Google just rolled out its answer to Midjourney – here’s how to try the AI image generator

Google has expanded access to its AI image generator, Imagen 3, to all users in the U.S., following its initial limited release during the May 2024 Google I/O event. Imagen 3 is described as a latent diffusion model that creates high-quality images from text prompts and is touted by Google as superior to other state-of-the-art models. However, users have reported mixed reactions, noting that the tool is now more sensitive to prompts, often censoring benign words like "sock" or "water." This increased sensitivity has made it harder for users to achieve desired results compared to the previous version, Imagen 2. On the positive side, the model has been praised for its high-quality output and improved texture and word recognition. The expansion comes amid ongoing controversies surrounding AI image generators, with tools like Elon Musk’s Grok-2 AI facing criticism for lacking guardrails, allowing the creation of inappropriate or offensive images. Google appears to be taking a cautious approach with Imagen 3, prioritizing restrictions to avoid similar issues. Whether this cautious stance will change as the model evolves remains uncertain. The increased availability of Imagen 3 will likely lead to further exploration of its boundaries by users.