Google has made its latest AI text-to-image generator, Imagen 3, available to users in the US. The tool can generate high-quality images with detailed features, rich lighting, and fewer artifacts compared to previous models. Users can access Imagen 3 on Google’s AI Test Kitchen and edit the generated images by describing specific changes.
The tool has some guardrails in place to prevent inappropriate content. For example, it won’t generate images of public figures, weapons, or copyrighted characters. However, users can still create similar characters by providing descriptive prompts.
In testing, Imagen 3 successfully generated images resembling popular characters like Sonic the Hedgehog and Mario, as well as company logos like Apple’s and Google’s. This sets Imagen 3 apart from other AI image generators, such as Grok, which has been used to create controversial content.
While Imagen 3 has its limitations, it still raises concerns about the potential for AI-generated images to be used inappropriately. This is not the first time Google’s AI tools have faced issues; earlier this year, they stopped a chatbot from generating historically inaccurate images. Despite these challenges, Imagen 3 represents an exciting development in the field of AI-generated content.
Source: https://www.theverge.com/2024/8/15/24221218/google-ai-image-generator-imagen-3-available