Google has announced wider availability of Imagen 2, launching the advanced text-to-image model across multiple products and services. Unveiled originally last December, Imagen 2 represents the second generation of Google's most capable generative AI image model.
Imagen 2's integration into Bard and ImageFX marks a significant expansion of its applications. Starting today, Bard users can now generate a wide range of high-quality images for various creative projects. Similarly, ImageFX, a new platform for image generation experimentation, incorporates Imagen 2 to allow users to create and modify images using simple text prompts and expressive chips, fostering a unique space for creativity and innovation.
This integration also extends Imagen 2's presence across Google's ecosystem, enhancing text-to-image functionalities in Google Ads, Duet AI in Workspace, the Search Generative Experience (SGE), and for enterprise developers through Google Cloud's Vertex AI.
While certainly not state-of-the-art, Imagen 2 has a number of key improvements in image quality, caption comprehension, and fine-grained control compared to the original Imagen. Specifically, Google focused on deepening the model's understanding of the relationship between images and text, leading to improvements in generating images that precisely reflect the intended context and nuances of user prompts. Combined with enhanced training data and methods, the model does a decent job rendering realistic details like human faces and hands.
More widespread access to Imagen 2 is important for Google as it attempts to catch up to Microsoft and OpenAI. Seamless integration into consumer-facing services like Bard and ImageFX in particular helps increase mainstream accessible and experimentation. Furthermore, allowing cloud platform access additionally caters to developers and enterprises seeking to build imaginative applications on top of Imagen 2's ever-evolving capabilities