Microsoft‘s billion-dollar investment in OpenAI continues to pay off. After clearing the waiting list and revealing that it was based on OpenAI’s latest language model: GPT-4, Microsoft this week released an image generator for its brand new Bing chatbot that will use an “advanced version” of OpenAI’s DALL-E model.
Called “Image Maker”, Bing’s new chatbot tool will allow users to create “out of thin air” images from the descriptions they type.
“For those in the Bing preview, Picture Maker will be fully integrated into the Bing chat experience, initially rolling out in Creative mode,” Yusuf Mehdi, head of consumer marketing at Microsoft, explains in a blog post. “By typing a description of an image, providing additional context such as location or activity, and choosing an artistic style, Picture Maker will generate an image from your own imagination.”
Like Bing’s own chatbot, the Image Builder will also appear in Microsoft Edge, and will appear as an icon in the browser’s sidebar. Microsoft has also clarified that, for the time being, the image builder will only be available in Creative mode. “Initially we will only include the Image Builder in Bing chat’s Creative mode, and our intention is to make it available in Balanced and Precise modes over time,” Mehdi says.
Bing’s three chat modes are Creative, Balanced and Accurate. While Creative mode includes responses that are “original and imaginative”, Precise mode favors accuracy and relevance for more objective and concise responses. On the other hand, Balanced mode allows for a middle ground between AI accuracy and creativity.
At the moment, the Image Builder only supports English instructions, although Microsoft says it will add support for more languages in the future. If you are not signed up for Bing Preview, you can try this new feature through this link.
