Earlier this year, Google showcased the text-to-image AI, Imagen, and now they have released a second version of it. This is a generative AI that, like Dall-e and Stable Diffusion, can create images based on text input by the user.
Google DeepMind, Google’s AI division, states that Imagen 2 can generate photorealistic images similar to those seen in the picture above. The new version of Imagen can also incorporate text in multiple languages, which could be used, for example, to create advertising campaigns and logos. Some examples of this can be found below.
Imagen 2 has also been equipped with a function where users can feed the AI an image and then receive a description of what the image depicts and also suggest captions for the picture.
Imagen 2 has not been released publicly but will be available for use by invited Google Cloud customers who also have access to Google’s machine learning platform Vertex AI.