Google has recently released the latest version of Imagen 3, its AI text-to-image generator, and made it available to all U.S. users through its Vertex AI platform. Along with that, it also published a research paper detailing the technology.
Google announced the updated Imagen 3 tool during I/O in May and limited it to selected Vertex AI users in June.
This upgraded tool is supposed to generate images with “better detail, richer lighting, and fewer distracting artifacts” compared to the previous versions. It can create detailed images based on your prompt. You can also edit the image by highlighting a certain part and explaining what you want to edit.
The company’s research team stated in their paper, “We introduce Imagen 3, a latent diffusion model that generates high-quality images from text prompts. Imagen 3 is preferred over other state-of-the-art models at the time of evaluation.”
The announcement was in the same week as xAI’s launch of Grok-2, an AI image generator with notably fewer restrictions, showing the different contrast in ethics and moderation.
Following the access, the reception to the update has been mixed. Some users are praising its improved texture and word recognition capabilities, while others complain about the strict filters and limited features.
The users have found many loopholes. It declines to generate images of public figures, such as Taylor Swift and refuses to generate images of weapons.
Although it doesn’t generate any copyrighted characters, you can get around this restriction by describing the characteristics of the character you want to create.
Despite these restrictions and features, Imagen 3 still stands apart from Elon Musk’s Grok, which has been used to generate all sorts of wild content, like drugs, violence, and public figures doing questionable things.
