Previously, this model was available only to users with a Gemini Advanced or Gemini for Workspace subscription.
Imagen 3 delivers a high level of photorealism, understands text prompts more accurately, and generates images with fewer visual artifacts than previous versions. The model can recognize details in complex queries and creates images with a resolution of 2048×2048 pixels. Thanks to the integration of the latest safety methods, the model avoids generating content that may be harmful or inappropriate, including photos of specific individuals.
Google pays special attention to ethics and responsibility in the operation of its model. The tool has undergone thorough data filtering, which reduces the likelihood of generating inappropriate content.
Imagen 3 is our highest-quality image model, delivering an even greater degree of photorealism, better instruction following, and fewer distracting artifacts than ever before.
However, Imagen 3 has certain limitations compared to competitors such as Midjourney and DALL-E 3. The model does not allow you to change the aspect ratio or resolution of images, which is usually important for professionals who need more flexibility when working with images.
You can try out Imagen 3’s capabilities via the Gemini chatbot. Simply log into the chat with your Google account and type in what image you want to generate. But keep in mind that in the free version, generating people is still not supported.
