TapTechNews May 15th news, Google announced the launch of Imagen3 at today's I/O 2024 developer conference, further enhancing its technology capabilities to generate images from text.
Demis Hassabis, head of Google's artificial intelligence research department DeepMind, said that compared to its predecessor Imagen2, Imagen3 can more accurately understand text prompts and convert them into images. The generated images are more 'creative and detailed', with fewer distracting elements and errors.
To dispel concerns about the possibility of Deepfake, Google stated that Imagen3 will use the SynthID method developed by DeepMind to apply invisible encryption watermarks in the media.
Users can register for a private preview of Imagen3 in Google's ImageFX tool. Google stated that this model will soon be made available to developers and enterprise customers using Google's enterprise generative artificial intelligence development platform VertexAI.
TapTechNews provides reference link
Imagen3
2024 Google I/O developer conference topic