Googles Advanced AI Image Generator Is Officially Launched

Google has released the latest version of Imagen 3, its AI tool for generating images from text that is currently only available to users in the US, VentureBeat reports. The tool, which you can access via Google's AI Test Kitchen, is said to produce images with "better detail, richer colors, lighting, and less distracting effects" than Google's previous AI models.

Google first announced the upgraded Imagen 3 tool at its I/O event in May, but it has only been made widely available through Google's Vertex AI platform in the past few days. Some Reddit users began experimenting with Imagen 3 last week, and Google published a research paper about the tool on Tuesday.

Like other AI image generators, Imagen 3 can generate detailed images based on the text you ask for. You can also edit images by highlighting a specific part and describing what you want to change.

Even though it's an advanced image-generating AI, there are some safety rules in place, like it won't generate images of famous people like Lady Gaga, and it won't generate images of weapons either. And while it will stop before generating characters with names, you can still create detailed images by describing the character you want to create.

And you can use this tool to create images of iconic characters like Mario, Sonic, Pikachu... while my colleague was able to create characters that looked like Mickey Mouse. I also found that it can create logos belonging to companies like Apple, Macy's, Hershey's, and even Google, as you can see in the image at the top of this article.

Despite its relatively lenient safety regulations, Imagen 3 still draws a sharp contrast to Grok, Elon Musk's X platform AI image-generating tool. Grok has been used to create somewhat "wild" content, including images of drugs, violence, and celebrities.

Google's AI tools have also had their share of problems, however. Earlier this year, Google stopped allowing users to create images with its Gemini AI chatbot after users discovered it was generating historically inaccurate images.