Imagen 4 arrives as Google's latest image generation AI model
What's the story
Google has unveiled the latest version of its image generation model, Imagen 4. Eli Collins, VP of product at Google DeepMind, said that the new model combines speed with precision to create stunning images. He highlighted Imagen 4's excellence in both photorealistic and abstract styles, and its remarkable clarity in fine details such as intricate fabrics, water droplets, and animal fur.
Enhanced features
Improved spelling and typography
Collins also stressed that the new AI model is much better at spelling and typography. This, he says, makes it easier to make greeting cards, posters, and comics. While OpenAI's recent addition of image generation to ChatGPT also showcased text rendering improvements, the sample images from Google suggest that Imagen 4 may have surpassed those capabilities. Sample images from Google show this improvement with perfectly legible text in a short comic and even small fonts on mock stamps.
Deployment
Availability and future plans
Imagen 4 will be available from today in Gemini app, Whisk, and Vertex AI. It will also be integrated into Google Slides, Docs, Vids, "and more in Workspace," Collins said. The tech giant also plans to launch a "fast variant" of Imagen 4 soon which is said to be up to 10x faster than its predecessor, Imagen 3.