Google has released the AI-generator Imagen 3 for US users
The third generation of the AI-generator model will have improved texture generation and word recognition, as well as better compliance with user prompts
According to VentureBeat, Google has released the latest version of its AI-powered text-to-image generator — Imagen 3.
Now, the new version is only available to users in the United States and users can try it through the ImageFX platform.
Imagen 3 can generate detailed images based on the user's specifications. You can also edit an image that has already been generated by highlighting a specific part of it and providing instructions on what needs to be changed. According to the company, Imagen 3 is supposed to generate images with "better detail, richer lighting, and fewer distracting artifacts" than previous Google models.
However, the new model have some limitations. For example, Imagine 3 won't be able to generate images of public figures and weapons. Users also can't create named characters that are protected by copyright, but they can get around these imitations by describing the character they want to create.
The screenshot below shows that Image 3 can create cartoon characters based on a description.
Google also assures that Image 3 confidently handles the generation of hands and fingers, ensuring all necessary proportions are observed.
The new Imagine 3 model has been compared to Midjourney v6, and it has been noted that the former surpasses the latter in several ways, particularly in terms of detail and realism of the images.
