Currently, not a day goes by without at least one small AI revolution for generative image creation being released. But until recently, many of these tools were reserved for experimenting users who like to try out open source applications with the command line. Which got us thinking about it two weeks ago, where are the big manufacturers with their own tools?.
But now, just in time for early spring 2023, the first commercial AI seedlings for end users are also coming to light. So Adobe surprised its users only the day before yesterday with Firefly. Perhaps also to take the wind out of Microsoft's sails just one day later.
Yesterday, the Redmond-based software giant also presented a browser-based Web service that enables anyone to generate high-quality synthetic images from their own prompts directly in the browser.
The Bing Image Creator (or somewhat bumpily dubbed "image creator") is still in an open beta preview phase, but can already be used for free by anyone with an active Microsoft account (which, as of late, should be almost every Windows user automatically).
In order to regulate the access according to scarce computing power resources, every user receives so-called boosts, with which images can be calculated especially fast. Currently, one gets 10 boosts per week for free. If these are out, additional boosts can be "bought" through so-called Microsoft Rewards. These rewards can be "earned" by different behaviors at Microsoft. For example, you can earn them for actively using the Bing search.
If you have no more Boosts, you can still have your images calculated slowly with fewer resources and receive a notification by mail when the images are ready - which took us about 6 hours yesterday.
Per prompt, the Bing Image Creator currently calculates four images in 1024 x 1024 pixel format, whereby there are (still?) no further intervention options via the interface - for example, to change the output aspect ratio. Also useful functions like negative prompting are not documented yet.
But Microsoft provides with the Bing Image Creator a really easy access for everyone to gain first experiences in prompting with a current text-to-image model. This is probably where the trend will go in the next few weeks: Namely, simple AI interfaces from providers such as Adobe or Microsoft, with which one can get started immediately...