Google has recently released Imagen 3, its advanced AI text-to-image generator, to users in the United States. This latest version aims to enhance image quality by providing better detail, improved lighting, and reducing distracting artifacts compared to Google's previous models. Initially announced at the I/O conference in May, Imagen 3 was made available on Google's Vertex AI platform in the past few days. Early experiments by Reddit users began last week, and a research paper on the tool was published by Google on Tuesday. Similar to other AI image generators, Imagen 3 can generate detailed images based on user prompts. Users can also edit the generated images by specifying desired modifications. While there are certain limitations in place, such as refusing to generate images of public figures like Taylor Swift and avoiding images of weapons, Imagen 3 can still produce characters resembling copyrighted figures if described rather than named. Personally, I was able to use the tool to create images resembling Sonic the Hedgehog and Mario, while my colleague successfully created characters resembling Mickey Mouse.

Furthermore, the tool generates logos for various companies including Apple, Macy's, Hershey's, and even Google. Even with these flexible boundaries, Imagen 3 takes a different approach compared to Grok, the AI image generator on Elon Musk's X platform. Grok has been known to generate controversial content, including images involving drugs, violence, and public figures engaged in questionable activities. Google's AI tools have had their own challenges as well. Earlier this year, Google had to restrict image generation capabilities of its Gemini AI chatbot due to concerns of historically inaccurate image creation.