Google's Updated AI Image Generator Creates Photorealistic Pictures With Legible Text

Google announced Imagen 3, the latest version of its image generation model, at its I/O developer event on Tuesday.

Imagen 3 yields more photorealistic images, including details like sunlight or whiskers on animals. It also remembers to include smaller details like wildflowers or birds in longer prompts.

Throughout 2024 so far, we've seen a flurry of image and video generation models from the likes of OpenAI, Adobe and Google itself as they seek to expand chatbots far beyond text and win over users. 

According to Douglas Eck, senior research director at Google, Imagen 3 is Google's best model yet for rendering text, which has been a challenge for image generation models.

"It understands prompts written the way people write," he said. "The more creative and detailed you are, the better."

The latest version also generates fewer distorted images.

It follows the release of Imagen 2 in December 2023.

Signups for access to Imagen 3 open today and it will be coming soon to developers and enterprise customers in Vertex AI.