On Wednesday, Google announced a significant upgrade to Gemini, its internal artificial intelligence (AI) model. The company announced that the chatbot’s image generation capability will now be powered by the Imagen 3 AI model for all users. Imagen 3 is the latest and most capable image generation model from the Mountain View-based tech giant. In addition to the Gemini app, the feature also extends to the API version of Gemini to allow developers to build apps and experiences based on this capability.
Gemini users get access to the Imagen 3 AI model
In a post on X (formerly known as Twitter), the official manager of the Google Gemini app revealed that all users, including those on the free tier, will be able to generate images using Imagen 3. The post highlighted that the AI model offers a high degree of photorealism, the better soon adhesion and adds fewer unwanted elements to images.
Gadgets 360 staff members were able to confirm that the Gemini app does indeed use Imagen 3 to generate images. To test its capabilities and compare it to Meta AI, we gave both chatbots the same query. The instruction was: “Draw a picture of a golden retriever sitting on a train bed and looking out the window at the Alps.” The train has a wooden interior, and the seats are green. All the other passengers on the train are also animals. One human conductor checks the tickets.”
Meta AI vs Gemini
The generated images can be seen above. While both AI models failed to include one or more of the elements listed in the query, Gemini was able to include multiple elements. Additionally, while Meta AI generates images at 1280 x 1280 resolution, Imagen 3 images are generated at 2048 x 2048 resolution.
Imagen 3 can generate images in a wide range of styles such as photorealistic, textured oil paintings and clay scenes. Users can also request that images look like they were taken with a specific camera such as a Nikon DSLR, GoPro style, wide angle lens and more.
Google said the AI model comes with built-in safeguards to reduce the risk of deep counterfeiting. Each generated image also comes watermarked with SynthID, a technology that adds an invisible AI tag within the pixels of the image. It cannot be cropped or removed and is present even in screenshots.