Microsoft again ahead of its competitors in the race to bring artificial intelligence to consumer applications. The company has announced the integration of DALL-E with Bing and Edge, which makes them the first search engine and the first browser with the ability to generate images from the text description that is entered. At the moment, only some users of the beta version of the browser are seeing this feature activated, but Microsoft has made available to everyone a version preview of Bing Image Creator to test the capabilities of the tool.
As the company explains in a statement, Bing Image Creator will be “fully integrated into the Bing Chat experience,” but initially it is being deploying only within the Creative mode of the app, one of the three “personalities” or styles that the Microsoft chatbot can respond with and that the user chooses from when starting a conversation. Later it will come to the Balanced and Precise modes and the access will be extended to all users.
In this way, both from the search engine and from the Bing Image Creator button located in the browser’s toolbar, the user will be able to generate any image they want using an “advanced” model of OpenAI AI which happens to its second version released last July.
Microsoft has added additional protections to the safeguards that DALL-E already had to prevent a potential misuse of the tool by the users. Thus, when the AI detects a potentially harmful image, it goes to Block the promptthe text description entered by the user, and warns him of the situation. And to avoid confusion with image authorship, Bing Image Creator includes a Bing icon in the bottom left corner of each creation.
So you can create images from text with Bing and Edge
With Bing Image Creator, Bing and Edge users will stop relying on images they can find on the web and create any by expressing your idea in natural language that the AI will understand. Next to the image description, Microsoft recommends providing additional context with information such as the location or the activity that you want to display, as well as indicating the artistic style that you want to apply.
in the version preview From the Bing image creator, suggestions are made to refine the images, noting that it is a process that the more it is repeated, the better results will be obtained. For example, telling the AI the time of the day in which the generated image or a mood that must be present in the creation as well as more technical aspects such as the image quality wave photographic lens with which the image is “taken”. Some are also noted DALL-E inconsistencies such as that faces and hands can appear distorted or that the text of the image does not present the correct appearance.
In any case, anyone can try the tool and see that the results are remarkable and represent an important competition with Midjourney, the most popular text-to-image AI at the moment and which has just released a new version. It can still be said that Midjourney surpasses DALL-E in quality, but certainly not in accessibility.