OpenAI, the forerunner AI firm, has released a preview of the third version of Dall-E, its renowned text-to-image generator.
Dall-E 3 presents a crucial upgrade as it lets customers use ChatGPT –the company’s impressive AI chatbot– for image prompt writing; it also improved the tool's interpretative abilities to understand more detail and nuance than ever before.
The result is a tool that simplifies the creative process and produces more accurate results.
The latest iteration of Dall-E is currently released in research preview and will be made available for paid and free users in the coming months.
In what constitutes a major upgrade, Dall-E 3 is built natively on OpenAI’s own advanced AI chatbot, ChatGPT, and enables users to use the AI chat tool to create detailed prompts (which are known to work best for text-to-image generation on this platform– from a simple sentence or short written instruction. You can simply enter a few words on what you want to create, and the assisting bot will generate a paragraph to use as a prompt that Dall-E will then synthesize into four image variations.
Using ChatGPT as a brainstorming assistant, you can significantly improve your text prompts and your AI-generated image results. Not to mention how much time and effort it can save if you don’t have to spend time engineering the perfect text prompt by hand.
Let's not forget that Dall-E is responsible for over 2 million AI image generations a day, according to visual AI stats.
Besides the ChatGPT integration, Dall-E 3 produces better images than its previous version, even from the same human-written prompt. This is because the lab has improved the generative model’s ability to understand context and other nuances in text prompts.
Equally relevant, they’ve also upgraded the safety measurements regarding potentially problematic content. This is one of the most pressing issues in visual generative AI.
Dall-E 3 continues to improve its system to decline the generation of harmful and violent imagery, including that marked by visual bias regarding overrepresented and underrepresented concepts. But more interestingly, it has new restrictions that don’t allow creating images of public figures or mimicking a living artist’s style, and even an opt-out feature where human artists can submit a removal request for their copyrighted work. In the future, the tool will also be able to reject prompts that seek to recreate an existing, human-made image. Finally, they’re also developing a feature to recognize and accurately identify AI-generated images.
These are all solutions aiming at a more responsible and ethical use of Dall-E and preventing lawsuits and other disagreements with artists and authorities.
As mentioned, the latest version of Dall-E was released in research preview only so far. The company expects to roll it out for ChatGPT Plus and Enterprise subscribers in early October 2023. It plans to reach its research labs and API service sometime during the fall.
There are so far no date estimations for the release of Dall-E 3 for free for the general public, but according to The Verge, they plan to stagger it.
What do you think about the upcoming ChatGPT-based Dall-E? We’d love to know your thoughts!
I am an experienced author with expertise in digital communication, stock media, design, and creative tools. I have closely followed and reported on AI developments in this field since its early days. I have gained valuable industry insight through my work with leading digital media professionals since 2014.
AI Secrets is a platform for tech decision-makers to learn about AI technology. Our team includes experts such as Amos Struck (20+ yrs ICT, Stock Photo, AI), Ivanna Attie (expert in digital comms, design, stock media), and more who share their views on AI.
Get AI news in your inbox & join thousands of engineers and managers using AI to boost sales and grow market share