Google has recently announced the re-enablement of the character image generation feature for its AI chatbot, Gemini. This decision stems from the controversial feature that was suspended earlier this year due to historically inaccurate reports. According to Google's statement, the early access version of this feature will be available to Gemini Advanced, Business, and Enterprise users in the coming days, currently supporting only English.
This upgrade is powered by Google's latest AI text-to-image generator, Imagen3. Initially introduced quietly through the AI Test Kitchen earlier this month, Imagen3 is now applied across all language versions of Gemini. The upgraded tool can generate a diverse range of content, from photo-realistic landscapes to textured oil paintings, with just a few descriptive sentences.
Dave Citron, Senior Director of Product Management for Google Gemini, stated that the new Imagen3 model is equipped with built-in safety measures, "outperforming other available image generation models." To avoid the previous issues of historical inaccuracies, Google has optimized the model to ensure the generated images are more accurate and appropriate.
It is worth noting that this feature of Gemini differs from Google's newly launched Remagine feature, which is specifically designed to integrate AI elements into photos taken with the Pixel9 phone series.
To further ensure content safety, Gemini does not allow users to create realistic images of public figures, content involving minors, or scenes of gore, violence, and sexual content. Citron acknowledged, "Of course, not every image created by Gemini is flawless, but we will continue to listen to feedback from early Gemini Advanced users and improve."
Google plans to extend the capability of AI-generated character images to more users and language versions in the future to meet the global demand. This move not only showcases Google's advancements in AI image generation technology but also reflects the company's commitment to addressing ethical and safety challenges in technology.