuances better than ever before, thus producing more precise results and simplifying the creative process.
The most recent version of Dall-E is presently available in a research preview and will be accessible to both paid and free users in the upcoming months.
The most significant enhancement in Dall-E 3 is that it has been developed on OpenAI’s proprietary advanced AI chatbot, ChatGPT. This allows users to employ the AI chat tool for crafting detailed prompts, which are known to be ideal for text-to-image creation on this platform – from a simple sentence or brief written guidance. By just inputting a few words about the image you want to create, the AI chatbot will generate a paragraph to use as a prompt that Dall-E will then convert into four image variations.
Using the ChatGPT as a brainstorming assistant significantly enhances text prompts and the resultant AI-generated images. Furthermore, it saves significant time and effort as you no longer need to craft the ideal text manually or learn the necessary skills to do it.
In addition to the ChatGPT integration, Dall-E 3 produces higher-quality images than its previous versions, even with the same human-penned prompt. The reasons behind this are the upgrades made to the generative model’s ability to comprehend context and other text prompt nuances, an aspect where its predecessors had fallen short.
Attaining safety measures for potentially problematic content has been the primary goal in visual generative AI, and OpenAI has worked on this in this iteration of its tool. Dall-E 3 continues to enhance its system to refuse the production of harmful and violent pictures, especially those marked with visual bias concerning overrepresented and underrepresented concepts. Additionally, it has added new restrictions for creating images of well-known figures or replicating a living artist’s style and a way for artists to opt their copyrighted work out of the tool.
These new features and improvements aim to make Dall-E a more responsible and ethical tool for AI-generated image creation, preventing potential legal issues and disagreements with artists and authorities.
This latest version of Dall-E has been released only in research preview mode. The company is expected to roll it out for ChatGPT Plus and Enterprise subscribers by early October 2023 and extend its reach to research labs and API service sometime during the fall of 2023. The exact date for Dall-E 3’s free public release is currently undetermined, as reported by The Verge.
Are you excited about this ChatGPT-based Dall-E? Do share your thoughts!