If you've been waiting for better tools before experimenting with AI-generated art for your business, you won't have to wait much longer. This week, OpenAI announced a major update to Dall-E, its popular text-to-image model. The upgraded model, called Dall-E 3, is significantly more powerful than its predecessors, and its new functionalities could make it a must-have for entrepreneurs when it's released in October. So what's different? The biggest change is that Dall-E 3 can now be accessed directly through ChatGPT, turning the chatbot into a brainstorming partner and prompt-refiner when generating original A.I. art. In an example video, a ChatGPT user asks for the chatbot to generate an image of a "super-duper sunflower hedgehog" for his daughter. The bot instantly creates four unique, more descriptive prompts, such as "A joyful super-duper sunflower hedgehog frolicking in a vibrant garden, surrounded by blooming flowers and butterflies."
Dall-E is at its best when given a lengthy and detailed description of what the user wants, so by using ChatGPT to create more detailed prompts, the model can generate better and more detailed visuals. Later in the example video, the user asks ChatGPT to generate a few stickers featuring the sunflower hedgehog, illustrating how the tech could be used to ideate or even fully design products, marketing materials, or packaging. Another major upgrade over the earlier versions of Dall-E is that the model can now accurately generate words. Previously, when trying to generate images of, say, a person wearing a college sweatshirt, the model would produce a jumbled set of characters. The model is also now capable of generating realistic-looking human hands. Dall-E 2 had well-documented issues generating hands and fingers. Of course, there are some limitations on what Dall-E 3 can do. For instance, it is programmed to refuse any attempts to generate an image of a known public figure. The company also says that it's working on a classifier to identify if an image was created using the tech. In addition, the tool will decline any requests that ask for an image in the style of a living artist, and artists can also opt out of allowing their work to be used to train new models.
For business leaders looking for some quick graphic design work, or who are hoping to experiment with potential new logos or visual identifiers, the updated Dall-E could be an incredibly useful tool. In addition, OpenAI says that the images you create with Dall-E 3 "are yours to use and you don't need our permission to reprint, sell, or merchandise them.
留言