Google’s upgraded AI image generator is now available

1 month ago 18

Google has released the latest mentation of Imagen 3, its AI text-to-image generator, to users successful the US, as spotted by VentureBeat. The tool, which you tin entree connected Google’s AI Test Kitchen, is supposed to generate images with “better detail, richer lighting, and less distracting artifacts” compared to Google’s erstwhile models.

Google archetypal announced its updated Imagen 3 instrumentality during I/O successful May, but it lone appears to person made the instrumentality mostly disposable done its Vertex AI level wrong the past respective days. Some Reddit users started experimenting with Imagen 3 past week, and Google published a probe paper connected the instrumentality connected Tuesday.

Like different AI representation generators, Imagen 3 tin make elaborate images based connected your prompt. You tin besides edit the representation by highlighting a definite portion and describing what you privation to change.

There look to beryllium immoderate guardrails successful place, arsenic the instrumentality volition diminution to make images of nationalist figures, similar Taylor Swift, and besides won’t nutrient images of weapons. And portion it volition halt abbreviated of generating named copyrighted characters, you tin get astir this beauteous easy by describing the quality you privation to create.

I got the instrumentality to make images that look precise overmuch similar Sonic the Hedgehog and Mario, portion my workfellow was capable to make characters resembling Mickey Mouse. I besides recovered that it volition make logos belonging to companies similar Apple, Macy’s, Hershey’s, and adjacent Google, which you tin spot successful the representation astatine the apical of this article.

Despite these somewhat flexible guardrails, Imagen 3 inactive stands successful stark opposition to Grok, the AI representation generator that lives connected Elon Musk’s X platform. Grok has been utilized to generate each sorts of chaotic content, including images with drugs, violence, and nationalist figures doing questionable things.

But Google’s AI tools person tally into immoderate issues arsenic well. Earlier this year, Google stopped letting radical generate images with its Gemini AI chatbot aft users recovered that it created historically inaccurate images.

Read Entire Article